For Peer Review Only

BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

The Cost-Effectiveness of HPV-Based Cervical Cancer Screening in the Public Health System in Nicaragua

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2016-015048

Article Type: Research

Date Submitted by the Author: 04-Nov-2016

Complete List of Authors: Campos, Nicole; Harvard T.H. Chan School of Public Health, Health Policy and Management Mvundura, Mercy; PATH, Devices and Tools Program Jeronimo, Jose; PATH, Reproductive Global Health Program Holme, Francesca; PATH, Reproductive Health Global Program Vodicka, Elisabeth; University of Washington, School of Pharmacy Kim, Jane; Harvard School of Public Health, Health Policy and Management

Primary Subject Health economics Heading:

Global health, Health economics, Health policy, Obstetrics and Secondary Subject Heading: gynaecology, Public health

Keywords: HEALTH ECONOMICS, Epidemiology < ONCOLOGY, PUBLIC HEALTH http://bmjopen.bmj.com/

on September 26, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 The Cost-Effectiveness of HPV-Based Cervical Cancer Screening in the Public Health System in 4 5 2 Nicaragua 6 3 Nicole G. Campos, PhD, a Mercy Mvundura, PhD,b Jose Jeronimo, MD,c Francesca Holme, MPH,c Elisabeth 7 8 d a 9 4 Vodicka, MHA, Jane J. Kim, PhD 10 11 5 12 13 6 a Center for Health Decision Science, Harvard T.H. Chan School of Public Health, 718 Huntington Avenue, 14 15 For peer review only 16 7 Boston, Massachusetts, USA 17 18 8 b PATH, Devices and Tools Program, P.O. Box 900922, Seattle, Washington, USA 19 20 9 c PATH, Reproductive Health Global Program, P.O. Box 900922, Seattle, Washington, USA 21

22 d 23 10 University of Washington, School of Pharmacy, Seattle, Washington, USA 24 25 11 Corresponding author: Nicole G. Campos, 718 Huntington Avenue, Boston, MA 02115; e-mail: 26 27 12 [email protected]; phone: 617-432-2019; fax: 617-432-0190 28 29 30 13 Word count (abstract): 300 31 32

33 14 Word count (text): 4,068 http://bmjopen.bmj.com/ 34 35 36 15 37 38 16 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 17 Abstract 4 5 18 Objectives: To evaluate the cost-effectiveness of HPV DNA testing (versus Pap-based screening) for 6 7 8 19 cervical cancer screening in Nicaragua. 9 10 20 Design: A previously developed Monte Carlo simulation model of the natural history of HPV infection 11 12 21 and cervical cancer was calibrated to epidemiologic data from Nicaragua. Cost data inputs were derived 13 14 22 using a micro-costing approach in Carazo, Chontales, and Chinandega departments; test performance 15 For peer review only 16 17 23 data were from a demonstration project in Masaya department. 18 19 24 Setting: Nicaragua’s public health sector facilities. 20 21 25 Participants: Women aged 30 to 59 years. 22 23 24 26 Interventions: Screening strategies included 1) Pap testing every 3 years, with referral to colposcopy for 25 26 27 women with an ASCUS+ result (“Pap”); 2) HPV testing every 5 years, with referral to cryotherapy for 27 28 28 HPV-positive eligible women (HPV screen-and-treat, or “HPV-ST”); 3) HPV testing every 5 years, with 29 30 31 29 referral to triage with visual inspection with acetic acid (VIA) for HPV-positive women (“HPV-VIA”); and 32

33 30 4) HPV testing every 5 years, with referral to Pap testing for HPV-positive women (“HPV-Pap”). http://bmjopen.bmj.com/ 34 35 31 Outcome measures: Reduction in lifetime risk of cancer; incremental cost-effectiveness ratios (ICER; 36 37 32 2015 US$ per year of life saved [YLS]). 38 39 40 33 Results: HPV-based screening strategies were more effective than Pap testing. HPV-ST was the least 41 on September 26, 2021 by guest. Protected copyright. 42 34 costly and most effective strategy, reducing lifetime cancer risk by 29.5% and outperforming HPV-VIA, 43 44 35 HPV-Pap, and Pap only, which reduced cancer risk by 19.4%, 12.2%, and 10.8%, respectively. With an 45 46 47 36 ICER of US$320/YLS, HPV-ST every 5 years would be very cost-effective using a threshold based on 48 49 37 Nicaragua’s per capita GDP of US$2,090. Findings were robust across sensitivity analyses on test 50 51 38 performance, coverage, compliance, and cost parameters. 52 53 54 55 56 57 58 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 39 Conclusions: HPV testing is very cost-effective compared to Pap testing in Nicaragua, due to higher test 4 5 6 40 sensitivity and the relatively lower number of visits required. Increasing compliance with recommended 7 8 41 follow-up will further improve the health benefits and value for public health dollars. 9 10 42 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 43 Strengths and limitations of this study: 4 5 6 44 • Using implementation data from the Scale-Up project— which aims to facilitate 7 8 45 institutionalization of HPV testing at the national level in Guatemala, Honduras, and Nicaragua— 9 10 46 we estimated the cost-effectiveness of HPV testing in Nicaragua’s public health system. 11 12 13 47 • The screening algorithms, as modeled, reflect the prototypical structure of a screening episode 14 15 48 and the typeFor of facility peer at which visits usuallyreview take place, but doonly not capture variation due to 16 17 49 geography or health facility capacity. 18 19 20 50 • While we adopted a micro-costing approach to leverage data from implementation in 21 22 51 Nicaragua, individual-level data for each woman were not available; furthermore, we did not 23 24 52 have cost data associated with HPV self-collection in community settings, where most self- 25 26 27 53 collection takes place. 28 29 54 • Despite these limitations, extensive sensitivity analyses on cost and screening assumptions 30 31 55 indicate that HPV-ST is robustly the most efficient strategy. 32

33 http://bmjopen.bmj.com/ 56 34 35 36 57 Funding statement: This work was supported by the Bill & Melinda Gates Foundation. 37 38 39 58 Disclaimer: The findings and conclusions contained within are those of the authors and do not 40 41 on September 26, 2021 by guest. Protected copyright. 42 59 necessarily reflect positions or policies of the Bill & Melinda Gates Foundation. The funders had no role 43 44 60 in study design; data collection, analysis, and interpretation; preparation of the manuscript; or decision 45 46 61 to submit the article for publication. 47 48 49 62 Competing interests disclosed: All authors have completed the ICMJE uniform disclosure form at 50 51 52 63 www.icmje.org/coi_disclosure.pdf and declare financial support from the Bill and Melinda Gates 53 54 64 Foundation for the submitted work; JJ was the director of the START-UP demonstration projects and 55 56 57 58 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 65 received all tests used in the study as a donation from Qiagen; no other relationships or activities that 4 5 6 66 could appear to have influenced the submitted work. 7 8 67 Contributors: All authors developed the analysis plan and interpreted the data. NC, MM, JJ, FH, and JK 9 10 68 conceptualized the study. MM, JJ, FH, and EV collected implementation data. NC conducted data 11 12 13 69 analysis and wrote the first draft of the report and revised subsequent drafts. JJ was the principal 14 15 70 investigator of theFor START-UP andpeer Scale-Up projects. review JK was the principal only investigator overseeing 16 17 71 microsimulation model development. All authors contributed to and approved the final report. 18 19 72 Data sharing statement: A supplementary appendix is available. 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 73 Background 4 5 6 74 Cervical cancer is the leading cause of cancer among women in Nicaragua, with an estimated 7 8 75 934 cases and 424 deaths each year [1]. Yet cervical cancer is preventable through screening that allows 9 10 76 for early detection and subsequent treatment of precancerous lesions caused by sexually transmitted 11 12 13 77 infection with human papillomavirus (HPV). While most HPV infections clear spontaneously within 1 to 2 14 15 78 years, a persistentFor infection withpeer one of approximately review 15 oncogenic onlyHPV genotypes may progress to 16 17 79 precancer, which, if untreated, may become invasive cancer [2, 3]. 18 19 80 In most high-income countries, routine screening with cervical cytology (i.e., Pap) testing has 20 21 22 81 substantially reduced the incidence of cervical cancer [4]. However, due to its low sensitivity to detect 23 24 82 precancer, Pap testing must be performed at regular frequent intervals in women of screening age. In 25 26 83 low- and middle-income settings, where many women do not have access to routine primary health care 27 28 29 84 and even fewer have access to higher-level facilities that offer diagnostic testing and treatment, Pap 30 31 85 testing has not been effective at reducing cervical cancer incidence and mortality [5]. In Nicaragua, an 32

33 86 estimated 31.5% of women aged 15 to 49 years have been screened within the last year, and nearly 30% http://bmjopen.bmj.com/ 34 35 36 87 of women in this age group have never been screened [6]. One recent survey found that 87% of women 37 38 88 in León, Nicaragua, were informed of their Pap results, but of those who were referred to follow-up, 39 40 89 only 67% received further care [7]. 41 on September 26, 2021 by guest. Protected copyright. 42 90 HPV DNA tests are highly sensitive to detect potentially oncogenic HPV infections and present 43 44 45 91 an alternative to Pap-based screening. Because HPV-negative women are at very low risk for developing 46 47 92 cervical cancer within the next 10 years [8], the interval between screenings can be extended to at least 48 49 93 5 years for this subset of women [9]. An additional advantage of HPV testing is that samples can be 50 51 52 94 collected by a provider or by the woman herself, reducing the burden on health workers and time 53 54 95 women spend seeking care and potentially increasing screening uptake [10-13]. Furthermore, a lower- 55 56 96 cost HPV DNA test known as careHPV has been clinically validated [14, 15] and is now commercialized. 57 58 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 97 Given these potential benefits, the World Health Organization (WHO) recommends HPV testing for 4 5 6 98 countries with sufficient resources [9]. 7 8 99 In 2011, the Screening Technologies to Advance Rapid Testing for Cervical Cancer 9 10 100 Prevention−Ulity and Program Planning (START-UP) project in Nicaragua’s Masaya department 11 12 13 101 demonstrated that screening with careHPV could be effectively implemented in public sector health 14 15 102 facilities [14]. TheFor Ministry of Healthpeer of Nicaragua review subsequently built onlyon these initial efforts, 16 17 103 incorporating HPV testing into public health care systems in three departments with technical assistance 18 19 104 from PATH under the Scale-Up project. Adoption of HPV testing within Nicaragua’s public health care 20 21 22 105 system is taking place in three phases [16]. In Phase 1, partner organizations worked with the Ministry of 23 24 106 Health to prepare for introduction of HPV screening into public health facilities by developing screening 25 26 107 and treatment algorithms; creating educational materials; organizing training sessions for health 27 28 29 108 workers and laboratory technicians; and bolstering referral and treatment systems for follow-up of 30 31 109 screen-positive women. Phase 2 piloted screening with 10,000 HPV tests in order to identify and address 32

33 110 barriers to implementation. Phase 3 will expand coverage to over 50,000 women within one year. http://bmjopen.bmj.com/ 34 35 36 111 To inform decision makers considering the national adoption and scale-up of HPV testing within 37 38 112 Nicaragua’s public health sector, this study aimed to 1) estimate the economic cost of cervical cancer 39 40 113 screening with careHPV testing; and 2) project the long-term health and economic impact and value 41 on September 26, 2021 by guest. Protected copyright. 42 114 (i.e., cost-effectiveness) of careHPV testing in Nicaragua relative to existing Pap-based screening. 43 44 45 115 46 47 116 Methods 48 49 117 Analytic overview 50 51 118 We used a micro-costing approach to measure and aggregate the cost of all resources used to 52 53 119 provide cervical cancer screening at the level of the individual patient within the public health sector in 54 55 120 Nicaragua. We considered direct medical costs (i.e., medical resources required for the intervention), 56 57 121 direct non-medical costs (i.e., other resources consumed as part of the intervention, such as patient 58 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 122 transportation costs), and patient time costs (i.e., time spent traveling and waiting for or receiving care). 4 5 6 123 These cost data were input into a previously developed Monte Carlo simulation model of the natural 7 8 124 history of HPV infection and cervical cancer that was calibrated to epidemiologic data from Nicaragua 9 10 125 [17, 18]. We then used the model to project the lifetime health and economic outcomes associated with 11 12 13 126 careHPV testing, using three different algorithms for the management of women who test HPV-positive, 14 15 127 and Pap-based screeningFor for womenpeer aged 30 toreview 59 years. only 16 17 128 Model outcomes included the lifetime risk of cervical cancer, total lifetime costs per woman (in 18 19 129 2015 United States dollars [US$]), and life expectancy. Incremental cost-effectiveness ratios (ICERs) 20 21 22 130 were calculated by dividing the additional cost of a particular strategy by its additional health benefit, 23 24 131 compared with the next most costly strategy. Dominated strategies (defined as more costly and either 25 26 132 less effective or having a higher cost-effectiveness ratio than more effective strategies) were eliminated. 27 28 29 133 There is no universal criterion that defines a threshold cost-effectiveness ratio, below which an 30 31 134 intervention is considered good value for money; we considered an intervention with an ICER less than 32

33 135 Nicaragua’s 2015 per capita GDP of US$2,090 to be “very cost-effective”, and an intervention with an http://bmjopen.bmj.com/ 34 35 36 136 ICER less than three times per capita GDP as “cost-effective” [19]. We followed guidelines for cost- 37 38 137 effectiveness by adopting a societal perspective, including costs irrespective of the payer in order to 39 40 138 capture the opportunity cost of resources used for the screening intervention. We discounted future 41 on September 26, 2021 by guest. Protected copyright. 42 139 costs and life-years at a rate of 3% per year to account for time preferences [20, 21]. 43 44 45 140 46 47 141 Mathematical simulation model 48 49 142 Descriptions of the natural history model of HPV infection and cervical carcinogenesis and 50 51 143 model parameterization process have been previously published [17, 18], but we summarize model 52 53 54 144 features here. Individual girls enter the model at age 9 years, prior to initiating sexual activity, and face 55 56 145 monthly transitions between mutually exclusive health states that reflect disease progression, including 57 58 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 146 type-specific HPV infection, grade of precancer (i.e., cervical intraepithelial neoplasia [CIN] grade 2 or 3), 4 5 6 147 and stage of invasive cancer. Transition probabilities may vary by age, HPV type, duration of infection or 7 8 148 precancerous lesion status, prior HPV infection, and exposure to screening and treatment of HPV or 9 10 149 precancer. Cervical cancer can be detected through symptoms or screening. Death can occur from non- 11 12 13 150 cervical causes or from cervical cancer after its onset. The model tracks each individual woman’s health 14 15 151 status, clinical events,For and economic peer outcomes review over her lifetime, and only aggregates outcomes to estimate 16 17 152 the expected costs and health outcomes over the lifetime of the cohort. 18 19 153 The model was calibrated to epidemiologic data on age-specific HPV prevalence and cervical 20 21 22 154 cancer incidence from Nicaragua [1, 14, 17]. We estimated baseline “prior” input parameter values for 23 24 155 natural history transitions using available longitudinal data, including age- and type-specific HPV 25 26 156 incidence data from Colombia [22-25]. To reflect potential differences in parameters that may vary by 27 28 29 157 setting (i.e., age- and type-specific HPV incidence, natural immunity following initial infection) and 30 31 158 uncertainty in progression and regression of precancer, we set plausible bounds around these input 32

33 159 values and performed repeated model simulations of disease natural history in the absence of any http://bmjopen.bmj.com/ 34 35 36 160 intervention. Each model simulation selected one random value within the bounds for each uncertain 37 38 161 parameter, creating a unique natural history input parameter set. By summing the log-likelihood of 39 40 162 model-projected outcomes for each parameter set relative to the epidemiologic data from Nicaragua, 41 on September 26, 2021 by guest. Protected copyright. 42 163 we computed a goodness-of-fit score. We selected the 50 top-fitting input parameter sets to use in 43 44 45 164 analysis. Results are reported as the mean across the top 50 parameter sets, and ICERs are reported as 46 47 165 the ratio of the mean costs divided by the mean effects of one strategy versus another across sets [26]. 48 49 166 Further details on model parameterization, including calibration, are available in the Appendix. 50 51 52 167 53 54 168 Cervical cancer screening strategies 55 56 57 58 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 169 We considered the following screening strategies for women aged 30 to 59 years (Figure 1): 1) 4 5 6 170 Pap testing every 3 years, with referral to colposcopy for all women with an atypical squamous cells of 7 8 171 undetermined significance or worse (ASCUS+) result (hereafter referred to as “Pap”); 2) careHPV testing 9 10 172 every 5 years, with referral to cryotherapy for all HPV-positive eligible women (HPV screen-and-treat, or 11 12 13 173 “HPV-ST”); 3) careHPV testing every 5 years, with referral to triage with visual inspection with acetic acid 14 15 174 (VIA) for all HPV-positiveFor women peer (“HPV-VIA”); reviewand 4) careHPV testing only every 5 years, with referral to Pap 16 17 175 testing for all HPV-positive women (“HPV-Pap”). The pathway of care for each strategy was based on 18 19 176 patterns of care in the Scale-Up project, national screening guidelines, and WHO recommendations. We 20 21 22 177 optimistically assumed 70% of women had access to routine screening and attended an initial visit at a 23 24 178 screening facility (i.e., a primary health care facility). Women could then return to receive screening 25 26 179 results and recommendations for any necessary follow-up care. Follow-up could include colposcopy, 27 28 29 180 cryotherapy, or triage testing at a referral (i.e., higher level) facility, with the exception of Pap triage 30 31 181 testing, which was assumed to take place at the screening clinic. At each encounter after the initial 32

33 182 screening visit, we assumed 85% of women complied with each subsequent visit to a screening facility, http://bmjopen.bmj.com/ 34 35 36 183 while 40% complied with each subsequent visit to a referral facility, consistent with data from Phase 2 of 37 38 184 the Scale-Up project. The minimum number of visits required for treatment in a single screening episode 39 40 185 was 4 for Pap, 3 for HPV-ST, 3 for HPV-VIA, and 5 for HPV-Pap. In the HPV-VIA and HPV-Pap strategies, 41 on September 26, 2021 by guest. Protected copyright. 42 186 women who were HPV-positive but negative on the selected triage test were referred to repeat HPV 43 44 45 187 testing in 1 year. In the HPV-ST and HPV-VIA strategies, women who were not eligible for treatment with 46 47 188 cryotherapy based on visual assessment were referred to colposcopy with biopsy to rule out cancer; in 48 49 189 the absence of cancer, these women were referred to treatment with loop electrosurgical excision 50 51 52 190 procedure (LEEP) or cryotherapy. 53 54 191 Screening and treatment parameters are presented in Table 1 [6, 14, 17, 27-32]. Screening test 55 56 192 performance data were drawn from the START-UP project in Nicaragua to reflect local test 57 58 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 193 characteristics. While the START-UP project did not evaluate VIA and Pap as triage tests, we used the 4 5 6 194 VIA and Pap positivity rates in HPV-positive women, along with published studies of triage test 7 8 195 performance, to inform triage test sensitivity and specificity [33-38]; in the base case, we optimistically 9 10 196 assumed high sensitivity of triage testing. 11 12 13 197 For all HPV testing strategies, we assumed 20% of women received provider-collection of 14 15 198 cervical specimensFor and 80% ofpeer women self-collected review vaginal specimens, only consistent with the proportions 16 17 199 in the Scale-Up project to date. We weighted cost and health outcomes for provider- and self-collection 18 19 200 accordingly when aggregating results for the HPV strategies. 20 21 22 201 23 24 202 Cost data 25 26 203 All costs were converted to 2015 United States dollars (US$) using GDP deflators and the official 27 28 204 exchange rate [39]. The direct medical costs of screening, diagnosis, and treatment of precancer were 29 30 31 205 drawn from the START-UP study (Masaya department) and the Scale-Up project (Carazo, Chontales, and 32

33 206 Chinandega departments). Direct medical costs included clinical staff time, clinical supplies, drugs, http://bmjopen.bmj.com/ 34 35 207 clinical equipment, laboratory staff time, laboratory supplies, and laboratory equipment. Direct non- 36 37 208 medical costs included women’s round-trip transportation costs to health facilities, and were based on 38 39 40 209 estimates provided by Scale-Up project staff to represent average transportation costs in the Carazo, 41 on September 26, 2021 by guest. Protected copyright. 42 210 Chontales, and Chinandega departments. To account for the opportunity cost of women’s time spent 43 44 211 traveling to, waiting for, or receiving care, we used time estimates from the START-UP and Scale-Up 45 46 47 212 projects, and valued women’s time using Nicaragua’s monthly minimum wage to serve as a proxy for the 48 49 213 societal value of women’s time. Figure 2 displays the categorical breakdown of undiscounted costs for 50 51 214 the screening visits over the course of a woman’s screening-eligible years, with Pap versus careHPV 52 53 54 215 testing. 55 56 57 58 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 216 Data on programmatic costs are limited, but for HPV strategies we included the cost of training 4 5 6 217 sessions for health providers offering HPV-based screening, outreach workers, laboratory technicians, 7 8 218 and providers offering VIA and cryotherapy. While women who self-collected HPV specimens in the 9 10 219 Scale-Up project primarily did so in a community setting, micro-costing data were not available for self- 11 12 13 220 collection performed outside of the clinic, so we assumed clinic-based self-collection. However, we 14 15 221 conservatively includedFor the cost peer of training outreach review workers to represent only a known programmatic cost 16 17 222 as self-collection efforts are shifting to community settings. 18 19 223 Data on the costs of treating cervical cancer were unavailable for Nicaragua, so we estimated 20 21 22 224 direct medical, direct non-medical, and patient and support person time costs using data from El 23 24 225 Salvador [29]. 25 26 226 Selected cost data are presented in Table 1. Further details on cost data are provided in the 27 28 29 227 Appendix. 30 31 228 32

33 229 Sensitivity analyses http://bmjopen.bmj.com/ 34 35 230 We performed sensitivity analysis to examine the impact of independently varying uncertain 36 37 231 parameters, including Pap test performance, triage test performance in HPV-positive women, 38 39 40 232 colposcopy performance, screening coverage, visit compliance, treatment effectiveness, and cost data. 41 on September 26, 2021 by guest. Protected copyright. 42 233 Ranges selected for sensitivity analysis are displayed in Table 1. 43 44 234 45 46 47 235 Results 48 49 236 Base case: Population-level health benefits and cost-effectiveness analysis 50 51 237 HPV-based screening strategies were more effective than Pap testing. Among the HPV 52 53 54 238 strategies, HPV-ST (every 5 years) was the most effective strategy; under base case assumptions, it 55 56 239 reduced the lifetime risk of cervical cancer by 29.5% on average (range: 25.2%−33.6%). HPV-VIA (every 5 57 58 59 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 240 years) reduced cancer risk by 19.4% (range: 16.2%−22.6%), while HPV-Pap reduced cancer risk by 12.2% 4 5 6 241 (range: 10.2%−14.5%). Pap (every 3 years) reduced cancer risk by 10.8% (range: 8.7%−13.4%). Under 7 8 242 base case assumptions, HPV-ST was both less costly and more effective than all other strategies, thus 9 10 243 dominating HPV-VIA, HPV-Pap, and Pap alone. With an ICER of US$320 per year of life saved (YLS), HPV- 11 12 13 244 ST every 5 years would be considered “very cost-effective” given Nicaragua’s per capita GDP of 14 15 245 US$2,090. The totalFor discounted peer lifetime cost perreview woman and life expectancy only associated with each 16 17 246 screening strategy is presented in Figure 3. 18 19 247 20 21 22 248 Sensitivity analysis 23 24 249 While HPV-ST remained the most effective strategy across all sensitivity analyses, the magnitude 25 26 250 of reduction in lifetime risk of cancer was dependent upon screening coverage of the target population 27 28 29 251 and compliance with recommended follow-up. When coverage was 50% and all other parameters were 30 31 252 held constant at base case values, HPV-ST reduced cancer risk by an average of 21.1%; HPV-VIA, HPV- 32

33 253 Pap, and Pap yielded average cancer risk reductions of 13.9%, 8.7%, and 7.7%, respectively. As coverage http://bmjopen.bmj.com/ 34 35 36 254 increased to 80%, all else being equal, HPV-ST reduced cancer risk by an average of 33.5%, while HPV- 37 38 255 VIA, HPV-Pap, and Pap yielded average cancer risk reductions of 22.1%, 14.0%, and 12.4%, respectively 39 40 256 (Appendix). Figure 4 displays the impact of visit compliance on lifetime risk of cancer. When compliance 41 on September 26, 2021 by guest. Protected copyright. 42 257 with visits to all facilities (i.e., for both screening and referral) was low at 40%, HPV-ST remained the 43 44 45 258 most effective strategy, but only reduced cancer risk by 16.2%; Pap had little health impact at this level 46 47 259 of compliance, reducing cancer risk by only 5.4%. As compliance at all facilities rose to 85%, HPV-ST 48 49 260 reduced cancer risk by 47.9%; HPV-VIA, HPV-Pap, and Pap reduced cancer risk by 42.0%, 40.7%, and 50 51 52 261 35.7% respectively. 53 54 262 In addition to remaining the most effective strategy across all sensitivity analyses, HPV-ST 55 56 263 remained the most efficient strategy as well. HPV-ST remained the least costly and most effective 57 58 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 264 strategy with a stable ICER of US$320 per YLS when 1) Pap test performance (as a primary screening 4 5 6 265 test) improved; 2) VIA and Pap triage test performance improved; and 3) colposcopy was assumed to be 7 8 266 perfect. Despite slight fluctuation in the ICER, HPV-ST also remained the least costly and most effective 9 10 267 strategy as 1) screening coverage varied from 50% to 80%; 2) visit compliance varied from 40% to 85% 11 12 13 268 per visit; 3) the screen-and-treat cryotherapy cure rate was reduced to 75%; 4) the treatment cure rate 14 15 269 following colposcopyFor was reduced peer to 85%; 5) onlyreview payer costs were considered,only excluding women’s time 16 17 270 and transportation costs; 6) the direct medical cost of HPV self-collection was varied from 75% to 125% 18 19 271 of the base case; 7) the direct medical cost of cryotherapy was increased to 170% of the base case; 8) 20 21 22 272 the direct medical cost of colposcopy was reduced to 35% of the base case; 9) programmatic costs 23 24 273 associated with HPV-based screening were varied from 50% to 150% of the base case; 10) women’s time 25 26 274 and transportation costs were reduced to 50% of the base case; and 11) the costs of cancer treatment 27 28 29 275 ranged from including only direct medical costs to 150% of the base case (Figure 5). Among variables 30 31 276 considered, compliance per visit appears to have the greatest impact on the ICER for HPV-ST, with 40% 32

33 277 compliance yielding an ICER of US$580 per YLS and 85% compliance yielding an ICER of US$190 per YLS. http://bmjopen.bmj.com/ 34 35 36 278 Even when visit compliance is low, HPV-ST would be considered very cost-effective. 37 38 279 When only payer costs were considered (i.e., women’s time and transportation costs were 39 40 280 excluded), the total lifetime cost per woman was lower for all strategies, and Pap every 3 years was 41 on September 26, 2021 by guest. Protected copyright. 42 281 slightly less costly (although still slightly less effective) than HPV-Pap every 5 years. HPV-ST remained the 43 44 45 282 most effective and efficient strategy, with an ICER of US$270 per YLS (Appendix). 46 47 283 The only scenario in which Pap testing every 3 years was the least costly strategy occurred when 48 49 284 the direct medical cost of Pap testing was US$3 (base case: US$7.26), a value commonly cited for the 50 51 52 285 cost of Pap in Nicaragua, though the source of this estimate is unknown. However, Pap remained the 53 54 286 least effective strategy, and HPV-ST had a lower cost-effectiveness ratio, maintaining an ICER of US$320 55 56 287 per YLS (Appendix). 57 58 59 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 288 4 5 6 289 Discussion 7 8 290 Using implementation data from the Scale-Up project— which aims to facilitate 9 10 291 institutionalization of HPV testing at the national level in Guatemala, Honduras, and Nicaragua— we 11 12 13 292 estimated the long-term health impact and value of careHPV testing in Nicaragua’s public health system. 14 15 293 We found that screeningFor algorithms peer consisting review of HPV testing at 5-year only intervals would be less costly and 16 17 294 more effective than screening with Pap testing at 3-year intervals. Furthermore, HPV testing followed by 18 19 295 treatment with cryotherapy for all eligible HPV-positive women would be less costly and more effective 20 21 22 296 than HPV testing followed by triage testing with either VIA or Pap for HPV-positive women. A screen- 23 24 297 and-treat HPV program would be a very cost-effective intervention in Nicaragua, with an ICER of US$320 25 26 298 per YLS under base case assumptions. These findings were robust across sensitivity analyses. The 27 28 29 299 comparatively large health benefits and efficiency of HPV-ST can largely be attributed to the relatively 30 31 300 low number of visits to health care facilities and the high sensitivity of the careHPV test to detect both 32

33 301 CIN2+ and oncogenic HPV infections with the potential to develop into precancer. http://bmjopen.bmj.com/ 34 35 36 302 We found that screening coverage of the target population had a considerable impact on 37 38 303 achievable reductions in cervical cancer risk, with HPV-ST yielding the greatest risk reduction. Due to 39 40 304 proportional increases in both costs and health benefits, the ICER for HPV-ST remained stable as 41 on September 26, 2021 by guest. Protected copyright. 42 305 coverage increased from 50% to 80%. Compliance with recommended follow-up was a key driver of 43 44 45 306 both achievable reductions in cancer risk and the ICER of HPV-ST. As the proportion of women who 46 47 307 returned for each clinical encounter (relative to the previous visit) increased from 40% to 85%, the 48 49 308 cancer benefit associated with HPV-ST rose from 16.2% to 47.9%; the ICER fell from US$580 per YLS to 50 51 52 309 US$190 per YLS. Thus, improved efforts to successfully navigate women to recommended follow-up will 53 54 310 enhance screening program effectiveness and efficiency. 55 56 57 58 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 311 While substantially reducing the cost of Pap testing to US$3 (less than half of the base case 4 5 6 312 value) made Pap the strategy with the lowest per-woman lifetime costs, the confluence of low test 7 8 313 sensitivity and the high number of health facility visits needed to complete screening, diagnostic follow- 9 10 314 up, and treatment made Pap the least effective strategy since women are lost to follow-up with each 11 12 13 315 additional required visit. Even in this low-cost Pap scenario, HPV-ST remained the most effective and 14 15 316 cost-effective strategy.For peer review only 16 17 317 There are several limitations to this analysis. We did not consider alternative screening intervals 18 19 318 or ages for each strategy, but rather restricted the analysis to the ages and intervals currently under 20 21 22 319 consideration by the Nicaraguan Ministry of Health. The screening algorithms, as modeled, reflect the 23 24 320 prototypical structure of a screening episode and the type of facility at which visits usually take place, 25 26 321 but do not capture variation due to geography or health facility capacity. Furthermore, the modeled 27 28 29 322 screening algorithms do not entirely reflect the complex downstream follow-up of screen-positive 30 31 323 women that is embodied in the Ministry of Health’s screening guidelines. In simplifying the downstream 32

33 324 follow-up for modeled strategies, we may have underestimated the costs and overestimated the http://bmjopen.bmj.com/ 34 35 36 325 benefits relative to the national guidelines, which call for additional follow-up prior to treatment. 37 38 326 However, the modeled strategies would likely bias the analysis in favor of Pap and HPV triage strategies. 39 40 327 While we adopted a micro-costing approach to leverage data from the START-UP and Scale-Up 41 on September 26, 2021 by guest. Protected copyright. 42 328 projects in Nicaragua, there remain limitations to our cost estimates. First, individual-level data for each 43 44 45 329 woman were not available; thus, our estimates represent average costs in the project populations. 46 47 330 Second, we did not have information on the costs associated with HPV self-collection in community 48 49 331 settings, where most self-collection takes place. Instead, we assumed all self-collection took place at the 50 51 52 332 clinic. Compared to clinic-based efforts, community-based self-collection may be associated with lower 53 54 333 costs for women’s time and travel, and higher direct medical and programmatic costs due to outreach 55 56 334 worker involvement in facilitating screening and delivering results. Third, our estimates of programmatic 57 58 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 335 costs were restricted to training sessions, and we did not have information on the costs of social 4 5 6 336 mobilization and outreach, patient navigation and support, or infrastructural improvements that would 7 8 337 be required to successfully scale-up a screening program. Fourth, we valued women’s time based on the 9 10 338 minimum wage in Nicaragua. This may be a conservative estimate if most women attending screening 11 12 13 339 are formally employed; conversely, it may overestimate the societal value of women’s time spent 14 15 340 working in the informalFor sector peer or at home. Finally, review we extrapolated theonly cost of cancer treatment using 16 17 341 data from El Salvador [29]. Despite these limitations, extensive sensitivity analyses on cost components 18 19 342 indicate that HPV-ST is robustly the most efficient strategy. 20 21 22 343 As implementation of HPV testing continues, particularly without triage testing, the health 23 24 344 system’s capacity to provide cryotherapy will likely need to increase. While the use of triage testing 25 26 345 (either with Pap or HPV) reduces the number of cryotherapy procedures performed, we found that the 27 28 29 346 lower sensitivity of triage testing (resulting in more false negatives) led to a decline in health benefits as 30 31 347 fewer women with persistent HPV infection and precancer received treatment. The cost savings 32

33 348 associated with fewer cryotherapy procedures were outweighed by increased costs of additional follow- http://bmjopen.bmj.com/ 34 35 36 349 up and cancer treatment in triage-negative women. A sensitivity analysis on the cost of cryotherapy 37 38 350 revealed that HPV-ST remained the dominant strategy even when costs increased to 170% of the base 39 40 351 case. However, we did not explicitly consider the costs of increasing access to cryotherapy machines or 41 on September 26, 2021 by guest. Protected copyright. 42 352 the implications of gas stock-outs, which have been identified as barriers in some low- and middle- 43 44 45 353 income countries [40]. New ablative technologies currently undergoing testing are smaller, portable, 46 47 354 and do not require gas. Thermal-coagulation has been used in United Kingdom for more than 30 years, 48 49 355 and now it is being used in several low- and middle-income countries, including as part of a “screen-and- 50 51 52 356 treat” program in Malawi [41], and is currently undergoing testing in Latin America. If newer 53 54 357 technologies demonstrate cure rates similar to cryotherapy, the cost-effectiveness of screen-and-treat 55 56 358 algorithms may improve along with access to treatment. 57 58 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 359 In summary, using data from the Scale-Up implementation project in Nicaragua, we found that 4 5 6 360 HPV testing followed by cryotherapy for eligible HPV-positive women (a “screen-and-treat” approach) 7 8 361 was a very cost-effective intervention in Nicaragua. As the HPV-ST algorithm was not implemented in 9 10 362 Phase 2 of the Scale-Up project, compliance and cost estimates may need to be further honed to reflect 11 12 13 363 improvements in capacity for cryotherapy if HPV-ST is implemented going forward. Still, extensive 14 15 364 sensitivity analysesFor indicate thepeer robustness of reviewfindings. An HPV-based only screening algorithm involving a 16 17 365 similar “screen-and-treat” approach was recently found to be good value for public health dollars in El 18 19 366 Salvador [29], where national scale-up is underway. It is important to note that a favorable cost- 20 21 22 367 effectiveness profile does not guarantee that HPV-ST will be affordable or feasible in a lower-middle 23 24 368 income country like Nicaragua. Both the cost-effectiveness ratio, budgetary impact, and health systems 25 26 369 infrastructure need to be favorable for screening programs to be sustainable. We present these findings 27 28 29 370 to inform evidence-based decision making around national screening guidelines, program design and 30 31 371 implementation, and budgeting for infrastructural improvements and procurement of HPV tests in 32

33 372 Nicaragua. http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 373 References 4 5 374 6 7 375 1. Ferlay J, Soerjomataram I, Ervik M, et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality 8 376 Worldwide: IARC CancerBase No. 11 [Internet], 2013, International Agency for Research on 9 377 Cancer: Lyon, France. 10 11 378 2. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of 12 379 invasive cervical cancer worldwide. J Pathol 1999;189(1):12-19. 13 380 3. Rodriguez AC, Schiffman M, Herrero R, et al. Longitudinal study of human papillomavirus 14 381 persistence and cervical intraepithelial neoplasia grade 2/3: critical role of duration of infection. 15 382 J Natl CancerFor Inst 2010;102(5):315-324. peer review only 16 383 4. Kitchener HC, Castle PE, and Cox JT. Chapter 7: Achievements and limitations of cervical cytology 17 18 384 screening. Vaccine 2006;24 Suppl 3:S3/63-70. 19 385 5. Denny L, Quinn M, and Sankaranarayanan R. Chapter 8: Screening for cervical cancer in 20 386 developing countries. Vaccine 2006;24 Suppl 3:S3/71-77. 21 387 6. Institut Catala Oncologia. HPV Information Centre, 2014. 22 388 7. Vastbinder MB, Castillo C, and Bekkers RL. Barriers to follow-up of abnormal Papanicolaou 23 389 smears among women in Leon, Nicaragua. Trop Doct 2010;40(1):22-26. 24 390 8. Khan MJ, Castle PE, Lorincz AT, et al. The elevated 10-year risk of cervical precancer and cancer 25 26 391 in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type- 27 392 specific HPV testing in clinical practice. J Natl Cancer Inst 2005;97(14):1072-1079. 28 393 9. World Health Organization. WHO Guidelines for Screening and Treatment of Precancerous 29 394 Lesions for Cervical Cancer Prevention. World Health Organization, 2013: Geneva. 30 395 10. Arrossi S, Thouyaret L, Herrero R, et al. Effect of self-collection of HPV DNA offered by 31 396 community health workers at home visits on uptake of screening for cervical cancer (the EMA 32 397 study): a population-based cluster-randomised trial. Lancet Glob Health 2015;3(2):e85-94.

33 http://bmjopen.bmj.com/ 34 398 11. Moses E, Pedersen HN, Mitchell SM, et al. Uptake of community-based, self-collected HPV 35 399 testing vs. visual inspection with acetic acid for cervical cancer screening in Kampala, Uganda: 36 400 preliminary results of a randomised controlled trial. Trop Med Int Health 2015;20(10):1355- 37 401 1367. 38 402 12. Lazcano-Ponce E, Lorincz AT, Cruz-Valdez A, et al. Self-collection of vaginal specimens for human 39 403 papillomavirus testing in cervical cancer prevention (MARCH): a community-based randomised 40 41 404 controlled trial. Lancet 2011;378(9806):1868-1873. on September 26, 2021 by guest. Protected copyright. 42 405 13. Racey CS, Withrow DR, and Gesink D. Self-collected HPV testing improves participation in 43 406 cervical cancer screening: a systematic review and meta-analysis. Can J Public Health 44 407 2013;104(2):e159-166. 45 408 14. Jeronimo J, Bansil P, Lim J, et al. A multicountry evaluation of careHPV testing, visual inspection 46 409 with acetic acid, and papanicolaou testing for the detection of cervical cancer. Int J Gynecol 47 410 Cancer 2014;24(3):576-585. 48 49 411 15. Qiao YL, Sellors JW, Eder PS, et al. A new HPV-DNA test for cervical-cancer screening in 50 412 developing regions: a cross-sectional study of clinical accuracy in rural China. Lancet Oncol 51 413 2008;9(10):929-936. 52 414 16. Jeronimo J, Holme F, Slavkovsky R, et al. Implementation of HPV testing in Latin America. J Clin 53 415 Virol 2016;76 Suppl 1:S69-73. 54 416 17. Campos NG, Tsu, V., Jeronimo, J., Mvundura, M., Lee, K., Kim, J.J. When and how often to screen 55 56 417 for cervical cancer in three low- and middle-income countries: A cost-effectiveness analysis. 57 418 Papillomavirus Research 2015. 58 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 419 18. Campos NG, Burger EA, Sy S, et al. An updated natural history model of cervical cancer: 4 420 derivation of model parameters. Am J Epidemiol 2014;180(5):545-555. 5 6 421 19. World Health Organization. Macroeconomics and health: Investing in health for economic 7 422 development: Report of the Commission on Macroeconomics and Health. World Health 8 423 Organization, 2001: Geneva. 9 424 20. Tan-Torres Edejer T, Baltussen, R., Adam, T., Hutubessy, R., Acharya, A., Evans, D.B., Murray, 10 425 C.J.L., ed. Making Choices in Health: WHO Guide to Cost-Effectiveness Analysis. 2003, World 11 426 Health Organization: Geneva, Switzerland. 12 13 427 21. Jamison DT, Breman, J.G., Measham, A.R., Alleyne, G., Claeson, M., Evans, D.B., Jha, P., Mills, A., 14 428 Musgrove, P., ed. Disease Control Priorities in Developing Countries. Second edition ed. 2006, 15 429 Oxford UniversityFor Press peer and The World review Bank: Washington, D.C. only 16 430 22. Munoz N, Mendez F, Posso H, et al. Incidence, duration, and determinants of cervical human 17 431 papillomavirus infection in a cohort of Colombian women with normal cytological results. J 18 432 Infect Dis 2004;190(12):2077-2087. 19 433 23. Herrero R, Hildesheim A, Rodriguez AC, et al. Rationale and design of a community-based 20 21 434 double-blind randomized clinical trial of an HPV 16 and 18 vaccine in Guanacaste, Costa Rica. 22 435 Vaccine 2008;26(37):4795-4808. 23 436 24. Surveillance Epidemiology, and End Results (SEER) Program. SEER*Stat Database: Incidence - 24 437 SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases Nov 2011 Sub (1973- 25 438 2009 varying) – Linked to County Attributes – Total U.S., 1969-2010 Counties. 26 439 25. Sankaranarayanan R, Swaminathan R, Brenner H, et al. Cancer survival in Africa, Asia, and 27 28 440 Central America: a population-based study. Lancet Oncol 2010;11(2):165-173. 29 441 26. Stinnett AA and Paltiel AD. Estimating CE ratios under second-order uncertainty: the mean ratio 30 442 versus the ratio of means. Med Decis Making 1997;17(4):483-489. 31 443 27. Mayrand MH, Duarte-Franco E, Rodrigues I, et al. Human papillomavirus DNA versus 32 444 Papanicolaou screening tests for cervical cancer. N Engl J Med 2007;357(16):1579-1588.

33 445 28. Ronco G, Dillner J, Elfstrom KM, et al. Efficacy of HPV-based screening for prevention of invasive http://bmjopen.bmj.com/ 34 446 cervical cancer: follow-up of four European randomised controlled trials. Lancet 35 36 447 2014;383(9916):524-532. 37 448 29. Campos NG, Maza M, Alfaro K, et al. The comparative and cost-effectiveness of HPV-based 38 449 cervical cancer screening algorithms in El Salvador. Int J Cancer 2015. 39 450 30. Chirenje ZM, Rusakaniko S, Kirumbi L, et al. Situation analysis for cervical cancer diagnosis and 40 451 treatment in east, central and southern African countries. Bull World Health Organ 41 452 2001;79(2):127-132. on September 26, 2021 by guest. Protected copyright. 42 453 31. Cremer M, Bullard K, Maza M, et al. Cytology versus visual inspection with acetic acid among 43 44 454 women treated previously with cryotherapy in a low-resource setting. Int J Gynaecol Obstet 45 455 2010;111(3):249-252. 46 456 32. Sauvaget C, Muwonge R, and Sankaranarayanan R. Meta-analysis of the effectiveness of 47 457 cryotherapy in the treatment of cervical intraepithelial neoplasia. Int J Gynaecol Obstet 48 458 2013;120(3):218-223. 49 459 33. Bigoni J, Gundar M, Tebeu PM, et al. Cervical cancer screening in sub-Saharan Africa: a 50 51 460 randomized trial of VIA versus cytology for triage of HPV-positive women. Int J Cancer 52 461 2015;137(1):127-134. 53 462 34. Catarino R, Vassilakos P, Scaringella S, et al. Smartphone Use for Cervical Cancer Screening in 54 463 Low-Resource Countries: A Pilot Study Conducted in Madagascar. PLoS One 55 464 2015;10(7):e0134309. 56 57 58 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 465 35. Pan QJ, Hu SY, Guo HQ, et al. Liquid-based cytology and human papillomavirus testing: a pooled 4 466 analysis using the data from 13 population-based cervical cancer screening studies from China. 5 6 467 Gynecol Oncol 2014;133(2):172-179. 7 468 36. Qiao YL, Jeronimo J, Zhao FH, et al. Lower cost strategies for triage of human papillomavirus 8 469 DNA-positive women. Int J Cancer 2014;134(12):2891-2901. 9 470 37. Tebeu PM, Fokom-Domgue J, Crofts V, et al. Effectiveness of a two-stage strategy with HPV 10 471 testing followed by visual inspection with acetic acid for cervical cancer screening in a low- 11 472 income setting. Int J Cancer 2015;136(6):E743-750. 12 13 473 38. Richardson LA, El-Zein M, Ramanakumar AV, et al. HPV DNA testing with cytology triage in 14 474 cervical cancer screening: Influence of revealing HPV infection status. Cancer Cytopathol 15 475 2015;123(12):745-754.For peer review only 16 476 39. World Development Indicators, 2016, World Bank. 17 477 40. Msyamboza KP, Phiri T, Sichali W, et al. Cervical cancer screening uptake and challenges in 18 478 Malawi from 2011 to 2015: retrospective cohort study. BMC Public Health 2016;16(1):806. 19 479 41. Campbell C, Kafwafwa S, Brown H, et al. Use of thermo-coagulation as an alternative treatment 20 21 480 modality in a 'screen and treat' programme of cervical screening in rural Malawi. Int J Cancer 22 481 2016. 23 482 42. Ronco G, Dillner J, Elfstrom KM, et al. Efficacy of HPV-based screening for prevention of invasive 24 483 cervical cancer: follow-up of four European randomised controlled trials. Lancet 25 484 2014;383(9916):524-532. 26 485 43. Starks D, Arriba LN, Enerson CL, et al. Mexican Cervical Cancer Screening Study II: 6-month and 27 28 486 2-year follow-up of HR-HPV women treated with cryotherapy in a low-resource setting. J Low 29 487 Genit Tract Dis 2014;18(4):333-337. 30 488 44. McClung EC and Blumenthal PD. Efficacy, safety, acceptability and affordability of cryotherapy: a 31 489 review of current literature. Minerva Ginecol 2012;64(2):149-171. 32 490 33 http://bmjopen.bmj.com/ 34 35 491 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 22 of 56

------75% 1.0/1.0 0.40/0.85 0.70/0.65 0.70/0.95 0.40/0.85 0.90/0.50 40% - 85% - 40% 40% - 85% - 40% 50% - 80% - 50% screening screening intervals screening intervals screening screening intervals Sensitivity Sensitivity Analysis: 0.95/0.68, with

40% 85% 92% 70% 22 22 0.95/0.68 0.41/0.94 0.67/0.86 0.78/0.89 treatment treatment CIN2: 85%CIN2: 75%CIN3: treatment Base Case: BMJ Open http://bmjopen.bmj.com/

Upon ruling out cancer, all referred to Upon ruling out cancer, all referred to a 100% ≤CIN1: on September 26, 2021 by guest. Protected copyright. [29] 0.60/0.75 0.85/0.55

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

For peer review only

c [14, 33, 35,[14, 38] 33, [14, 33, 34, 36, 37]

b Test/Parameter

[14, 27, 42] [14,

[30-32] [6] [14, 17]

[14] [14] Screen-and-treat cryotherapy HPV+ rate cure (for or HPV+/VIA+) VIA (triage of (triage VIA HPV+) Visit compliance,Visit referral facility Pap (primary) Colposcopy (HPV+ Colposcopy (HPV+ women ineligible for cryotherapy) Colposcopy (ASCUS+ women) to routineAccess screening, theof % target population Visit compliance,Visit screening facility careHPV (primary),careHPV provider-collection vaginalof samples Colposcopy (HPV+/VIA+ women ineligible ST for cryotherapy) Colposcopy (HPV+/Pap+ women) Upon ruling out cancer, all referred to Pap HPV+)of (triage careHPV (primary),careHPV provider-collection cervicalof samples Eligibility screen-and-treatfor cryotherapy Table 1. Baseline values and ranges for model variables. model for ranges and values Baseline 1. Table Screening/triage Screening/triage test performance (sensitivity/specificity to CIN2+)detect Treatment eligibility and efficacy Colposcopy performance (sensitivity/specificity to CIN1+)detect andCoverage compliance 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

f e

3 3 ------85% 6.91 30.54 0% to 50% ofto 0% 50% case base 0% to 50% ofto 0% 50% case base ofto 0% 50% case base ofto 0% 50% case base ofto 0% 50% case base 0% to 50% ofto 0% 50% case base 50% - 150% - 50% case base of 75% - 125% - 75% case base of 50% - 150% - 50% case base of 50% - 150% - 50% case base of 50% - 150% - 50% case base of -- --

944 197 346 96% 4.19 1.75 0.82 3.10 0.48 7.26 2.81 0.41 19.91 68.36 11.96 11.04 18.16 23 23 collection) cryotherapy BMJ Open http://bmjopen.bmj.com/ 10% 15% 1.51 per woman1.51 receiving VIA and/or 0.04 per woman0.04 screened with careHPV 0.08 per woman0.08 screened with (self-careHPV

d [44] on September 26, 2021 by guest. Protected copyright.

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml [29]

For peer review only [30]

[43] d

[14]

Direct medicalDirect Direct non-medicalDirect Women’s timeWomen’s

Colposcopy/biopsy LEEP LEEP Local cancer - - careHPV testcareHPV (provider-collection) testcareHPV (self-collection) VIA triage testtriage VIA Proportion maintaining women of an HPV infection colposcopicfollowing diagnosis and treatment Cryotherapy Outreach workers/auxiliary nurses (careHPV self- collection) Laboratory technicians (careHPV) Health personnelcare (careHPV) per woman 0.09 screened with careHPV - Treatment rate cure following colposcopy (LEEP for CIN2+; cryotherapy)else Proportion maintaining women of an HPV infection cryotherapyfollowing Transportation to screening facility (round-trip) Wait referral time, facility Transport time, screening facility Transport time, referral facility Transportation to referral facility (round trip) Wait screening time, facility Health providerscare (VIA and cryotherapy) Pap test Cost of cancerof Cost treatment (2015 US$) Programmatic costs, training US$) (2015 Direct medical Direct costs, screening and of treatment (2015precancer US$) Women’s time Women’s and transportation (2015 US$) costs

Page 23 of 56 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 24 of 56

case) case) case) case) 918 - 2,290 - 918 (direct only;medical of 150% base 944 - 2,2920 - 944 (direct medical baseonly;of 150%

918 390 640 1,946 1,486 24 24 BMJ Open http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

Direct non-medicalDirect Women’s timeWomen’s Direct medicalDirect

- Total - Regional andRegional distant cancer - Total Test performance characteristics of colposcopy in the START-UP demonstration project were derived from the worst diagnosis of the local the local of diagnosis the worst from derived were project demonstration START-UP the in colposcopy of characteristics Test performance In sensitivity analysis, we considered the direct medical cost of colposcopy to be equivalent to the cost of colposcopy alone, without biopsy. biopsy. without alone, colposcopy of the cost to equivalent be to of colposcopy cost medical the direct considered we analysis, sensitivity In Unpublished data from the Scale-Up Nicaragua project. Further details on costing data are provided in the Appendix. Appendix. in the are provided data on costing details Further project. Nicaragua Scale-Up the from data Unpublished meals. and housing, temporary facility, tertiary to a transportation include costs Direct non-medical Compliance is defined as the proportion of women who return for each clinical encounter, relative to the previous visit. visit. previous the to relative encounter, clinical each for return who women of proportion the as defined is Compliance In sensitivity analysis, we considered the direct medical cost of cryotherapy to include the upper bound of cryotherapy equipment costs costs equipment cryotherapy of bound the upper include to of cryotherapy cost medical the direct considered we analysis, sensitivity In Includes woman’s time an support person’s time. time. person’s support an time woman’s Includes ASCUS+: atypical squamous cells of undetermined significance or higher; CIN 1+: cervical intraepithelial neoplasia grade 1 or higher; CIN2+: CIN2+: higher; or 1 grade neoplasia intraepithelial cervical 1+: CIN higher; or significance of undetermined cells squamous atypical ASCUS+: that were inadequate or with a histological classification other than negative, CIN1, CIN2, CIN3, or cancer. Because CIN1 is not a true underlying underlying a true not is CIN1 Because cancer. or CIN3, CIN2, CIN1, negative, than other classification a histological with or inadequate were that HPV no lesion, of states health underlying on the based is model the in colposcopy of performance model, microsimulation the in state health country- on the based HPV with women for of colposcopy sensitivity weighted we CIN1, of threshold a treatment For or CIN3. CIN2, infection, studies. START-UP the in infections HPV with women among of CIN1 prevalence specific (assuming the lowest number of women treated per year per facility, in Carazo, Chontales, or Chinandega) and the upper bound of cryotherapy of cryotherapy bound the upper and Chinandega) or Chontales, Carazo, in facility, year per per treated women of number lowest the (assuming or Chinandega). Chontales, Carazo, in facility any in tank gas per treated women of number lowest the (assuming costs supply

a b c d e f g h cervical intraepithelial neoplasia grade 2 or higher; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; ST: screen-and- ST: procedure; excision electrosurgical LEEP: loop papillomavirus; human HPV: higher; or 2 grade neoplasia intraepithelial cervical acid. acetic with inspection visual VIA: dollars; States United 2015 US$: treat; pathologist relative to the worst diagnosis by a quality control pathologist (gold standard); we applied the treatment threshold of CIN1+, CIN1+, of threshold the treatment we applied standard); (gold pathologist control quality by a diagnosis worst the to relative pathologist classifications histological excluded we colposcopy, of performance test derive To START-UP. in threshold treatment the was not this although 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 25 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Figure legends. 4 5 6 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible 7 8 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, 9 10 conditional on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 11 12 13 visits for screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-ST), which 14 15 requires 3 or moreFor visits for screeningpeer and necessary review treatment; c) HPVonly testing followed by visual 16 17 inspection with acetic acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 18 19 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV- 20 21 22 positive women every 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary 23 24 treatment. CIN: cervical intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical 25 26 excision procedure; VIA: visual inspection with acetic acid. 27 28 29 Figure 2. Cervical cancer screening cost per woman over duration of screening eligibility, by cost 30 31 component: Pap testing (every 3 years) versus careHPV testing (every 5 years). Bars indicate the 32

33 undiscounted cost (2015 US$) of screening with Pap testing (offered 10 times between ages 30 and 59 http://bmjopen.bmj.com/ 34 35 36 years) versus careHPV testing (offered 6 times between ages 30 and 59 years), by cost component. Only 37 38 screening costs are shown; costs associated with recommended management following a positive 39 40 screening test are not included. 6x: delivered 6 times over the course of screening eligible ages 30 to 59; 41 on September 26, 2021 by guest. Protected copyright. 42 10x: delivered 10 times over the course of screening eligible ages. 43 44 45 Figure 3. Cost-effectiveness analysis: Base case. The graph displays the discounted lifetime costs (x-axis; 46 47 in 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing every 3 48 49 years; careHPV screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual inspection 50 51 52 with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap 53 54 triage of HPV-positive women [HPV-Pap]), under base case assumptions. The cost-effectiveness 55 56 associated with a change from one strategy to a more costly alternative is represented by the difference 57 58 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 in cost divided by the difference in life expectancy associated with the two strategies. The curve 4 5 6 indicates the strategies that are efficient because they are more effective and either 1) cost less; or 2) 7 8 have a more attractive cost-effectiveness ratio than less effective options. The incremental cost- 9 10 effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two strategies under 11 12 13 comparison. In the base case analysis, HPV-ST every 5 years was less costly and more effective than 14 15 other screening strategiesFor considered, peer and was review thus a dominant strategy only with an ICER of US$320 per 16 17 year of life saved. HPV-Pap: HPV testing with Pap triage of HPV-positive women; HPV-ST: HPV screen- 18 19 and-treat strategy; HPV-VIA: HPV testing with visual inspection with acetic acid (VIA) triage of HPV- 20 21 22 positive women; yrs: years. 23 24 Figure 4. Reduction in lifetime risk of cervical cancer, by compliance level. Bars indicate the percent 25 26 reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 years; 27 28 29 careHPV screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual inspection with 30 31 acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of 32

33 HPV-positive women [HPV-Pap]) as compliance per visit within a screening episode increases. http://bmjopen.bmj.com/ 34 35 36 Compliance is defined as the proportion of women who return for each clinical encounter, relative to 37 38 the previous visit. Coverage of the target population is assumed to be 70%. While the base case analysis 39 40 assumed 85% compliance for visits at screening facilities and 40% compliance for visits at referral 41 on September 26, 2021 by guest. Protected copyright. 42 facilities (for diagnosis and treatment), the graph displays cancer risk reduction assuming the specified 43 44 45 compliance level at all visits, regardless of facility type. 46 47 Figure 5. Base case and sensitivity analysis: Incremental cost-effectiveness ratios, HPV screen-and- 48 49 treat (HPV-ST) strategy. Incremental cost-effectiveness ratios (ICERs) are presented (x-axis, 2015 US$ 50 51 52 per year of life saved) for the base case and sensitivity analyses (y-axis). The blue bars represent the 53 54 range of the ICERs for HPV-ST every 5 years across the 50 input parameter sets, with the ICER of the 55 56 mean costs divided by the mean effects demarcated by a black line. The dashed blue line indicates 57 58 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Nicaragua’s per capita gross domestic product (GDP), at US$2,090, assuming this is the threshold that 4 5 6 designates interventions as “very cost-effective”. 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1A. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-ST), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. CIN: cervical 39 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 40 visual inspection with acetic acid. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1B. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-ST), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. CIN: cervical 39 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 40 visual inspection with acetic acid. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1C. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-ST), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. CIN: cervical 39 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 40 visual inspection with acetic acid. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1D. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-ST), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. CIN: cervical 39 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 40 visual inspection with acetic acid. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 2. Cervical cancer screening cost per woman over duration of screening eligibility, by cost

33 component: Pap testing (every 3 years) versus careHPV testing (every 5 years). Bars indicate the http://bmjopen.bmj.com/ 34 undiscounted cost (2015 US$) of screening with Pap testing (offered 10 times between ages 30 and 59 35 years) versus careHPV testing (offered 6 times between ages 30 and 59 years), by cost component. Only 36 screening costs are shown; costs associated with recommended management following a positive screening 37 test are not included. 6x: delivered 6 times over the course of screening eligible ages 30 to 59; 10x: delivered 10 times over the course of screening eligible ages. 38

33 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing every 3 years; http://bmjopen.bmj.com/ 34 careHPV screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual inspection with acetic 35 acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of HPV- 36 positive women [HPV-Pap]), under base case assumptions. The cost-effectiveness associated with a change 37 from one strategy to a more costly alternative is represented by the difference in cost divided by the difference in life expectancy associated with the two strategies. The curve indicates the strategies that are 38 efficient because they are more effective and either 1) cost less; or 2) have a more attractive cost- 39 effectiveness ratio than less effective options. The incremental cost-effectiveness ratio (ICER) is the 40 reciprocal of the slope of the line connecting the two strategies under comparison. In the base case analysis, 41 HPV-ST every 5 years was less costly and more effective than other screening strategies considered, and on September 26, 2021 by guest. Protected copyright. 42 was thus a dominant strategy with an ICER of US$320 per year of life saved. HPV-Pap: HPV testing with Pap 43 triage of HPV-positive women; HPV-ST: HPV screen-and-treat strategy; HPV-VIA: HPV testing with visual 44 inspection with acetic acid (VIA) triage of HPV-positive women; yrs: years. 45 254x190mm (300 x 300 DPI) 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 4. Reduction in lifetime risk of cervical cancer, by compliance level. Bars indicate the percent

33 reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 years; careHPV http://bmjopen.bmj.com/ 34 screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual inspection with acetic acid 35 [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of HPV-positive 36 women [HPV-Pap]) as compliance per visit within a screening episode increases. Compliance is defined as 37 the proportion of women who return for each clinical encounter, relative to the previous visit. Coverage of the target population is assumed to be 70%. While the base case analysis assumed 85% compliance for 38 visits at screening facilities and 40% compliance for visits at referral facilities (for diagnosis and treatment), 39 the graph displays cancer risk reduction assuming the specified compliance level at all visits, regardless of 40 facility type. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 5. Base case and sensitivity analysis: Incremental cost-effectiveness ratios, HPV screen-and-treat

33 (HPV-ST) strategy. Incremental cost-effectiveness ratios (ICERs) are presented (x-axis, 2015 US$ per year http://bmjopen.bmj.com/ 34 of life saved) for the base case and sensitivity analyses (y-axis). The blue bars represent the range of the 35 ICERs for HPV-ST every 5 years across the 50 input parameter sets, with the ICER of the mean costs divided 36 by the mean effects demarcated by a black line. The dashed blue line indicates Nicaragua’s per capita gross 37 domestic product (GDP), at US$2,090, assuming this is the threshold that designates interventions as “very cost-effective”. 38

39 254x190mm (300 x 300 DPI) 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 Supplemental Appendix: 4 2 The Cost-Effectiveness of HPV-Based Screening in the Public Health System in Nicaragua 5 6 7 3 8 4 MODEL CALIBRATION 9 5 Overview of the calibration process 10 6 Details of model development for the U.S. and Nicaragua models, including initial parameterization and 11 7 calibration, have been published elsewhere [1, 2]. Baseline HPV incidence rates, as a function of 12 8 genotype and age, were derived from a published prospective study of sexually active women aged 15- 13 14 9 85 years in Bogota, Colombia [3]. Because HPV incidence is not necessarily similar in Colombia and 15 10 Nicaragua due to Fordifferences inpeer sexual behavior, review we considered age-specific only HPV incidence and natural 16 11 immunity following initial infection as candidate parameters for calibration. Type-specific data on CIN2 17 12 and CIN3 regression and progression are limited [4-9], so these highly uncertain parameters were also 18 13 candidates for calibration. Time-dependent rates of HPV clearance and progression by genotype were 19 14 informed by primary longitudinal data from the control arm of the Costa Rica Vaccine Trial [10], 20 15 assuming that rates of HPV clearance and progression to precancer are similar across populations and 21 22 16 vary only by time since infection and HPV genotype. 23 17 24 18 To calibrate the model, we selected empirical epidemiologic data from Nicaragua, including age-specific 25 19 prevalence of high-risk HPV and age-specific cancer incidence (see below for further detail). We then set 26 20 plausible search ranges around baseline model input values for age- and type-specific HPV incidence; 27 21 natural immunity following initial infection; and progression and regression of CIN. We performed 28 29 22 repeated model simulations in the absence of any preventive intervention (i.e., natural history 30 23 simulations). For each simulation, we randomly selected a single value for each of the uncertain 31 24 parameters from the identified search range, creating a unique vector of parameter values (i.e., 32 25 parameter “set”). From over 1,475,000 repeated samplings, we identified the parameter sets with the

33 26 highest correspondence to the empirical calibration target data from Nicaragua by calculating and http://bmjopen.bmj.com/ 34 27 aggregating the log-likelihood of model-projected outcomes. We used the 50 parameter sets with the 35 28 highest likelihood scores (i.e., best overall fit to the empirical data from Nicaragua) for analysis to 36 37 29 capture uncertainty in the model parameters as a form of probabilistic sensitivity analysis. We report 38 30 results as a mean of outcomes across these top 50 parameter sets; incremental cost-effectiveness ratios 39 31 are reported as the ratio of the mean costs divided by the mean effects of one strategy versus another 40 32 across sets. 41 33 on September 26, 2021 by guest. Protected copyright. 42 34 Calibration targets 43 35 We assessed model fit by comparing projected model outcomes of age-specific prevalence of high-risk 44 45 36 HPV and age-specific cancer incidence relative to empirical data from Nicaragua. The scoring algorithm 46 37 included age-specific prevalence of high-risk HPV and age-specific cervical cancer incidence. 47 38 48 39 Age-specific prevalence of high-risk HPV was drawn from the Screening Technologies to Advance Rapid 49 40 Testing for Cervical Cancer Prevention–Utility and Program Planning (START-UP) project data on 50 41 careHPV positivity using a cut-off ratio cut-point of 0.5 relative light units (Table A.1). For each age 51 52 42 group, we derived a 95% binomial confidence interval around the point prevalence, which comprised 53 43 the calibration target. The likelihood function for each age group was assumed to follow a binomial 54 44 distribution. 55 45 56 46 Age-specific cancer incidence was drawn from Globocan due to the lack of a local cancer registry [11] 57 47 (Table A.2). The likelihood function for each age group was assumed to follow a normal distribution. 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 48 4 49 Composite goodness-of-fit scores for each input parameter set were generated by summing the log 5 6 50 likelihood of each model outcome (i.e., age-specific HPV prevalence, age-specific cancer incidence). The 7 51 50 input parameter sets with the highest goodness-of-fit scores yielded the model outputs that were 8 52 simultaneously closest to all calibration targets, and these were selected for analysis. Figures A.1 and 9 53 A.2 display model fit to epidemiologic data on age-specific prevalence of high-risk HPV and age-specific 10 54 cancer incidence. 11 55 12 13 56 COST DATA 14 57 15 58 Direct Medical Costs:For Screening, peer Diagnosis, andreview Treatment of Precancerous only Lesions 16 59 The direct medical costs of screening, diagnosis, and treatment of precancerous lesions were drawn 17 60 from the START-UP demonstration study in Masaya department and from the Scale-Up Nicaragua 18 61 Project in the departments of Carazo, Chontales, and Chinandega. Direct medical costs included clinical 19 62 staff time, clinical supplies, drugs, clinical equipment, laboratory staff time, laboratory supplies, and 20 21 63 laboratory equipment. Costs from the START-UP study were collected in 2010 local currency units and 22 64 inflated to year 2015 levels using Nicaragua GDP deflators and then converted to US$ using the official 23 65 exchange rate [12]; the exception was for equipment, which was generally procured in the United States 24 66 and was inflated to year 2015 levels using US GDP deflators. Costs from the Scale-Up Nicaragua project 25 67 (including equipment) were collected in 2015 local currency units and converted to US$ using the official 26 68 exchange rate [12]. 27 28 69 29 70 Costs are reported in Table A.3 (screening procedures) and Table A.4 (precancer diagnostic and 30 71 treatment procedures). 31 72 32 73 Women’s Time and Transportation Costs

33 74 To account for the opportunity cost of women’s time spent traveling to, waiting for, or receiving care, http://bmjopen.bmj.com/ 34 75 we valued women’s time using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 US$. 35 36 76 While the minimum wage may be a conservative estimate for valuation of women’s time if they are 37 77 formally employed, it can serve as a proxy for the societal value of women’s time spent working in the 38 78 informal sector or at home. We converted this to an hourly wage rate assuming 171 hours of work per 39 79 month (40 hours per week). 40 80 41 81 Estimates for time spent traveling to, waiting for, and receiving care were dependent upon the facility on September 26, 2021 by guest. Protected copyright. 42 82 level where care was assumed to take place (Table A.5). While screening and delivery of primary 43 44 83 screening results were assumed to take place at basic primary health facilities, we assumed cryotherapy 45 84 and LEEP treatment were only provided at referral facilities. Estimates of women’s time spent receiving 46 85 a procedure were based on site-specific data from the START-UP demonstration projects, with staff time 47 86 spent on the procedure (excluding preparation and registration time, which we assumed were built into 48 87 patient waiting time) used as a proxy for women’s procedure time. Estimates of women’s wait time for 49 88 screening and delivery of primary screening results was based on the average wait time for screening at 50 51 89 facilities in the Chinandega department; wait time for cryotherapy, colposcopy, and LEEP was based on 52 90 the average wait time at referral centers in the Chontales department. Round-trip transportation time 53 91 was based on Scale-Up project staff estimates of the typical amount of time women spent traveling to 54 92 screening or referral facilities, using a sample of 9 facilities the Carazo department and 10 facilities in the 55 93 Chontales and Chinandega departments. We averaged across screening and referral facilities within 56 94 each department to obtain a department average, and then averaged across departments to obtain 57 58 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 95 average round-trip transportation times based on facility level. Estimates of women’s time and time 4 96 costs are presented in Table A.6. 5 6 97 7 98 Women’s transportation costs were based on Scale-Up project staff reports of the typical mode of 8 99 transportation to a particular facility by women in the catchment area, using a sample of 9 facilities in 9 100 Carazo and 10 facilities each in Chontales or Chinandega. We used the most commonly reported mode 10 101 of transportation for each department (for screening facilities: walking for Carazo; walking for 11 102 Chontales; bus for Chinandega; for referral facilities: bus in all three departments). The cost of a bus trip 12 13 103 was based on typical bus fare for the distance to the health facility, as reported by Scale-Up project staff. 14 104 We averaged the typical transportation costs across departments to obtain an estimate of women’s 15 105 round-trip transportationFor costs peer to and from a healthreview facility, assuming only the same mode of transport was 16 106 used for the trip to and from the health facility. Women’s transportation costs are presented in Table 17 107 A.7. 18 108 19 109 Programmatic Costs 20 21 110 We included the programmatic cost of the following types of training sessions that were conducted 22 111 during the Scale-Up project: 1) for health care personnel to offer careHPV screening and counsel women 23 112 on results; 2) for laboratory technicians to run the careHPV assay; 3) for community outreach workers 24 113 and auxiliary nurses to promote careHPV screening and offer self-collection of HPV samples; and 4) for 25 114 health care providers to offer VIA and cryotherapy. We included the cost of trainer time and trainee 26 115 time (valued using average monthly salaries converted to an hourly rate; volunteer outreach worker 27 28 116 time was valued using the minimum wage), as well as the cost of supplies, materials, transportation, and 29 117 refreshments. To derive an average health care personnel training cost per woman screened, we divided 30 118 the total training cost for health care personnel by 100,000, corresponding to the number of women 31 119 screened during the Scale-Up project. To derive an average laboratory technician training cost per 32 120 woman screened, we divided the total training cost for laboratory technicians by 100,000. To derive an

33 121 average community outreach worker training cost per woman screened, we divided the total training http://bmjopen.bmj.com/ 34 122 cost for outreach workers by 80,000, corresponding to the number of women who participated in self- 35 36 123 collection of HPV samples. To derive an average health care provider training cost per woman receiving 37 124 VIA and cryotherapy, we divided the total VIA/cryotherapy training cost for health care providers by 38 125 16,000, corresponding to the approximate number of women who tested HPV-positive in Scale-Up and 39 126 would thus require visual assessment (to determine eligibility for cryotherapy) and, if eligible, 40 127 cryotherapy. 41 128 on September 26, 2021 by guest. Protected copyright. 42 129 All women screened with HPV testing incurred the per-woman training costs for health care personnel 43 44 130 and laboratory technicians. Only women screened with HPV self-collection incurred the per-woman cost 45 131 of training outreach workers to offer self-collection (due to lack of data on self-collection in a 46 132 community setting, in the base case analysis we assumed self-collection took place at the clinic; 47 133 however, we assumed outreach workers would be involved in mobilizing and educating women about 48 134 self-collection, and conservatively included the cost of training so as to represent a known 49 135 programmatic cost as self-collection efforts shift to also include community settings). Only women who 50 51 136 received either VIA (as triage) or screen-and-treat cryotherapy incurred the per-woman cost of training 52 137 health providers to perform VIA and cryotherapy. Training costs are presented in Table A.8. 53 138 54 139 Cost of Cancer Care by Stage 55 140 Costs associated with cancer care by stage (Local versus Regional or Distant), including direct medical 56 141 costs, direct non-medical costs (i.e., transport to a cancer center, and meals and temporary housing 57 142 during the course of treatment), and patient and support person time were derived from a previous 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 143 analysis in El Salvador [14]. Based on expert opinion in El Salvador, we assumed 50% of local cancer 4 144 patients presented with FIGO stage 1a, requiring radical hysterectomy, and 50% of local cancer patients 5 6 145 presented with FIGO stage 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions 7 146 followed by brachytherapy. We assumed all women presenting with FIGO stage 2b or higher received 8 147 cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. 9 148 10 149 To adjust the direct medical costs of cancer treatment in El Salvador to the setting of Nicaragua, we 11 150 multiplied by the ratio of WHO-CHOICE inpatient bed-day costs at a teaching hospital in Nicaragua 12 13 151 relative to El Salvador. We assumed each patient and a support person spent the same amount of time 14 152 traveling, waiting for, and receiving care as in El Salvador, and valued this time at the 2013 minimum 15 153 wage rate in NicaraguaFor (inflated peer and converted review to 2015 US$). To adjust only temporary housing and 16 154 transportation costs to the setting of Nicaragua, we multiplied the Salvadoran costs by the ratio of GDP 17 155 per capita in Nicaragua relative to El Salvador. Costs are reported in Table A.9. 18 156 19 157 SUPPLEMENTARY RESULTS 20 21 158 Additional results described in the main manuscript are presented in Figures A3 to A5. 22 159 23 160 24 161 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 162 Table A.1. Age-specific prevalence of high-risk HPV, Nicaragua [15].a 4 5 Age group Number of women Number of women Prevalence (95% CI) 6 with high-risk HPV 7 30 – 34 years 1,693 310 0.18 (0.17, 0.20) 8 35 – 39 years 1,141 184 0.16 (0.14, 0.18) 9 40 – 44 years 933 125 0.13 (0.11, 0.16) 10 45 – 49 years 878 121 0.14 (0.12, 0.16) 11 a 12 163 HPV positivity was based on a cut-off of 0.5 relative light units. 13 14 164 15 165 Table A.2. Age-specificFor cervical peer cancer incidence, review Nicaragua (GLOBOCAN only 2012) [11]. 16 Age group Cases Rate per 100,000 women (95% 17 CI) 18 40 – 44 years 123 78.7 (64.8, 92.6) 19 20 45 – 49 years 112 85.4 (69.6, 101.2) 21 50 – 54 years 102 88.4 (71.2, 105.6) 22 55 – 59 years 85 88.1 (69.4, 106.8) 23 60 – 64 years 51 84.0 (61.0, 107.1) 24 65 – 69 years 37 80.8 (54.8, 106.8) 25 70 – 74 years 30 74.6 (47.9, 101.3) 26 ≥75 years 45 70.3 (49.8, 90.8) 27 28 166 29 167 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 6 6 START- c START-UP d

other) Source ( START-UP START-UP START-UP (other) e UP (Cytology/VIA)UP personnel); fuel);

0.00 Scale-Up (transport 0.00 0.00 Scale-Up (transport 3.81 0.02 0.36 Scale-Up (HPV); 4.19

1.37 0.92 3.81 0.82 7.26 Cytology triage VIA

b a a

0.41 0.81 2.88 6.72 BMJ Open 11.04 http://bmjopen.bmj.com/ HPV test HPV (self-collection)

b on September 26, 2021 by guest. Protected copyright.

0.41 0.81 3.81 6.72 11.96 HPV test HPV For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml provider-collection) ( For peer review only The cost of laboratory staff time was drawn from the START-UP project, with the exception of preparing and transporting samples to the the to samples transporting and of preparing exception the with project, START-UP the from drawn was time staff laboratory of cost The For HPV testing, all lab supplies are bundled in the category of clinical supplies, with the exception of fuel for laboratory transport, which is is which transport, for laboratory of fuel exception the with supplies, of clinical category the in are bundled supplies lab all testing, For HPV HPV testing with both provider-collection of cervical samples and self-collection of vaginal samples was assumed to take place at the clinic. The The clinic. at the place take to assumed was samples vaginal of self-collection and samples of cervical provider-collection both with testing HPV HPV: human papillomavirus; Scale-Up: Scale-Up demonstration project (2015); START-UP: Screening Technologies to Advance Rapid Testing for for Testing Rapid Advance to Technologies Screening START-UP: (2015); project demonstration Scale-Up Scale-Up: papillomavirus; human HPV: The cost of clinical supplies for HPV testing was drawn from the Scale-Up project and includes the test kit, plate sealers, brush, collection collection brush, sealers, plate kit, test the includes and project Scale-Up the from was drawn testing for HPV supplies of clinical cost The Clinical equipmentClinical 0.01 0.01 0.01 Lab staff estimates of the number of samples transported per load, the distance to the lab from 9 to 10 facilities in each department, and the usual type the usual and department, each in 10 facilities 9 to from lab the to distance the load, per transported samples of the number of estimates [16] and types vehicle major of economy fuel the average using sample per cost fuel the average we derived department, each in used vehicle of [12]. Nicaragua in liter per price gas the average medium, pipette tips (3 types), gloves, and towels to decant and dry. dry. and decant to towels and gloves, types), (3 tips pipette medium, laboratory, which was drawn from the Scale-Up project. Staff time spent preparing samples for transport in the Scale-Up project was based on based was project the Scale-Up in transport for samples preparing spent time Staff project. the Scale-Up from was drawn which laboratory, sample. per minutes 7 to 5 from ranged which Chinandega), and Chontales, (Carazo, departments the three of each in estimates staff project Table A.3. Screening: Direct medical costs per procedure (2015 US$). (2015 procedure per costs medical Direct Screening: A.3. Table Cost category Lab equipment 0.21 0.21 0.32 Total Total direct medical cost a b c d e Clinical suppliesClinical Clinical staffClinical Cervical Cancer Prevention–Utility and Program Planning demonstration project (2010); US$: United States dollars; VIA: visual inspection with with inspection visual VIA: dollars; States United US$: (2010); project demonstration Planning Program and Prevention–Utility Cancer Cervical acid. acetic Lab supplies included under lab supplies. For both HPV and cytology testing, we assumed the same cost of fuel per sample. Using Scale-Up project staff staff project Scale-Up Using sample. per of fuel cost same the assumed we testing, cytology and HPV For both supplies. lab under included Staff time for round-trip transport of samples was averaged across 9 to 10 facilities in each of the three departments, and valued based on the on the based valued and departments, three the of each in facilities 10 to 9 across averaged was of samples transport round-trip for time Staff difference in costs between collection modalities is due to reduced clinical staff time when a woman self-collects her own sample. sample. own her self-collects a woman when time staff clinical reduced to is due modalities collection between costs in difference average monthly wage of the personnel responsible for transporting the samples in each department. We derived an average cost per sample sample per cost average an derived We department. each in samples the transporting for responsible personnel of the wage monthly average and specimens), cytology and for HPV equivalent was sample per time of staff cost the (assuming department each for transported and prepared then averaged these costs across the three departments. departments. three the across costs these averaged then Page 41 of 56 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 42 of 56 7 7

Source START-UP START-UP START-UP START-UP START-UP START-UP (other)START-UP START-UP (other) c b gas); unit);

0.10 4.09 4.12 3.35 7.34 LEEP 13.29 Scale-Up (cryotherapy 36.05 Scale-Up (cryotherapy 68.36

0.30 0.00 0.00 0.00 7.54 9.13 3.54 18.16

0.13 4.09 4.12 3.35 6.29 0.53 1.39 BMJ Open 19.91 http://bmjopen.bmj.com/ Colposcopy and biopsy Cryotherapy on September 26, 2021 by guest. Protected copyright.

6.91 0.00 0.00 0.00 5.92 0.49 0.38 Colposcopy For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only We used estimates from Scale-Up project staff to inform the cost of gas used per cryotherapy procedure, based on the cost of 1 cylinder and and cylinder 1 of the cost on based procedure, cryotherapy per ofused gas cost the inform to staff project Scale-Up from estimates We used To estimate the cost per procedure of a cryotherapy unit, we assumed this equipment item was a tradable good, and accordingly converted the the converted accordingly good, and tradable was a item equipment this assumed we unit, a cryotherapy of procedure per cost the estimate To LEEP: loop electrosurgical excision procedure; Scale-Up: Scale-Up demonstration project (2015); START-UP: Screening Technologies to Advance Advance to Technologies Screening START-UP: (2015); project demonstration Scale-Up Scale-Up: procedure; excision electrosurgical loop LEEP: Drugs 0.13 Clinical equipmentClinical Lab supplies Lab equipment Total Total direct medical cost Lab staff US$). (2015 procedure per costs medical Direct Precancer: of Treatment and Diagnosis A.4. Table a b c Clinical suppliesClinical Clinical staffClinical Cost category Rapid Testing for Cervical Cancer Prevention–Utility and Program Planning demonstration project (2010); US$: United States dollars. dollars. States United US$: (2010); project demonstration Planning Program and Prevention–Utility Cancer Cervical for Testing Rapid purchase price from local currency units to US$ using the official exchange rate. We annualized the cost with a 3% interest rate and assumed an an assumed and rate interest a 3% with cost the We annualized rate. exchange the official using US$ to units currency local from price purchase the divided we considered, departments three the across year per machine per of procedures number the derive To years. 5 of life economic by the cost unit cryotherapy the annualized divided We then units. cryotherapy of number the by procedures of cryotherapy number annual the number of women treated per tank. We did not include the cost of the tank deposit. deposit. tank of the cost the include not did We tank. per treated women of the number number of procedures per machine per year. year. per machine per of procedures number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 43 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.5. Location of service delivery for screening, diagnosis, and treatment of precancerous lesions 4 and cancer.a 5 b 6 Procedure Location of services 7 HPV DNA test Primary facility 8 Cytology test Primary facility 9 VIA test Primary facility 10 Colposcopy/biopsy Referral facility 11 Cryotherapy Referral facility 12 13 LEEP Referral facility 14 Follow-up visits (after cryotherapy or LEEP) Primary facility 15 Cancer treatmentFor peer reviewTertiary facility only 16 a HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection with 17 acetic acid. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Tabe A.6. Women’s time estimates.a 4 Visit Minutes Cost (2015 US$)b Source 5 c 6 Screening 7 Wait time 33 0.48 Scale-Up 8 Procedure time 20 0.29 START-UP 9 Transport time (round- 57 0.82 Scale-Up 10 trip) 11

12 c 13 Results 14 Wait time 33 0.48 Scale-Up 15 Procedure time For peer2 review0.03 only START-UP 16 Transport time (round- 57 0.82 Scale-Up 17 trip) 18 19 Cryotherapyc 20 21 Wait time 120 1.75 Scale-Up 22 Procedure time 35 0.51 START-UP 23 Transport time (round- 213 3.10 Scale-Up 24 trip) 25 26 Colposcopy/biopsy 27 28 Wait time 120 1.75 Scale-Up 29 Procedure time 37 0.54 START-UP 30 Transport time (round- 213 3.10 Scale-Up 31 trip) 32

33 LEEP http://bmjopen.bmj.com/ 34 Wait time 120 1.75 Scale-Up 35 36 Procedure time 25 0.36 START-UP 37 Transport time (round- 213 3.10 Scale-Up 38 trip) 39 a LEEP: loop electrosurgical excision procedure; Scale-Up: Scale-Up demonstration project (2015); START- 40 UP: Screening Technologies to Advance Rapid Testing for Cervical Cancer Prevention–Utility and 41 Program Planning demonstration project (2010); US$: United States dollars. on September 26, 2021 by guest. Protected copyright. 42 b 43 Women’s time was valued using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 44 US$. We converted this to an hourly wage rate assuming 171 hours of work per month (40 hours per 45 week). 46 c Wait and transport time costs were accrued for each visit to a health facility, but we did not double- 47 count wait and transport times when multiple procedures were performed in the same visit (e.g., VIA 48 triage of screen-positive women and cryotherapy of VIA+ women accrued the wait time and transport 49 time costs associated with a single visit to a referral facility, but accrued the time costs of both 50 51 procedures.) 52 53 54 55 56 57 58 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.7. Women’s transportation costs.a 4 b 5 Facility Round-trip transportation cost Source 6 (2015 US$) 7 Screening facility 0.41 Scale-Up 8 Referral facility 2.81 Scale-Up 9 a Scale-Up: Scale-Up demonstration project (2015); US$: United States dollar. 10 b Screening with cytology or HPV testing was assumed to take place at a screening facility (i.e., a primary 11 12 level health facility). Triage testing with visual inspection with acetic acid (VIA), cryotherapy, 13 colposcopy/biopsy, and loop electrosurgical excision procedure (LEEP) were assumed to take place at 14 referral facilities. 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.8. Programmatic training costs for HPV-based screening.a 4 Type of training Total cost of Total cost of personnel Per woman costd 5 b c 6 supplies/materials (2015 time (2015 US$) (2015 US$) 7 US$) 8 Health care personnel 2,758 5,784 0.09 9 offering careHPV 10 Laboratory technicians 3,503 526 0.04 11 processing careHPV 12 13 assays 14 Community outreach 2,662 3,618 0.08 15 workers promotingFor and peer review only 16 offering self-collection 17 Health care providers 8,103 16,069 1.51 18 offering VIA and 19 20 cryotherapy a 21 HPV: human papillomavirus; US$: United States dollars; VIA: visual inspection with acetic acid. 22 b The total cost of supplies, materials, transportation, and refreshments for all training sessions of each 23 type was converted from 2015 local currency units using the official exchange rate. 24 c The total cost of personnel time includes time costs for both the trainer and the trainees, valued using 25 average monthly salary rates of each type of personnel present (and the proportion of trainees of each 26 personnel type) and based on the length of the training session. 27 d 28 To derive an average health care personnel training cost per woman screened, we divided the total 29 training cost for health care personnel by 100,000, corresponding to the number of women screened 30 during the Scale-Up project. To derive an average laboratory technician training cost per woman 31 screened, we divided the total training cost for laboratory technicians by 100,000. To derive an average 32 community outreach worker training cost per woman screened with self-collection, we divided the total

33 training cost for outreach workers by 80,000, corresponding to the number of women who participated http://bmjopen.bmj.com/ 34 35 in self-collection of HPV samples (due to lack of data on self-collection in a community setting, in the 36 base case analysis we assumed self-collection took place at the clinic; however, we assumed outreach 37 workers would be involved in mobilizing and educating women about self-collection, and conservatively 38 included the cost of training so as to represent a known programmatic cost as self-collection efforts shift 39 to also include community settings). To derive an average health care provider training cost per woman 40 receiving VIA or cryotherapy, we divided the total VIA/cryotherapy training cost for health care 41 providers by 16,000, corresponding to the approximate number of women who tested HPV-positive in on September 26, 2021 by guest. Protected copyright. 42 43 Scale-Up and would thus require visual assessment and cryotherapy. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.9. Cancer treatment costs (2015 US$). 4 Cancer stage Local cancer Regional cancer Distant cancer 5 a 6 Direct medical costs 943.62 917.83 917.83 7 Woman and support 345.77 639.14 639.14 8 person time 9 Transportation, temporary 196.78 389.50 389.50 10 housing, and meals 11 Total cancer treatment 1486 1946 1946 12 13 cost a 14 Direct medical costs are based on the premise that 50% of local cancer patients presented with FIGO 15 stage 1a, requiringFor radical hysterectomy, peer and 50%review of local cancer patients only presented with FIGO stage 16 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. We 17 assumed all women presenting with FIGO stage 2b or higher received cobalt therapy followed by 5 18 chemotherapy sessions followed by brachytherapy. 19 b 20 Patient and support person time costs value time using the minimum wage in Nicaragua, as described 21 in the text. 22 c Direct non-medical costs include round-trip transportation for visits to a cancer center and meals and 23 temporary housing during the course of treatment. 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 Prevalence of High-Risk HPV, Nicaragua 7 8 0.50 9 10 0.45 11 12 0.40 13 14 15 0.35 For peer review only 16 17 0.30 18 19 0.25 20 21 0.20 22 23 Prevalence Prevalence ofHR HPV 24 0.15 25 26 0.10 27 28 0.05 29 30 0.00 31 30-34 35-39 40-44 45-49 32 Age group

33 http://bmjopen.bmj.com/ 34 35 36 Figure A1. Selected model output from the top 50 input parameter sets compared with empirical data 37 (i.e., calibration targets) on age-specific prevalence of high-risk HPV in Nicaragua, based on a relative 38 light unit cut-off value of 0.5 in the START-UP studies [2, 15]. Bold lines represent the 95% confidence 39 40 intervals around the empirical data, and gray circles represent model output from each of the top 50 41 input parameter sets. on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 Cancer Incidence, Nicaragua 7 300 8 9 10 11 250 12 13 14 15 200 For peer review only 16 17 18 19 150 20 21 22 23 100 24 25 Incidence Women100,000 per 26 50 27 28 29 30 0 31 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ 32 Age group

33 http://bmjopen.bmj.com/ 34 35 36 Figure A2. Selected model output from the top 50 input parameter sets compared with empirical data 37 (i.e., calibration targets) on age-specific cancer incidence in Nicaragua [2, 11]. Bold lines represent the 38 95% confidence intervals around the empirical data, and gray circles represent model output from each 39 40 of the top 50 input parameter sets. 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 50% Screening coverage: 50% Screening coverage: 70% Screening coverage: 80% 5 6 45% 7 8 40% 9 10 11 35% 12 13 30% 14 15 For peer review only 16 25% 17 18 20% 19 20 21 15% 22 23 10% 24 Reduction in lifetime risk of cervical cancer, % 25 5% 26 27 28 0% 29 Pap (3 yrs) HPV-ST (5 yrs) HPV-VIA (5 yrs) HPV-Pap (5 yrs) 30 Screening strategy 31 32

33 http://bmjopen.bmj.com/ Figure A3. Reduction in lifetime risk of cervical cancer, by screening coverage level. Bars indicate the 34 35 percent reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 36 years; careHPV screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual inspection 37 with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap 38 39 triage of HPV-positive women [HPV-Pap]) as screening coverage of the target population increases. 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 Figure A4. Cost-effectiveness analysis: Payer cost only. The graph displays the discounted lifetime costs 35 36 (x-axis; in 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing 37 every 3 years; careHPV screen-and-treat every 5 years [HPV-ST]; careHPV every 5 years with visual 38 inspection with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years 39 40 with Pap triage of HPV-positive women [HPV-Pap]), when only costs to the payer are included (i.e., 41

women’s time and transportation costs are excluded). The cost-effectiveness associated with a change on September 26, 2021 by guest. Protected copyright. 42 from one strategy to a more costly alternative is represented by the difference in cost divided by the 43 44 difference in life expectancy associated with the two strategies. The curve indicates the strategies that 45 are efficient because they are more effective and either 1) cost less; or 2) have a more attractive cost- 46 effectiveness ratio than less effective options. The incremental cost-effectiveness ratio (ICER) is the 47 48 reciprocal of the slope of the line connecting the two strategies under comparison. In this sensitivity 49 analysis focused on costs to the payer, HPV-ST every 5 years was less costly and more effective than 50 other screening strategies considered, and was thus a dominant strategy with an ICER of US$270 per 51 52 year of life saved. 53 54 55 56 57 58 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 Figure A5. Cost-effectiveness analysis: Direct medical cost of Pap testing, US$3. The graph displays the 35 36 discounted lifetime costs (x-axis; in 2015 US$) and life expectancy (y-axis) associated with each 37 screening strategy (Pap testing every 3 years; careHPV screen-and-treat every 5 years [HPV-ST]; careHPV 38 every 5 years with visual inspection with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and 39 40 careHPV every 5 years with Pap triage of HPV-positive women [HPV-Pap]), when the direct medical cost 41

of Pap testing was assumed to be as low as US$3 (base case: US$7.26). The cost-effectiveness associated on September 26, 2021 by guest. Protected copyright. 42 with a change from one strategy to a more costly alternative is represented by the difference in cost 43 44 divided by the difference in life expectancy associated with the two strategies. The curve indicates the 45 strategies that are efficient because they are more effective and either 1) cost less; or 2) have a more 46 attractive cost-effectiveness ratio than less effective options. The incremental cost-effectiveness ratio 47 48 (ICER) is the reciprocal of the slope of the line connecting the two strategies under comparison. In this 49 sensitivity analysis examining the impact of a lower Pap test cost, HPV-ST every 5 years was more 50 effective and had a more attractive cost-effectiveness ratio than Pap testing every 3 years; furthermore, 51 52 HPV-ST was more effective and less costly than HPV-VIA and HPV-Pap, and was thus a dominant strategy 53 with an ICER of US$320 per year of life saved. 54 55 56 57 58 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 References 7 8 1. Campos, N.G., et al., An updated natural history model of cervical cancer: derivation of model 9 parameters. Am J Epidemiol, 2014. 180(5): p. 545-55. 10 2. Campos, N.G., Tsu, V., Jeronimo, J., Mvundura, M., Lee, K., Kim, J.J., When and how often to 11 screen for cervical cancer in three low- and middle-income countries: A cost-effectiveness 12 analysis. Papillomavirus Research, 2015. 13 3. Munoz, N., et al., Incidence, duration, and determinants of cervical human papillomavirus 14 15 infection inFor a cohort ofpeer Colombian women review with normal cytological only results. J Infect Dis, 2004. 16 190(12): p. 2077-87. 17 4. McCredie, M.R., et al., Natural history of cervical neoplasia and risk of invasive cancer in women 18 with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol, 2008. 9(5): 19 p. 425-34. 20 5. Meyskens, F.L., Jr., et al., Enhancement of regression of cervical intraepithelial neoplasia II 21 22 (moderate dysplasia) with topically applied all-trans-retinoic acid: a randomized trial. J Natl 23 Cancer Inst, 1994. 86(7): p. 539-43. 24 6. Keefe, K.A., et al., A randomized, double blind, Phase III trial using oral beta-carotene 25 supplementation for women with high-grade cervical intraepithelial neoplasia. Cancer Epidemiol 26 Biomarkers Prev, 2001. 10(10): p. 1029-35. 27 7. Castle, P.E., et al., Evidence for frequent regression of cervical intraepithelial neoplasia-grade 2. 28 Obstet Gynecol, 2009. 113(1): p. 18-25. 29 30 8. Wang, S.M., et al., Six-year regression and progression of cervical lesions of different human 31 papillomavirus viral loads in varied histological diagnoses. Int J Gynecol Cancer, 2013. 23(4): p. 32 716-23.

33 9. Moscicki, A.B., et al., Rate of and risks for regression of cervical intraepithelial neoplasia 2 in http://bmjopen.bmj.com/ 34 adolescents and young women. Obstet Gynecol, 2010. 116(6): p. 1373-80. 35 10. Herrero, R., et al., Rationale and design of a community-based double-blind randomized clinical 36 trial of an HPV 16 and 18 vaccine in Guanacaste, Costa Rica. Vaccine, 2008. 26(37): p. 4795-808. 37 38 11. Ferlay, J., et al., GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC 39 CancerBase No. 11 [Internet], 2013, International Agency for Research on Cancer: Lyon, France. 40 12. World Development Indicators, 2016, World Bank. 41

13. Database, I., Country Profiles: Nicaragua, 2016. on September 26, 2021 by guest. Protected copyright. 42 14. Campos, N.G., et al., The comparative and cost-effectiveness of HPV-based cervical cancer 43 screening algorithms in El Salvador. Int J Cancer, 2015. 44 45 15. Jeronimo, J., et al., A multicountry evaluation of careHPV testing, visual inspection with acetic 46 acid, and papanicolaou testing for the detection of cervical cancer. Int J Gynecol Cancer, 2014. 47 24(3): p. 576-85. 48 16. Energy, U.S.D.o. Average Fuel Economy of Major Vehicle Categories. August 1, 2016]; Available 49 from: http://www.afdc.energy.gov/data/10310. 50 51 52 53 54 55 56 57 58 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only review peer For http://bmjopen.bmj.com/ BMJ Open on September 26, 2021 by guest. Protected copyright. Page 54 of 56 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Page 55 of 56

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only review peer For http://bmjopen.bmj.com/ BMJ Open on September 26, 2021 by guest. Protected copyright. BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1

The Cost-Effectiveness of HPV-Based Cervical Cancer Screening in the Public Health System in Nicaragua

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2016-015048.R1

Article Type: Research

Date Submitted by the Author: 08-Feb-2017

Primary Subject Health economics Heading:

Global health, Health economics, Health policy, Obstetrics and Secondary Subject Heading: gynaecology, Public health

Keywords: HEALTH ECONOMICS, Epidemiology < ONCOLOGY, PUBLIC HEALTH http://bmjopen.bmj.com/

on September 26, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 The Cost-Effectiveness of HPV-Based Cervical Cancer Screening in the Public Health System in 4 5 2 Nicaragua 6 3 Nicole G. Campos, PhD, a Mercy Mvundura, PhD,b Jose Jeronimo, MD,c Francesca Holme, MPH,c Elisabeth 7 8 d a 9 4 Vodicka, MHA, Jane J. Kim, PhD 10 11 5 12 13 6 a Center for Health Decision Science, Harvard T.H. Chan School of Public Health, 718 Huntington Avenue, 14 15 For peer review only 16 7 Boston, Massachusetts, USA 17 18 8 b PATH, Devices and Tools Program, P.O. Box 900922, Seattle, Washington, USA 19 20 9 c PATH, Reproductive Health Global Program, P.O. Box 900922, Seattle, Washington, USA 21

33 14 Word count (text): 4,527 http://bmjopen.bmj.com/ 34 35 36 15 37 38 16 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 17 Abstract 4 5 18 Objectives: To evaluate the cost-effectiveness of HPV DNA testing (versus Pap-based screening) for 6 7 8 19 cervical cancer screening in Nicaragua. 9 10 20 Design: A previously developed Monte Carlo simulation model of the natural history of HPV infection 11 12 21 and cervical cancer was calibrated to epidemiologic data from Nicaragua. Cost data inputs were derived 13 14 22 using a micro-costing approach in Carazo, Chontales, and Chinandega departments; test performance 15 For peer review only 16 17 23 data were from a demonstration project in Masaya department. 18 19 24 Setting: Nicaragua’s public health sector facilities. 20 21 25 Participants: Women aged 30 to 59 years. 22 23 24 26 Interventions: Screening strategies included 1) Pap testing every 3 years, with referral to colposcopy for 25 26 27 women with an ASCUS+ result (“Pap”); 2) HPV testing every 5 years, with referral to cryotherapy for 27 28 28 HPV-positive eligible women (HPV cryotherapy, or “HPV-Cryo”); 3) HPV testing every 5 years, with 29 30 31 29 referral to triage with visual inspection with acetic acid (VIA) for HPV-positive women (“HPV-VIA”); and 32

33 30 4) HPV testing every 5 years, with referral to Pap testing for HPV-positive women (“HPV-Pap”). http://bmjopen.bmj.com/ 34 35 31 Outcome measures: Reduction in lifetime risk of cancer; incremental cost-effectiveness ratios (ICER; 36 37 32 2015 US$ per year of life saved [YLS]). 38 39 40 33 Results: HPV-based screening strategies were more effective than Pap testing. HPV-Cryo was the least 41 on September 26, 2021 by guest. Protected copyright. 42 34 costly and most effective strategy, reducing lifetime cancer risk by 29.5% and outperforming HPV-VIA, 43 44 35 HPV-Pap, and Pap only, which reduced cancer risk by 19.4%, 12.2%, and 10.8%, respectively. With an 45 46 47 36 ICER of US$320/YLS, HPV-Cryo every 5 years would be very cost-effective using a threshold based on 48 49 37 Nicaragua’s per capita GDP of US$2,090. Findings were robust across sensitivity analyses on test 50 51 38 performance, coverage, compliance, and cost parameters. 52 53 54 55 56 57 58 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 39 Conclusions: HPV testing is very cost-effective compared to Pap testing in Nicaragua, due to higher test 4 5 6 40 sensitivity and the relatively lower number of visits required. Increasing compliance with recommended 7 8 41 follow-up will further improve the health benefits and value for public health dollars. 9 10 42 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 43 Strengths and limitations of this study: 4 5 6 44 • Using implementation data from the Scale-Up project— which aims to facilitate 7 8 45 institutionalization of HPV testing at the national level in Guatemala, Honduras, and Nicaragua— 9 10 46 we estimated the cost-effectiveness of HPV testing in Nicaragua’s public health system. 11 12 13 47 • The screening algorithms, as modeled, reflect the prototypical structure of a screening episode 14 15 48 and the typeFor of facility peer at which visits usuallyreview take place, but doonly not capture variation due to 16 17 49 geography or health facility capacity. 18 19 20 50 • While we adopted a micro-costing approach to leverage data from implementation in 21 22 51 Nicaragua, individual-level data for each woman were not available; furthermore, we did not 23 24 52 have cost data associated with HPV self-collection in community settings, where most self- 25 26 27 53 collection takes place. 28 29 54 • Despite these limitations, extensive sensitivity analyses on cost and screening assumptions 30 31 55 indicate that HPV-Cryo is robustly the most efficient strategy. 32

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 65 received all tests used in the study as a donation from Qiagen; no other relationships or activities that 4 5 6 66 could appear to have influenced the submitted work. 7 8 67 Contributors: All authors developed the analysis plan and interpreted the data. NC, MM, JJ, FH, and JK 9 10 68 conceptualized the study. MM, JJ, FH, and EV collected implementation data. NC conducted data 11 12 13 69 analysis and wrote the first draft of the report and revised subsequent drafts. JJ was the principal 14 15 70 investigator of theFor START-UP andpeer Scale-Up projects. review JK was the principal only investigator overseeing 16 17 71 microsimulation model development. All authors contributed to and approved the final report. 18 19 72 Data sharing statement: PATH provides technical assistance to the government of Nicaragua and has 20 21 22 73 access to screening indicators through partnership with Movicáncer, a local non-governmental 23 24 74 organization partner in Nicaragua. Movicáncer designed the health information system used by the 25 26 75 Ministry of Health to track screening and treatment of women’s cancers. The Ministry of Health protects 27 28 29 76 individually identifiable information; PATH received de-identified and consolidated data on visit 30 31 77 compliance from Movicáncer. Costing data were collected by Movicáncer through consultation with 32

33 78 Ministry of Health personnel, mostly via phone interviews. No individual identifiers were collected. http://bmjopen.bmj.com/ 34 35 36 79 These data are not available from any public source. A supplementary appendix describing the costing 37 38 80 data and microsimulation model is available. 39 40 81 Acknowledgements: We gratefully acknowledge the efforts of Movicáncer of Nicaragua in gathering 41 on September 26, 2021 by guest. Protected copyright. 42 82 costing data. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 83 Background 4 5 6 84 Cervical cancer is the leading cause of cancer among women in Nicaragua, with an estimated 7 8 85 934 cases and 424 deaths each year [1]. Yet cervical cancer is preventable through screening that allows 9 10 86 for early detection and subsequent treatment of precancerous lesions caused by sexually transmitted 11 12 13 87 infection with human papillomavirus (HPV). While most HPV infections clear spontaneously within 1 to 2 14 15 88 years, a persistentFor infection withpeer one of approximately review 15 oncogenic onlyHPV genotypes may progress to 16 17 89 precancer, which, if untreated, may become invasive cancer [2, 3]. 18 19 90 In most high-income countries, routine screening with cervical cytology (i.e., Pap) testing has 20 21 22 91 substantially reduced the incidence of cervical cancer [4]. However, due to its low sensitivity to detect 23 24 92 precancer, Pap testing must be performed at regular frequent intervals in women of screening age. In 25 26 93 low- and middle-income settings, where many women do not have access to routine primary health care 27 28 29 94 and even fewer have access to higher-level facilities that offer diagnostic testing and treatment, Pap 30 31 95 testing has not been effective at reducing cervical cancer incidence and mortality [5]. In Nicaragua, an 32

33 96 estimated 31.5% of women aged 15 to 49 years have been screened within the last year, and nearly 30% http://bmjopen.bmj.com/ 34 35 36 97 of women in this age group have never been screened [6]. One recent survey found that 87% of women 37 38 98 in León, Nicaragua, were informed of their Pap results, but of those who were referred to follow-up, 39 40 99 only 67% received further care [7]. 41 on September 26, 2021 by guest. Protected copyright. 42 100 HPV DNA tests are highly sensitive to detect potentially oncogenic HPV infections and present 43 44 45 101 an alternative to Pap-based screening. Because HPV-negative women are at very low risk for developing 46 47 102 cervical cancer within the next 10 years [8], the interval between screenings can be extended to at least 48 49 103 5 years for this subset of women [9]. An additional advantage of HPV testing is that samples can be 50 51 52 104 collected by a provider or by the woman herself, reducing the burden on health workers and time 53 54 105 women spend seeking care and potentially increasing screening uptake [10-13]. Furthermore, a lower- 55 56 106 cost HPV DNA test known as careHPV has been clinically validated [14, 15] and is now commercialized. 57 58 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 107 Given these potential benefits, the World Health Organization (WHO) recommends HPV testing for 4 5 6 108 countries with sufficient resources [9]. 7 8 109 In 2011, the Screening Technologies to Advance Rapid Testing for Cervical Cancer 9 10 110 Prevention−Ulity and Program Planning (START-UP) project in Nicaragua’s Masaya department 11 12 13 111 demonstrated that screening with careHPV could be effectively implemented in public sector health 14 15 112 facilities [14]. TheFor Ministry of Healthpeer of Nicaragua review subsequently built onlyon these initial efforts, 16 17 113 incorporating HPV testing into public health care systems in three departments with technical assistance 18 19 114 from PATH under the Scale-Up project. Adoption of HPV testing within Nicaragua’s public health care 20 21 22 115 system is taking place in three phases [16]. In Phase 1, partner organizations worked with the Ministry of 23 24 116 Health to prepare for introduction of HPV screening into public health facilities by developing screening 25 26 117 and treatment algorithms; creating educational materials; organizing training sessions for health 27 28 29 118 workers and laboratory technicians; and bolstering referral and treatment systems for follow-up of 30 31 119 screen-positive women. Phase 2 piloted screening with 10,000 HPV tests in order to identify and address 32

33 120 barriers to implementation. Phase 3 will expand coverage to over 50,000 women within one year. http://bmjopen.bmj.com/ 34 35 36 121 To inform decision makers considering the national adoption and scale-up of HPV testing within 37 38 122 Nicaragua’s public health sector, this study aimed to 1) estimate the economic cost of cervical cancer 39 40 123 screening with careHPV testing; and 2) project the long-term health and economic impact and value 41 on September 26, 2021 by guest. Protected copyright. 42 124 (i.e., cost-effectiveness) of careHPV testing in Nicaragua relative to existing Pap-based screening. 43 44 45 125 46 47 126 Methods 48 49 127 Analytic overview 50 51 128 We used a micro-costing approach to measure and aggregate the cost of all resources used to 52 53 129 provide cervical cancer screening at the level of the individual patient within the public health sector in 54 55 130 Nicaragua. We considered direct medical costs (i.e., medical resources required for the intervention), 56 57 131 direct non-medical costs (i.e., other resources consumed as part of the intervention, such as patient 58 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 132 transportation costs), and patient time costs (i.e., time spent traveling and waiting for or receiving care). 4 5 6 133 These cost data were input into a previously developed Monte Carlo simulation model (programmed in 7 8 134 C++) of the natural history of HPV infection and cervical cancer that was calibrated to epidemiologic data 9 10 135 from Nicaragua [17, 18]. We then used the model to project the lifetime health and economic outcomes 11 12 13 136 associated with careHPV testing, using three different algorithms for the management of women who 14 15 137 test HPV-positive,For and Pap-based peer screening for review women aged 30 to 59 only years. 16 17 138 Model outcomes included the lifetime risk of cervical cancer, total lifetime costs per woman (in 18 19 139 2015 United States dollars [US$]), and life expectancy. Incremental cost-effectiveness ratios (ICERs) 20 21 22 140 were calculated by dividing the additional cost of a particular strategy by its additional health benefit, 23 24 141 compared with the next most costly strategy. Dominated strategies (defined as more costly and either 25 26 142 less effective or having a higher cost-effectiveness ratio than more effective strategies) were eliminated. 27 28 29 143 There is no universal criterion that defines a threshold cost-effectiveness ratio, below which an 30 31 144 intervention is considered good value for money; we considered an intervention with an ICER less than 32

33 145 Nicaragua’s 2015 per capita GDP of US$2,090 to be “very cost-effective”, and an intervention with an http://bmjopen.bmj.com/ 34 35 36 146 ICER less than three times per capita GDP as “cost-effective” [19]. We followed guidelines for cost- 37 38 147 effectiveness by adopting a societal perspective, including costs irrespective of the payer in order to 39 40 148 capture the opportunity cost of resources used for the screening intervention. We discounted future 41 on September 26, 2021 by guest. Protected copyright. 42 149 costs and life-years at a rate of 3% per year to account for time preferences [20, 21]. 43 44 45 150 46 47 151 Mathematical simulation model 48 49 152 Descriptions of the natural history model of HPV infection and cervical carcinogenesis and 50 51 153 model parameterization process have been previously published [17, 18], but we summarize model 52 53 54 154 features here. Individual girls enter the model at age 9 years, prior to initiating sexual activity, and face 55 56 155 monthly transitions between mutually exclusive health states that reflect disease progression, including 57 58 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 156 type-specific HPV infection, grade of precancer (i.e., cervical intraepithelial neoplasia [CIN] grade 2 or 3), 4 5 6 157 and stage of invasive cancer. Transition probabilities may vary by age, HPV type, duration of infection or 7 8 158 precancerous lesion status, prior HPV infection, and exposure to screening and treatment of HPV or 9 10 159 precancer. Cervical cancer can be detected through symptoms or screening. Death can occur from non- 11 12 13 160 cervical causes or from cervical cancer after its onset. The model tracks each individual woman’s health 14 15 161 status, clinical events,For and economic peer outcomes review over her lifetime, and only aggregates outcomes to estimate 16 17 162 the expected costs and health outcomes over the lifetime of the cohort. 18 19 163 The model was calibrated to epidemiologic data on age-specific HPV prevalence and cervical 20 21 22 164 cancer incidence from Nicaragua [1, 14, 17]. We estimated baseline “prior” input parameter values for 23 24 165 natural history transitions using available longitudinal data, including age- and type-specific HPV 25 26 166 incidence data from Colombia [22-25]. To reflect potential differences in parameters that may vary by 27 28 29 167 setting (i.e., age- and type-specific HPV incidence, natural immunity following initial infection) and 30 31 168 uncertainty in progression and regression of precancer, we set plausible bounds around these input 32

33 169 values and performed repeated model simulations of disease natural history in the absence of any http://bmjopen.bmj.com/ 34 35 36 170 intervention. Each model simulation selected one random value within the bounds for each uncertain 37 38 171 parameter, creating a unique natural history input parameter set. By summing the log-likelihood of 39 40 172 model-projected outcomes for each parameter set relative to the epidemiologic data from Nicaragua, 41 on September 26, 2021 by guest. Protected copyright. 42 173 we computed a goodness-of-fit score. We selected the 50 top-fitting input parameter sets to use in 43 44 45 174 analysis. Results are reported as the mean across the top 50 parameter sets, and ICERs are reported as 46 47 175 the ratio of the mean costs divided by the mean effects of one strategy versus another across sets [26]. 48 49 176 Further details on model parameterization, including calibration, are available in the Appendix. 50 51 52 177 53 54 178 Cervical cancer screening strategies 55 56 57 58 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 179 We considered the following screening strategies for women aged 30 to 59 years (Figure 1): 1) 4 5 6 180 Pap testing every 3 years, with referral to colposcopy for all women with an atypical squamous cells of 7 8 181 undetermined significance or worse (ASCUS+) result (hereafter referred to as “Pap”); 2) careHPV testing 9 10 182 every 5 years, with referral to cryotherapy for all HPV-positive eligible women (HPV cryotherapy, or 11 12 13 183 “HPV-Cryo”); 3) careHPV testing every 5 years, with referral to triage with visual inspection with acetic 14 15 184 acid (VIA) for all HPV-positiveFor womenpeer (“HPV-VIA”); review and 4) careHPV testingonly every 5 years, with referral to 16 17 185 Pap testing for all HPV-positive women (“HPV-Pap”). The pathway of care for each strategy was based 18 19 186 on patterns of care in the Scale-Up project, national screening guidelines, and WHO recommendations. 20 21 22 187 We optimistically assumed 70% of women had access to routine screening and attended an initial visit at 23 24 188 a screening facility (i.e., a primary health care facility). Women could then return to receive screening 25 26 189 results and recommendations for any necessary follow-up care. Follow-up could include colposcopy, 27 28 29 190 cryotherapy, or triage testing at a referral (i.e., higher level) facility, with the exception of Pap triage 30 31 191 testing, which was assumed to take place at the screening clinic. At each encounter after the initial 32

33 192 screening visit, we assumed 85% of women complied with each subsequent visit to a screening facility, http://bmjopen.bmj.com/ 34 35 36 193 while 40% complied with each subsequent visit to a referral facility, consistent with data from Phase 2 of 37 38 194 the Scale-Up project. The minimum number of visits required for treatment in a single screening episode 39 40 195 was 4 for Pap, 3 for HPV-Cryo, 3 for HPV-VIA, and 5 for HPV-Pap. In the HPV-VIA and HPV-Pap strategies, 41 on September 26, 2021 by guest. Protected copyright. 42 196 women who were HPV-positive but negative on the selected triage test were referred to repeat HPV 43 44 45 197 testing in 1 year. In the HPV-Cryo and HPV-VIA strategies, women who were not eligible for treatment 46 47 198 with cryotherapy based on visual assessment were referred to colposcopy with biopsy to rule out 48 49 199 cancer; in the absence of cancer, these women were referred to treatment with loop electrosurgical 50 51 52 200 excision procedure (LEEP) or cryotherapy. 53 54 201 Screening and treatment parameters are presented in Table 1 [6, 14, 17, 27-32]. Screening test 55 56 202 performance data were drawn from the START-UP project in Nicaragua to reflect local test 57 58 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 203 characteristics. While the START-UP project did not evaluate VIA and Pap as triage tests, we used the 4 5 6 204 VIA and Pap positivity rates in HPV-positive women, along with published studies of triage test 7 8 205 performance, to inform triage test sensitivity and specificity [33-38]; in the base case, we optimistically 9 10 206 assumed high sensitivity of triage testing. 11 12 13 207 For all HPV testing strategies, we assumed 20% of women received provider-collection of 14 15 208 cervical specimensFor and 80% ofpeer women self-collected review vaginal specimens, only consistent with the proportions 16 17 209 in the Scale-Up project to date. We weighted cost and health outcomes for provider- and self-collection 18 19 210 accordingly when aggregating results for the HPV strategies. 20 21 22 211 23 24 212 Cost data 25 26 213 All costs were converted to 2015 United States dollars (US$) using GDP deflators and the official 27 28 214 exchange rate [39]. The direct medical costs of screening, diagnosis, and treatment of precancer were 29 30 31 215 drawn from the START-UP study (Masaya department) and the Scale-Up project (Carazo, Chontales, and 32

33 216 Chinandega departments). Direct medical costs included clinical staff time, clinical supplies, drugs, http://bmjopen.bmj.com/ 34 35 217 clinical equipment, laboratory staff time, laboratory supplies, and laboratory equipment. Direct non- 36 37 218 medical costs included women’s round-trip transportation costs to health facilities, and were based on 38 39 40 219 estimates provided by Scale-Up project staff to represent average transportation costs in the Carazo, 41 on September 26, 2021 by guest. Protected copyright. 42 220 Chontales, and Chinandega departments. To account for the opportunity cost of women’s time spent 43 44 221 traveling to, waiting for, or receiving care, we used time estimates from the START-UP and Scale-Up 45 46 47 222 projects, and valued women’s time using Nicaragua’s monthly minimum wage to serve as a proxy for the 48 49 223 societal value of women’s time. Figure 2 displays the categorical breakdown of undiscounted costs for 50 51 224 the screening visits over the course of a woman’s screening-eligible years, with Pap versus careHPV 52 53 54 225 testing. 55 56 57 58 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 226 Data on programmatic costs are limited, but for HPV strategies we included the cost of training 4 5 6 227 sessions for health providers offering HPV-based screening, outreach workers, laboratory technicians, 7 8 228 and providers offering VIA and cryotherapy. While women who self-collected HPV specimens in the 9 10 229 Scale-Up project primarily did so in a community setting, micro-costing data were not available for self- 11 12 13 230 collection performed outside of the clinic, so we assumed clinic-based self-collection. However, we 14 15 231 conservatively includedFor the cost peer of training outreach review workers to represent only a known programmatic cost 16 17 232 as self-collection efforts are shifting to community settings. 18 19 233 Data on the costs of treating cervical cancer were unavailable for Nicaragua, so we estimated 20 21 22 234 direct medical, direct non-medical, and patient and support person time costs using data from El 23 24 235 Salvador [29]. 25 26 236 Selected cost data are presented in Table 1. Further details on cost data are provided in the 27 28 29 237 Appendix. 30 31 238 32

33 239 Sensitivity analyses http://bmjopen.bmj.com/ 34 35 240 We performed sensitivity analysis to examine the impact of independently varying uncertain 36 37 241 parameters, including Pap test performance, triage test performance in HPV-positive women, 38 39 40 242 colposcopy performance, screening coverage, visit compliance, eligibility for cryotherapy following a 41 on September 26, 2021 by guest. Protected copyright. 42 243 positive screening and triage test, treatment effectiveness, discount rate, and cost data. Ranges selected 43 44 244 for sensitivity analysis are displayed in Table 1. 45 46 47 245 48 49 246 Scenario analysis 50 51 247 The base case and sensitivity analysis assumed the availability of all strategies (i.e., Pap; HPV- 52 53 54 248 Cryo; HPV-VIA; and HPV-Pap). Additionally, we performed a scenario analysis in which we assumed HPV- 55 56 57 58 59 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 249 Cryo was not available for logistical and programmatic reasons (i.e., limited access to cryotherapy 4 5 6 250 equipment and gas). 7 8 251 9 10 252 Results 11 12 13 253 Base case: Population-level health benefits and cost-effectiveness analysis 14 15 254 HPV-basedFor screening strategiespeer were morereview effective than Pap only testing. Among the HPV 16 17 255 strategies, HPV-Cryo (every 5 years) was the most effective strategy; under base case assumptions, it 18 19 256 reduced the lifetime risk of cervical cancer by 29.5% on average (range: 25.2%−33.6%). HPV-VIA (every 5 20 21 22 257 years) reduced cancer risk by 19.4% (range: 16.2%−22.6%), while HPV-Pap reduced cancer risk by 12.2% 23 24 258 (range: 10.2%−14.5%). Pap (every 3 years) reduced cancer risk by 10.8% (range: 8.7%−13.4%). Under 25 26 259 base case assumptions, HPV-Cryo was both less costly and more effective than all other strategies, thus 27 28 29 260 dominating HPV-VIA, HPV-Pap, and Pap alone. With an ICER of US$320 per year of life saved (YLS), HPV- 30 31 261 Cryo every 5 years would be considered “very cost-effective” given Nicaragua’s per capita GDP of 32

33 262 US$2,090. The total discounted lifetime cost per woman and life expectancy associated with each http://bmjopen.bmj.com/ 34 35 36 263 screening strategy is presented in Figure 3. 37 38 264 39 40 265 Sensitivity analysis 41 on September 26, 2021 by guest. Protected copyright. 42 266 While HPV-Cryo remained the most effective strategy across all sensitivity analyses, the 43 44 45 267 magnitude of reduction in lifetime risk of cancer was dependent upon screening coverage of the target 46 47 268 population and compliance with recommended follow-up. When coverage was 50% and all other 48 49 269 parameters were held constant at base case values, HPV-Cryo reduced cancer risk by an average of 50 51 52 270 21.1%; HPV-VIA, HPV-Pap, and Pap yielded average cancer risk reductions of 13.9%, 8.7%, and 7.7%, 53 54 271 respectively. As coverage increased to 80%, all else being equal, HPV-Cryo reduced cancer risk by an 55 56 272 average of 33.5%, while HPV-VIA, HPV-Pap, and Pap yielded average cancer risk reductions of 22.1%, 57 58 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 273 14.0%, and 12.4%, respectively (Appendix). Figure 4 displays the impact of visit compliance on lifetime 4 5 6 274 risk of cancer. When compliance with visits to all facilities (i.e., for both screening and referral) was low 7 8 275 at 40%, HPV-Cryo remained the most effective strategy, but only reduced cancer risk by 16.2%; Pap had 9 10 276 little health impact at this level of compliance, reducing cancer risk by only 5.4%. As compliance at all 11 12 13 277 facilities rose to 85%, HPV-Cryo reduced cancer risk by 47.9%; HPV-VIA, HPV-Pap, and Pap reduced 14 15 278 cancer risk by 42.0%,For 40.7%, andpeer 35.7% respectively. review only 16 17 279 In addition to remaining the most effective strategy across all sensitivity analyses, HPV-Cryo 18 19 280 remained the most efficient strategy as well. HPV-Cryo remained the least costly and most effective 20 21 22 281 strategy with a stable ICER of US$320 per YLS when 1) Pap test performance (as a primary screening 23 24 282 test) improved; 2) VIA and Pap triage test performance improved; 3) colposcopy was assumed to be 25 26 283 perfect; and 4) the direct medical cost of LEEP was varied from 75% to 125% of the base case. Despite 27 28 29 284 slight fluctuation in the ICER, HPV-Cryo also remained the least costly and most effective strategy as 1) 30 31 285 screening coverage varied from 50% to 80%; 2) visit compliance varied from 40% to 85% per visit; 3) the 32

33 286 screen-and-treat cryotherapy cure rate was reduced to 75%; 4) eligibility for cryotherapy was reduced to http://bmjopen.bmj.com/ 34 35 36 287 75% or less; 5) the treatment cure rate following colposcopy was reduced to 85%; 6) women receiving 37 38 288 treatment following colposcopy in the Pap and HPV-Pap strategies were assumed to receive cryotherapy 39 40 289 instead of the more costly LEEP; 7) the discount rate was varied from 0% to 5%; 8) only payer costs were 41 on September 26, 2021 by guest. Protected copyright. 42 290 considered, excluding women’s time and transportation costs; 9) the direct medical cost of HPV self- 43 44 45 291 collection was varied from 75% to 125% of the base case; 10) the direct medical cost of cryotherapy was 46 47 292 increased to 170% of the base case; 11) the direct medical cost of colposcopy was reduced to 35% of the 48 49 293 base case; 12) programmatic costs associated with HPV-based screening were varied from 50% to 150% 50 51 52 294 of the base case; 13) women’s time and transportation costs were reduced to 50% of the base case; and 53 54 295 14) the costs of cancer treatment ranged from including only direct medical costs to 150% of the base 55 56 296 case (Figure 5). Among variables considered, compliance per visit appears to have the greatest impact 57 58 59 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 297 on the ICER for HPV-Cryo, with 40% compliance yielding an ICER of US$580 per YLS and 85% compliance 4 5 6 298 yielding an ICER of US$190 per YLS. Even when visit compliance is low, HPV-Cryo would be considered 7 8 299 very cost-effective. 9 10 300 When only payer costs were considered (i.e., women’s time and transportation costs were 11 12 13 301 excluded), the total lifetime cost per woman was lower for all strategies, and Pap every 3 years was 14 15 302 slightly less costlyFor (although stillpeer slightly less effective)review than HPV-Pap onlyevery 5 years. HPV-Cryo remained 16 17 303 the most effective and efficient strategy, with an ICER of US$270 per YLS (Appendix). 18 19 304 The only scenario in which Pap testing every 3 years was the least costly strategy occurred when 20 21 22 305 the direct medical cost of Pap testing was US$3 (base case: US$7.26), a value commonly cited for the 23 24 306 cost of Pap in Nicaragua, though the source of this estimate is unknown. However, Pap remained the 25 26 307 least effective strategy, and HPV-Cryo had a lower cost-effectiveness ratio, maintaining an ICER of 27 28 29 308 US$320 per YLS (Appendix). 30 31 309 32

33 310 Scenario analysis: HPV-Cryo unavailable http://bmjopen.bmj.com/ 34 35 36 311 When we assumed HPV-Cryo was not available as a screening strategy, HPV-VIA was the least 37 38 312 costly and most effective strategy in the base case and most sensitivity analyses, with a base case ICER 39 40 313 of US$550 per YLS (Appendix). Exceptions included the following circumstances: 1) when the direct 41 on September 26, 2021 by guest. Protected copyright. 42 314 medical cost of Pap was US$3, Pap alone had a more attractive ICER (US$530), albeit was less effective 43 44 45 315 than HPV-VIA (US$630 per YLS); 2) Pap performance in the general screening population was improved 46 47 316 and Pap alone became the most effective strategy, with an ICER of US$540 per YLS; 3) VIA test sensitivity 48 49 317 in HPV-positive women was only 0.40, in which case HPV-VIA had an ICER of US$726 per YLS, but HPV- 50 51 52 318 Pap was more effective with an ICER of US$3,260. 53 54 319 55 56 320 Discussion 57 58 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 321 Using implementation data from the Scale-Up project— which aims to facilitate 4 5 6 322 institutionalization of HPV testing at the national level in Guatemala, Honduras, and Nicaragua— we 7 8 323 estimated the long-term health impact and value of careHPV testing in Nicaragua’s public health system. 9 10 324 We found that screening algorithms consisting of HPV testing at 5-year intervals would be less costly and 11 12 13 325 more effective than screening with Pap testing at 3-year intervals. Furthermore, HPV testing followed by 14 15 326 treatment with cryotherapyFor forpeer all eligible HPV-positive review women would only be less costly and more effective 16 17 327 than HPV testing followed by triage testing with either VIA or Pap for HPV-positive women. A screen- 18 19 328 and-treat HPV program would be a very cost-effective intervention in Nicaragua, with an ICER of US$320 20 21 22 329 per YLS under base case assumptions. These findings were robust across sensitivity analyses. The 23 24 330 comparatively large health benefits and efficiency of HPV-Cryo can largely be attributed to the relatively 25 26 331 low number of visits to health care facilities and the high sensitivity of the careHPV test to detect both 27 28 29 332 CIN2+ and oncogenic HPV infections with the potential to develop into precancer. 30 31 333 We found that screening coverage of the target population had a considerable impact on 32

33 334 achievable reductions in cervical cancer risk, with HPV-Cryo yielding the greatest risk reduction. Due to http://bmjopen.bmj.com/ 34 35 36 335 proportional increases in both costs and health benefits, the ICER for HPV-Cryo remained stable as 37 38 336 coverage increased from 50% to 80%. Compliance with recommended follow-up was a key driver of 39 40 337 both achievable reductions in cancer risk and the ICER of HPV-Cryo. As the proportion of women who 41 on September 26, 2021 by guest. Protected copyright. 42 338 returned for each clinical encounter (relative to the previous visit) increased from 40% to 85%, the 43 44 45 339 effectiveness or cancer benefit associated with HPV-Cryo rose from 16.2% to 47.9% as more women 46 47 340 were linked to treatment; the ICER fell from US$580 per YLS to US$190 per YLS as more cancers were 48 49 341 averted. Thus, improved efforts to successfully navigate women to recommended follow-up will 50 51 52 342 enhance screening program effectiveness and efficiency. 53 54 343 While substantially reducing the cost of Pap testing to US$3 (less than half of the base case 55 56 344 value) made Pap the strategy with the lowest per-woman lifetime costs, the confluence of low test 57 58 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 345 sensitivity and the high number of health facility visits needed to complete screening, diagnostic follow- 4 5 6 346 up, and treatment made Pap the least effective strategy since women are lost to follow-up with each 7 8 347 additional required visit. Even in this low-cost Pap scenario, HPV-Cryo remained the most effective and 9 10 348 cost-effective strategy. 11 12 13 349 When we assumed a scenario in which HPV-Cryo was not available for logistic and programming 14 15 350 reasons, we foundFor that HPV-VIA peer was usually thereview least costly and most only effective of the remaining 16 17 351 strategies. Although the ICER was less attractive than the ICER associated with HPV-Cryo in the main 18 19 352 analysis, it was below Nicaragua’s per capita GDP. 20 21 22 353 There are several limitations to this analysis. We did not consider alternative screening intervals 23 24 354 or ages for each strategy, but rather restricted the analysis to the ages and intervals currently under 25 26 355 consideration by the Nicaraguan Ministry of Health. It is likely that an increased screening interval and 27 28 29 356 fewer lifetime screens will also be cost-effective, although health benefits may be reduced; we 30 31 357 demonstrated in a previous analysis that screening once or three times in a woman’s lifetime with 32

33 358 careHPV would be very cost-effective in Nicaragua [17]. The screening algorithms, as modeled, reflect http://bmjopen.bmj.com/ 34 35 36 359 the prototypical structure of a screening episode and the type of facility at which visits usually take 37 38 360 place, but do not capture variation due to geography or health facility capacity. Furthermore, the 39 40 361 modeled screening algorithms do not entirely reflect the complex downstream follow-up of screen- 41 on September 26, 2021 by guest. Protected copyright. 42 362 positive women that is embodied in the Ministry of Health’s screening guidelines. In simplifying the 43 44 45 363 downstream follow-up for modeled strategies, we may have underestimated the costs and 46 47 364 overestimated the benefits relative to the national guidelines, which call for additional follow-up prior to 48 49 365 treatment. However, the modeled strategies would likely bias the analysis in favor of Pap and HPV triage 50 51 52 366 strategies. We also did not consider a strategy in which women with ASCUS+ were referred directly to 53 54 367 cryotherapy instead of colposcopy because that approach is not recommended by the WHO guidelines, 55 56 57 58 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 368 or by any professional medical society to the best of our knowledge. We did not consider quality of life 4 5 6 369 impact or the potential disutilities or harms that might be associated with overtreatment. 7 8 370 While we adopted a micro-costing approach to leverage data from the START-UP and Scale-Up 9 10 371 projects in Nicaragua, there remain limitations to our cost estimates. First, individual-level data for each 11 12 13 372 woman were not available; thus, our estimates represent average costs in the project populations. 14 15 373 Second, we did notFor have information peer on the costsreview associated with HPV only self-collection in community 16 17 374 settings, where most self-collection takes place. Instead, we assumed all self-collection took place at the 18 19 375 clinic. Compared to clinic-based efforts, community-based self-collection may be associated with lower 20 21 22 376 costs for women’s time and travel, and higher direct medical and programmatic costs due to outreach 23 24 377 worker involvement in facilitating screening and delivering results. Third, our estimates of programmatic 25 26 378 costs were restricted to training sessions, and we did not have information on the costs of social 27 28 29 379 mobilization and outreach, patient navigation and support, or infrastructural improvements that would 30 31 380 be required to successfully scale-up a screening program. Fourth, we valued women’s time based on the 32

33 381 minimum wage in Nicaragua. This may be a conservative estimate if most women attending screening http://bmjopen.bmj.com/ 34 35 36 382 are formally employed; conversely, it may overestimate the societal value of women’s time spent 37 38 383 working in the informal sector or at home. Finally, we extrapolated the cost of cancer treatment using 39 40 384 data from El Salvador [29]. Despite these limitations, extensive sensitivity analyses on cost components 41 on September 26, 2021 by guest. Protected copyright. 42 385 indicate that HPV-Cryo is robustly the most efficient strategy. 43 44 45 386 As implementation of HPV testing continues, particularly without triage testing, the health 46 47 387 system’s capacity to provide cryotherapy will likely need to increase. While the use of triage testing 48 49 388 (either with Pap or HPV) reduces the number of cryotherapy procedures performed, we found that the 50 51 52 389 lower sensitivity of triage testing (resulting in more false negatives) led to a decline in health benefits as 53 54 390 fewer women with persistent HPV infection and precancer received treatment. The cost savings 55 56 391 associated with fewer cryotherapy procedures were outweighed by increased costs of additional follow- 57 58 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 392 up and cancer treatment in triage-negative women. A sensitivity analysis on the cost of cryotherapy 4 5 6 393 revealed that HPV-Cryo remained the dominant strategy even when costs increased to 170% of the base 7 8 394 case. However, we did not explicitly consider the costs of increasing access to cryotherapy machines or 9 10 395 the implications of gas stock-outs, which have been identified as barriers in some low- and middle- 11 12 13 396 income countries [40]. New ablative technologies currently undergoing testing are smaller, portable, 14 15 397 and do not requireFor gas. Thermal-coagulation peer hasreview been used in United only Kingdom for more than 30 years, 16 17 398 and now it is being used in several low- and middle-income countries, including as part of a “screen-and- 18 19 399 treat” program in Malawi [41], and is currently undergoing testing in Latin America. If newer 20 21 22 400 technologies demonstrate cure rates similar to cryotherapy, the cost-effectiveness of screen-and-treat 23 24 401 algorithms may improve along with access to treatment. 25 26 402 In summary, using data from the Scale-Up implementation project in Nicaragua, we found that 27 28 29 403 HPV testing followed by cryotherapy for eligible HPV-positive women (a “screen-and-treat” approach) 30 31 404 was a very cost-effective intervention in Nicaragua. As the HPV-Cryo algorithm was not implemented in 32

33 405 Phase 2 of the Scale-Up project, compliance and cost estimates may need to be further honed to reflect http://bmjopen.bmj.com/ 34 35 36 406 improvements in capacity for cryotherapy if HPV-Cryo is implemented going forward. While it is too 37 38 407 early to assume that costs and health impact from Phase 2 of implementation are generalizable to other 39 40 408 departments in Nicaragua or other settings in Central America, extensive sensitivity analyses indicate 41 on September 26, 2021 by guest. Protected copyright. 42 409 the robustness of findings. An HPV-based screening algorithm involving a similar “screen-and-treat” 43 44 45 410 approach was recently found to be good value for public health dollars in El Salvador [29], where 46 47 411 national scale-up is underway. It is important to note that a favorable cost-effectiveness profile does not 48 49 412 guarantee that HPV-Cryo will be affordable or feasible in a lower-middle income country like Nicaragua. 50 51 52 413 Both the cost-effectiveness ratio, budgetary impact, and health systems infrastructure need to be 53 54 414 favorable for screening programs to be sustainable. We present these findings to inform evidence-based 55 56 57 58 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 415 decision making around national screening guidelines, program design and implementation, and 4 5 6 416 budgeting for infrastructural improvements and procurement of HPV tests in Nicaragua. 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 417 References 4 5 418 6 7 419 1. Ferlay J, Soerjomataram I, Ervik M, et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality 8 420 Worldwide: IARC CancerBase No. 11 [Internet], 2013, International Agency for Research on 9 421 Cancer: Lyon, France. 10 11 422 2. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of 12 423 invasive cervical cancer worldwide. J Pathol 1999;189(1):12-19. 13 424 3. Rodriguez AC, Schiffman M, Herrero R, et al. Longitudinal study of human papillomavirus 14 425 persistence and cervical intraepithelial neoplasia grade 2/3: critical role of duration of infection. 15 426 J Natl CancerFor Inst 2010;102(5):315-324. peer review only 16 427 4. Kitchener HC, Castle PE, and Cox JT. Chapter 7: Achievements and limitations of cervical cytology 17 18 428 screening. Vaccine 2006;24 Suppl 3:S3/63-70. 19 429 5. Denny L, Quinn M, and Sankaranarayanan R. Chapter 8: Screening for cervical cancer in 20 430 developing countries. Vaccine 2006;24 Suppl 3:S3/71-77. 21 431 6. ICO. HPV Information Centre, 2014. 22 432 7. Vastbinder MB, Castillo C, and Bekkers RL. Barriers to follow-up of abnormal Papanicolaou 23 433 smears among women in Leon, Nicaragua. Trop Doct 2010;40(1):22-26. 24 434 8. Khan MJ, Castle PE, Lorincz AT, et al. The elevated 10-year risk of cervical precancer and cancer 25 26 435 in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type- 27 436 specific HPV testing in clinical practice. J Natl Cancer Inst 2005;97(14):1072-1079. 28 437 9. WHO, in WHO Guidelines for Screening and Treatment of Precancerous Lesions for Cervical 29 438 Cancer Prevention2013: Geneva. 30 439 10. Arrossi S, Thouyaret L, Herrero R, et al. Effect of self-collection of HPV DNA offered by 31 440 community health workers at home visits on uptake of screening for cervical cancer (the EMA 32 441 study): a population-based cluster-randomised trial. Lancet Glob Health 2015;3(2):e85-94.

33 http://bmjopen.bmj.com/ 34 442 11. Moses E, Pedersen HN, Mitchell SM, et al. Uptake of community-based, self-collected HPV 35 443 testing vs. visual inspection with acetic acid for cervical cancer screening in Kampala, Uganda: 36 444 preliminary results of a randomised controlled trial. Trop Med Int Health 2015;20(10):1355- 37 445 1367. 38 446 12. Lazcano-Ponce E, Lorincz AT, Cruz-Valdez A, et al. Self-collection of vaginal specimens for human 39 447 papillomavirus testing in cervical cancer prevention (MARCH): a community-based randomised 40 41 448 controlled trial. Lancet 2011;378(9806):1868-1873. on September 26, 2021 by guest. Protected copyright. 42 449 13. Racey CS, Withrow DR, and Gesink D. Self-collected HPV testing improves participation in 43 450 cervical cancer screening: a systematic review and meta-analysis. Can J Public Health 44 451 2013;104(2):e159-166. 45 452 14. Jeronimo J, Bansil P, Lim J, et al. A multicountry evaluation of careHPV testing, visual inspection 46 453 with acetic acid, and papanicolaou testing for the detection of cervical cancer. Int J Gynecol 47 454 Cancer 2014;24(3):576-585. 48 49 455 15. Qiao YL, Sellors JW, Eder PS, et al. A new HPV-DNA test for cervical-cancer screening in 50 456 developing regions: a cross-sectional study of clinical accuracy in rural China. Lancet Oncol 51 457 2008;9(10):929-936. 52 458 16. Jeronimo J, Holme F, Slavkovsky R, et al. Implementation of HPV testing in Latin America. J Clin 53 459 Virol 2016;76 Suppl 1:S69-73. 54 460 17. Campos NG, Tsu, V., Jeronimo, J., Mvundura, M., Lee, K., Kim, J.J. When and how often to screen 55 56 461 for cervical cancer in three low- and middle-income countries: A cost-effectiveness analysis. 57 462 Papillomavirus Research 2015;1:38-58. 58 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 463 18. Campos NG, Burger EA, Sy S, et al. An updated natural history model of cervical cancer: 4 464 derivation of model parameters. Am J Epidemiol 2014;180(5):545-555. 5 6 465 19. WHO. Macroeconomics and health: Investing in health for economic development: Report of the 7 466 Commission on Macroeconomics and Health, 2001, World Health Organization: Geneva. 8 467 20. Tan-Torres Edejer T, Baltussen, R., Adam, T., Hutubessy, R., Acharya, A., Evans, D.B., Murray, 9 468 C.J.L., ed. Making Choices in Health: WHO Guide to Cost-Effectiveness Analysis. 2003, World 10 469 Health Organization: Geneva, Switzerland. 11 470 21. Jamison DT, Breman, J.G., Measham, A.R., Alleyne, G., Claeson, M., Evans, D.B., Jha, P., Mills, A., 12 13 471 Musgrove, P., ed. Disease Control Priorities in Developing Countries. Second edition ed. 2006, 14 472 Oxford University Press and The World Bank: Washington, D.C. 15 473 22. Munoz N,For Mendez F, Possopeer H, et al. Incidence, review duration, and onlydeterminants of cervical human 16 474 papillomavirus infection in a cohort of Colombian women with normal cytological results. J 17 475 Infect Dis 2004;190(12):2077-2087. 18 476 23. Herrero R, Hildesheim A, Rodriguez AC, et al. Rationale and design of a community-based 19 477 double-blind randomized clinical trial of an HPV 16 and 18 vaccine in Guanacaste, Costa Rica. 20 21 478 Vaccine 2008;26(37):4795-4808. 22 479 24. Surveillance, Epidemiology, and End Results (SEER) Program. SEER*Stat Database: Incidence - 23 480 SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases Nov 2011 Sub (1973- 24 481 2009 varying) – Linked to County Attributes – Total U.S., 1969-2010 Counties. Bethesda, MD: 25 482 National Cancer Institute, Division of Cancer Control and Population Sciences, SUrveillance 26 483 Research Program, Surveillance Systems Branch; 2012. 27 28 484 25. Sankaranarayanan R, Swaminathan R, Brenner H, et al. Cancer survival in Africa, Asia, and 29 485 Central America: a population-based study. Lancet Oncol 2010;11(2):165-173. 30 486 26. Stinnett AA and Paltiel AD. Estimating CE ratios under second-order uncertainty: the mean ratio 31 487 versus the ratio of means. Med Decis Making 1997;17(4):483-489. 32 488 27. Mayrand MH, Duarte-Franco E, Rodrigues I, et al. Human papillomavirus DNA versus

33 489 Papanicolaou screening tests for cervical cancer. N Engl J Med 2007;357(16):1579-1588. http://bmjopen.bmj.com/ 34 490 28. Ronco G, Dillner J, Elfstrom KM, et al. Efficacy of HPV-based screening for prevention of invasive 35 36 491 cervical cancer: follow-up of four European randomised controlled trials. Lancet 37 492 2014;383(9916):524-532. 38 493 29. Campos NG, Maza M, Alfaro K, et al. The comparative and cost-effectiveness of HPV-based 39 494 cervical cancer screening algorithms in El Salvador. Int J Cancer 2015. 40 495 30. Chirenje ZM, Rusakaniko S, Kirumbi L, et al. Situation analysis for cervical cancer diagnosis and 41 496 treatment in east, central and southern African countries. Bull World Health Organ on September 26, 2021 by guest. Protected copyright. 42 497 2001;79(2):127-132. 43 44 498 31. Cremer M, Bullard K, Maza M, et al. Cytology versus visual inspection with acetic acid among 45 499 women treated previously with cryotherapy in a low-resource setting. Int J Gynaecol Obstet 46 500 2010;111(3):249-252. 47 501 32. Sauvaget C, Muwonge R, and Sankaranarayanan R. Meta-analysis of the effectiveness of 48 502 cryotherapy in the treatment of cervical intraepithelial neoplasia. Int J Gynaecol Obstet 49 503 2013;120(3):218-223. 50 51 504 33. Bigoni J, Gundar M, Tebeu PM, et al. Cervical cancer screening in sub-Saharan Africa: a 52 505 randomized trial of VIA versus cytology for triage of HPV-positive women. Int J Cancer 53 506 2015;137(1):127-134. 54 507 34. Catarino R, Vassilakos P, Scaringella S, et al. Smartphone Use for Cervical Cancer Screening in 55 508 Low-Resource Countries: A Pilot Study Conducted in Madagascar. PLoS One 56 509 2015;10(7):e0134309. 57 58 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 510 35. Pan QJ, Hu SY, Guo HQ, et al. Liquid-based cytology and human papillomavirus testing: a pooled 4 511 analysis using the data from 13 population-based cervical cancer screening studies from China. 5 6 512 Gynecol Oncol 2014;133(2):172-179. 7 513 36. Qiao YL, Jeronimo J, Zhao FH, et al. Lower cost strategies for triage of human papillomavirus 8 514 DNA-positive women. Int J Cancer 2014;134(12):2891-2901. 9 515 37. Tebeu PM, Fokom-Domgue J, Crofts V, et al. Effectiveness of a two-stage strategy with HPV 10 516 testing followed by visual inspection with acetic acid for cervical cancer screening in a low- 11 517 income setting. Int J Cancer 2015;136(6):E743-750. 12 13 518 38. Richardson LA, El-Zein M, Ramanakumar AV, et al. HPV DNA testing with cytology triage in 14 519 cervical cancer screening: Influence of revealing HPV infection status. Cancer Cytopathol 15 520 2015;123(12):745-754.For peer review only 16 521 39. World Development Indicators, 2016, World Bank. 17 522 40. Msyamboza KP, Phiri T, Sichali W, et al. Cervical cancer screening uptake and challenges in 18 523 Malawi from 2011 to 2015: retrospective cohort study. BMC Public Health 2016;16(1):806. 19 524 41. Campbell C, Kafwafwa S, Brown H, et al. Use of thermo-coagulation as an alternative treatment 20 21 525 modality in a 'screen and treat' programme of cervical screening in rural Malawi. Int J Cancer 22 526 2016. 23 527 42. Gage JC, Rodriguez AC, Schiffman M, et al. Treatability by cryotherapy in a screen-and-treat 24 528 strategy. J Low Genit Tract Dis 2009;13(3):174-181. 25 529 43. Starks D, Arriba LN, Enerson CL, et al. Mexican Cervical Cancer Screening Study II: 6-month and 26 530 2-year follow-up of HR-HPV women treated with cryotherapy in a low-resource setting. J Low 27 28 531 Genit Tract Dis 2014;18(4):333-337. 29 532 44. McClung EC and Blumenthal PD. Efficacy, safety, acceptability and affordability of cryotherapy: a 30 533 review of current literature. Minerva Ginecol 2012;64(2):149-171. 31 32 534

33 http://bmjopen.bmj.com/ 535 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 24 of 72

-- -- 1.0/1.0 0.40/0.85 0.70/0.65 0.70/0.95 0.40/0.85 0.90/0.50 CIN2: 60%CIN2: 40% - 85% - 40% 40% - 85% - 40% 50% - 80% - 50% ≤CIN1: 75%≤CIN1: screening screening intervals screening intervals screening screening intervals Sensitivity Sensitivity Analysis: Alternative sensitivity/specificity pairs: 0.95/0.68, with

40% 85% 70% 24 24 0.41/0.94 Alternative sensitivity/specificity pair: 0.95/0.68 0.67/0.86 0.78/0.89 0.85/0.55 treatment treatment treatment CIN2: 85%CIN2: Base Case: BMJ Open http://bmjopen.bmj.com/

Upon ruling out cancer, all referred to Upon ruling out cancer, all referred to a 100% ≤CIN1: on September 26, 2021 by guest. Protected copyright. [29, 42] [29, 0.60/0.75 Alternative sensitivity/specificity pairs:

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

d For peer review only

c [14, 33, 35,[14, 38] 33, [14, 33, 34, 36, 37] b Test/Parameter

[6] [14, 17]

[14] [14] Visit compliance,Visit referral facility VIA (triage of (triage VIA HPV+) Pap 27, 28](primary) [14, Colposcopy (HPV+ Colposcopy (HPV+ women ineligible for cryotherapy) Colposcopy (ASCUS+ women) to routineAccess screening, theof % target population Visit compliance,Visit screening facility Colposcopy (HPV+/VIA+ women ineligible ST for cryotherapy) careHPV (primary),careHPV self-collection vaginal of samples Colposcopy (HPV+/Pap+ women) Upon ruling out cancer, all referred to Pap HPV+)of (triage careHPV (primary),careHPV provider-collection cervicalof samples Eligibility screen-and-treatfor cryotherapy Table 1. Baseline values and ranges for model variables. model for ranges and values Baseline 1. Table Screening/triage Screening/triage test performance (sensitivity/specificity to CIN2+)detect Treatment eligibility and efficacy Colposcopy performance (sensitivity/specificity to CIN1+)detect andCoverage compliance 536 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

f e

3 3 -- -- 85% 75% 6.91 30.54 0% - 5% - 0% CIN3: 49%CIN3: 0% to 50% ofto 0% 50% case base 0% to 50% ofto 0% 50% case base ofto 0% 50% case base ofto 0% 50% case base ofto 0% 50% case base 0% to 50% ofto 0% 50% case base 50% - 150% - 50% case base of 75% - 125% - 75% case base of 75% - 125% - 75% case base of 50% - 150% - 50% case base of 50% - 150% - 50% case base of 50% - 150% - 50% case base of -- --

3% 96% 92% 4.19 1.75 0.82 3.10 0.48 7.26 2.81 0.41 19.91 68.36 11.96 11.04 18.16 25 25 CIN3: 75%CIN3: collection) cryotherapy BMJ Open http://bmjopen.bmj.com/ 10% 15% 1.51 per woman1.51 receiving VIA and/or 0.04 per woman0.04 screened with careHPV 0.08 per woman0.08 screened with (self-careHPV

d [44] on September 26, 2021 by guest. Protected copyright.

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

For peer review only [30]

[43] d

[14] [30-32]

Colposcopy/biopsy Proportion maintaining women of an HPV infection colposcopicfollowing diagnosis and treatment LEEP LEEP careHPV testcareHPV (provider-collection) testcareHPV (self-collection) VIA triage testtriage VIA Treatment rate cure following colposcopy (LEEP for CIN2+; cryotherapy)else Cryotherapy Outreach workers/auxiliary nurses (careHPV self- collection) Laboratory technicians (careHPV) Health personnelcare (careHPV) per woman 0.09 screened with careHPV Screen-and-treat cryotherapy HPV+ rate cure (for or HPV+/VIA+) Transportation to screening facility (round-trip) Wait referral time, facility Transport time, screening facility Transport time, referral facility Transportation to referral facility (round trip) Wait screening time, facility Health providerscare (VIA and cryotherapy) Proportion maintaining women of an HPV infection cryotherapyfollowing Pap test Discount Discount rate costsfor and life-years medical Direct costs, screening and of treatment (2015precancer US$) Programmatic costs, training US$) (2015 Women’s time Women’s and transportation (2015 US$) costs

Page 25 of 72 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 26 of 72

case) case) case) case) 918 - 2,920 - 918 (direct only;medical of 150% base 944 - 2,229 - 944 (direct only;medical of 150% base

918 390 640 944 197 346 1,946 1,486 26 26 BMJ Open http://bmjopen.bmj.com/

on September 26, 2021 by guest. Protected copyright. [29] For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

g g

h h

Direct non-medicalDirect Direct medicalDirect Women’s timeWomen’s timeWomen’s Direct medicalDirect Direct non-medicalDirect

- Total Total Regional andRegional distant cancer - - - Local cancer - - Test performance characteristics of colposcopy in the START-UP demonstration project were derived from the worst diagnosis of the local the local of diagnosis the worst from derived were project demonstration START-UP the in colposcopy of characteristics Test performance In sensitivity analysis, we considered the direct medical cost of colposcopy to be equivalent to the cost of colposcopy alone, without biopsy. biopsy. without alone, colposcopy of the cost to equivalent be to of colposcopy cost medical the direct considered we analysis, sensitivity In Unpublished data from the Scale-Up Nicaragua project. Further details on costing data are provided in the Appendix. Appendix. in the are provided data on costing details Further project. Nicaragua Scale-Up the from data Unpublished meals. and housing, temporary facility, tertiary to a transportation include costs Direct non-medical Compliance is defined as the proportion of women who return for each clinical encounter, relative to the previous visit. visit. previous the to relative encounter, clinical each for return who women of proportion the as defined is Compliance In sensitivity analysis, we considered the direct medical cost of cryotherapy to include the upper bound of cryotherapy equipment costs costs equipment cryotherapy of bound the upper include to of cryotherapy cost medical the direct considered we analysis, sensitivity In Includes woman’s time an support person’s time. time. person’s support an time woman’s Includes ASCUS+: atypical squamous cells of undetermined significance or higher; CIN 1+: cervical intraepithelial neoplasia grade 1 or higher; CIN2+: CIN2+: higher; or 1 grade neoplasia intraepithelial cervical 1+: CIN higher; or significance of undetermined cells squamous atypical ASCUS+: (assumptions); programmatic costsw (assumptions); cancer costs (assumptions). (assumptions). costs cancer (assumptions); costsw programmatic (assumptions); treat; US$: 2015 United States dollars; VIA: visual inspection with acetic acid. Parameter values for sensitivity analysis were determined as as determined were analysis sensitivity for values Parameter acid. acetic with inspection visual VIA: dollars; States United 2015 US$: treat; follows: screening and triage test performance (cited literature); colposcopy performance (assumption); coverage and compliance compliance and coverage (assumption); performance colposcopy literature); (cited performance test triage and screening follows:

a b c d e f g h Cost of cancerof Cost treatment (rounded) (2015 US$) (assumptions); treatment eligibility and efficacy (cited literature); discount rate (assumptions); direct medical costs (assumptions, with the the with (assumptions, costs medical direct (assumptions); rate discount literature); (cited efficacy and eligibility treatment (assumptions); that were inadequate or with a histological classification other than negative, CIN1, CIN2, CIN3, or cancer. Because CIN1 is not a true underlying underlying a true not is CIN1 Because cancer. or CIN3, CIN2, CIN1, negative, than other classification a histological with or inadequate were that (assuming the lowest number of women treated per year per facility, in Carazo, Chontales, or Chinandega) and the upper bound of cryotherapy of cryotherapy bound the upper and Chinandega) or Chontales, Carazo, in facility, year per per treated women of number lowest the (assuming exception of colposcopy and cryotherapy, which were based on ranges suggested by Scale-Up and START-UP data); women’s costs costs women’s data); START-UP and by Scale-Up suggested on ranges based were which cryotherapy, and of colposcopy exception health state in the microsimulation model, performance of colposcopy in the model is based on the underlying health states of no lesion, HPV HPV no lesion, of states health underlying on the based is model the in colposcopy of performance model, microsimulation the in state health supply costs (assuming the lowest number of women treated per gas tank in any facility in Carazo, Chontales, or Chinandega). or Chinandega). Chontales, Carazo, in facility any in tank gas per treated women of number lowest the (assuming costs supply cervical intraepithelial neoplasia grade 2 or higher; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; ST: screen-and- ST: procedure; excision electrosurgical LEEP: loop papillomavirus; human HPV: higher; or 2 grade neoplasia intraepithelial cervical infection, CIN2, or CIN3. For a treatment threshold of CIN1, we weighted sensitivity of colposcopy for women with HPV based on the country- on the based HPV with women for of colposcopy sensitivity weighted we CIN1, of threshold a treatment For or CIN3. CIN2, infection, pathologist relative to the worst diagnosis by a quality control pathologist (gold standard); we applied the treatment threshold of CIN1+, CIN1+, of threshold the treatment we applied standard); (gold pathologist control quality by a diagnosis worst the to relative pathologist specific prevalence of CIN1 among women with HPV infections in the START-UP studies. studies. START-UP the in infections HPV with women among of CIN1 prevalence specific although this was not the treatment threshold in START-UP. To derive test performance of colposcopy, we excluded histological classifications classifications histological excluded we colposcopy, of performance test derive To START-UP. in threshold treatment the was not this although 542 538 543 551 539 558 540 546 554 552 557 541 547 555 537 548 553 544 549 545 550 556 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 27 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 559 Figure legends. 4 5 6 560 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible 7 8 561 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, 9 10 562 conditional on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 11 12 13 563 visits for screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which 14 15 564 requires 3 or moreFor visits for screeningpeer and necessary review treatment; c) HPVonly testing followed by visual 16 17 565 inspection with acetic acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 18 19 566 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV- 20 21 22 567 positive women every 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary 23 24 568 treatment. ASCUS+: Pap result of atypical squamous cells of undetermined significance or worse; CIN: 25 26 569 cervical intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision 27 28 29 570 procedure; VIA: visual inspection with acetic acid. 30 31 571 Figure 2. Cervical cancer screening cost per woman over duration of screening eligibility, by cost 32

33 572 component: Pap testing (every 3 years) versus careHPV testing (every 5 years). Bars indicate the http://bmjopen.bmj.com/ 34 35 36 573 undiscounted cost (2015 US$) of screening with Pap testing (offered 10 times between ages 30 and 59 37 38 574 years) versus careHPV testing (offered 6 times between ages 30 and 59 years), by cost component. Only 39 40 575 screening costs are shown; costs associated with recommended management following a positive 41 on September 26, 2021 by guest. Protected copyright. 42 576 screening test are not included. 6x: delivered 6 times over the course of screening eligible ages 30 to 59; 43 44 45 577 10x: delivered 10 times over the course of screening eligible ages. 46 47 578 Figure 3. Cost-effectiveness analysis: Base case results. The graph displays the discounted lifetime costs 48 49 579 (x-axis; in 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing 50 51 52 580 every 3 years; careHPV screen-and-treat every 5 years [HPV-Cryo]; careHPV every 5 years with visual 53 54 581 inspection with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years 55 56 582 with Pap triage of HPV-positive women [HPV-Pap]), under base case assumptions. The cost-effectiveness 57 58 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 583 associated with a change from one strategy to a more costly alternative is represented by the difference 4 5 6 584 in cost divided by the difference in life expectancy associated with the two strategies. The curve 7 8 585 indicates the strategies that are efficient because they are more effective and either 1) cost less; or 2) 9 10 586 have a more attractive cost-effectiveness ratio than less effective options. The incremental cost- 11 12 13 587 effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two strategies under 14 15 588 comparison. In theFor base case analysis,peer HPV-Cryo review every 5 years was lessonly costly and more effective than 16 17 589 other screening strategies considered, and was thus a dominant strategy with an ICER of US$320 per 18 19 590 year of life saved. HPV-Pap: HPV testing with Pap triage of HPV-positive women; HPV-Cryo: HPV screen- 20 21 22 591 and-treat strategy; HPV-VIA: HPV testing with visual inspection with acetic acid (VIA) triage of HPV- 23 24 592 positive women; yrs: years. 25 26 593 Figure 4. Reduction in lifetime risk of cervical cancer, by compliance level. Bars indicate the percent 27 28 29 594 reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 years; 30 31 595 careHPV screen-and-treat every 5 years [HPV-Cryo]; careHPV every 5 years with visual inspection with 32

33 596 acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of http://bmjopen.bmj.com/ 34 35 36 597 HPV-positive women [HPV-Pap]) as compliance per visit within a screening episode increases. 37 38 598 Compliance is defined as the proportion of women who return for each clinical encounter, relative to 39 40 599 the previous visit. Coverage of the target population is assumed to be 70%. While the base case analysis 41 on September 26, 2021 by guest. Protected copyright. 42 600 assumed 85% compliance for visits at screening facilities and 40% compliance for visits at referral 43 44 45 601 facilities (for diagnosis and treatment), the graph displays cancer risk reduction assuming the specified 46 47 602 compliance level at all visits, regardless of facility type. 48 49 603 Figure 5. Base case and sensitivity analysis: Incremental cost-effectiveness ratios, HPV cryotherapy 50 51 52 604 (HPV-Cryo) strategy. Incremental cost-effectiveness ratios (ICERs) are presented (x-axis, 2015 US$ per 53 54 605 year of life saved) for the base case and sensitivity analyses (y-axis). The blue bars represent the range of 55 56 606 the ICERs for HPV-Cryo every 5 years across the 50 input parameter sets, with the ICER of the mean 57 58 59 60 28

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 607 costs divided by the mean effects demarcated by a black line. The dashed blue line indicates Nicaragua’s 4 5 6 608 per capita gross domestic product (GDP), at US$2,090, assuming this is the threshold that designates 7 8 609 interventions as “very cost-effective”. 9 10 610 11 12 13 611 14 15 612 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 29

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-STHPV-Cryo), which 36 requires 3 or more visits for screening and necessary treatment; c) HPV testing followed by visual inspection 37 with acetic acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive 38 women every 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. 39 ASCUS+: Pap result of atypical squamous cells of undetermined significance or worse; CIN: cervical 40 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 41 visual inspection with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 2. Cervical cancer screening cost per woman over duration of screening eligibility, by cost

33 axis; in 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing every 3 http://bmjopen.bmj.com/ 34 years; careHPV screen-and-treat every 5 years [HPV-Cryo]; careHPV every 5 years with visual inspection 35 with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage 36 of HPV-positive women [HPV-Pap]), under base case assumptions. The cost-effectiveness associated with a 37 change from one strategy to a more costly alternative is represented by the difference in cost divided by the difference in life expectancy associated with the two strategies. The curve indicates the strategies that are 38 efficient because they are more effective and either 1) cost less; or 2) have a more attractive cost- 39 effectiveness ratio than less effective options. The incremental cost-effectiveness ratio (ICER) is the 40 reciprocal of the slope of the line connecting the two strategies under comparison. In the base case analysis, 41 HPV-Cryo every 5 years was less costly and more effective than other screening strategies considered, and on September 26, 2021 by guest. Protected copyright. 42 was thus a dominant strategy with an ICER of US$320 per year of life saved. HPV-Pap: HPV testing with Pap 43 triage of HPV-positive women; HPV-Cryo: HPV screen-and-treat strategy; HPV-VIA: HPV testing with visual 44 inspection with acetic acid (VIA) triage of HPV-positive women; yrs: years. 45 254x190mm (300 x 300 DPI) 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 4. Reduction in lifetime risk of cervical cancer, by compliance level. Bars indicate the percent

33 reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 years; careHPV http://bmjopen.bmj.com/ 34 screen-and-treat every 5 years [HPV-Cryo]; careHPV every 5 years with visual inspection with acetic acid 35 [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of HPV-positive 36 women [HPV-Pap]) as compliance per visit within a screening episode increases. Compliance is defined as 37 the proportion of women who return for each clinical encounter, relative to the previous visit. Coverage of the target population is assumed to be 70%. While the base case analysis assumed 85% compliance for 38 visits at screening facilities and 40% compliance for visits at referral facilities (for diagnosis and treatment), 39 the graph displays cancer risk reduction assuming the specified compliance level at all visits, regardless of 40 facility type. 41 on September 26, 2021 by guest. Protected copyright. 42 254x190mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 5. Base case and sensitivity analysis: Incremental cost-effectiveness ratios, HPV cryotherapy (HPV-

33 Cryo) strategy. Incremental cost-effectiveness ratios (ICERs) are presented (x-axis, 2015 US$ per year of http://bmjopen.bmj.com/ 34 life saved) for the base case and sensitivity analyses (y-axis). The blue bars represent the range of the 35 ICERs for HPV-Cryo every 5 years across the 50 input parameter sets, with the ICER of the mean costs 36 divided by the mean effects demarcated by a black line. The dashed blue line indicates Nicaragua’s per 37 capita gross domestic product (GDP), at US$2,090, assuming this is the threshold that designates interventions as “very cost-effective”. 38

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 90 90 /126 /126 /126 /126 #/ line # line #/ on page page on Reported Reported Page 38 of 72 NA NA 24/527 24/527 24/527 24/527 8/129 8/129 8/139 8/139 8/129 8/129 10/170- 1 8/137 8/137 7/119; 8 7/119; 8 7/112 7/112 2/18-42 2/18-42 1/1 1/1

Recommendation BMJ Open http://bmjopen.bmj.com/ : Describe fully Describe : the design features single of the effectivenessstudy and why the single :fully Describe themethods for used identification of included studies and synthesis of on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

study sufficient study wasa source of clinical effectivenessdata If applicable, describeIf the applicable, population andmethods used to preferences elicit for outcomes Synthesis-based estimates clinical clinical effectivenessdata. Single study-based Single estimates Describe Describe what outcomes used were asthe measure(s) theof benefit in theirevaluation and relevance for the typeof Report the choice Report ofdiscount rate(s) for used costs and outcomes say and why appropriate. State the State time horizon(s) costs over which consequences and are beingevaluated and say whyappropriate. Describe the Describe interventions strategiesor being compared and state they why werechosen. Describe the Describe perspective study of the andthis relate to costs the being evaluated. State relevant State aspectsof the system(s) the decision(s) which in need(s) be made.to Describe characteristics Describe base of case the population and subgroups analysed, including they chosen. were why Provide anProvide statement explicit of the broadercontext for study. the study Present the question relevance and its for health policy practiceor decisions. Provide a structured a Provide summary of objectives, perspective, setting, methods study(including design and inputs), (includingresults case base and uncertainty analyses),and conclusions. Identify the Identify study asan economicmoreuseor evaluation specific terms suchas“cost-effectiveness analysis”, and the describe interventionscompared. For peer review only analysis performed.

2 1 4 9 6 8 7 5 12 # # 11b 11a

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist AND ABSTRACT AND Section/topic Measurementand valuation of preference based

Measurementof effectiveness Choice of health outcomes 10 Discount rate Discount Timehorizon Comparators Study Study perspective Settingand location Targetpopulation and subgroups METHODS METHODS Backgroundand objectives 3 INTRODUCTION INTRODUCTION Title Title TITLE TITLE Abstract Abstract Abstract 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

ppendix ppendix 57-299 57-299 49 49 67 67 67 .2/ 63-.2/ .2-4/ 58-.2-4/ Reported Reported on page # page on p. 1-4 p. 1-4 p. 6/172 p. 6/172 68 68 155 155 13-14/ 13-14/ 2 NA NA 27/569 27/569 24/527- 5 9/154- 1 8/143; 8/143; 8/143- 1 NA NA p p Appendix Appendix A Appendix, Appendix, Appendix, Appendix,

hecklist item hecklist C BMJ Open http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

sourcesfor to distributions used represent uncertainty appropriate. where table Providinga to the show input values recommended. strongly is costs to the coststhe ofto reported costs year if Describe necessary. methods for converting costscommon a into base currency the and exchangerate. Model-basedeconomic evaluation Describe : the effects the results on for of uncertainty all parameters, input and Single study-based Single economic evaluation Describe : the effects sampling uncertaintyfor of the estimated incremental For each Foreach report intervention, meanvalues for main the categories of estimated costs outcomes and as of interest, Report the Report values,ranges, references, and, if used, distributions probability for all parameters/ reasons Report or Describe all Describe analytical methods supporting the evaluation. This could methods include for skewed,dealing with censoredor missing, data; methods; extrapolation methods for pooling approachesdata; to validate or make adjustmentscorrections) (such cycle ashalf to model; a and methods for handling populationheterogeneity and uncertainty. Describe all Describe structural otheror assumptions underpinning the decision-analytical model. Describe and Describe reasons give forspecific of type the decision-analyticalmodel used.figure Providing a to show model structurestrongly is recommended. Report the Report datesestimated of the resource quantities costs.and unit Describe methods for adjustingestimated unit Model-basedeconomic evaluation Describe : approaches and datasources used to estimate resource use Single study-based Single economic evaluation Describe : approaches to used estimate resource use associated with the alternativeinterventions. Describe primary secondaryor research methods for valuing each resource item in termsof cost. unit Describe its adjustments any made approximate to opportunity to costs. uncertaintyrelated structure to the model of the andassumptions. cost and costand incrementaleffectiveness parameters, together withthe impact methodologicalof assumptions (suchas study discountrate, perspective). well asmean well differences between the comparator groups. If applicable, report incremental cost-effectiveness ratios. associatedmodel health with states. Describe primary secondaryor research methods for resourcevaluing each For peertermsitem in cost. of unit Describe its adjustments any made approximate to opportunity to costs. review only

19 14 18 16 15 17 # 20b 13b 13a

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist Section/topic outcomes

Characterising Characterising uncertainty 20a Incrementalcosts and outcomes Study Study parameters RESULTS RESULTS Choiceof model Currency, Currency, price anddate, conversion

Estimating resources Estimating and costs

Analyticalmodels Assumptions Page 39 of 72 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

12-407 12-407 Page 40 of 72 Reported Reported on page # page on 4/ 63-67 4/ 63-67 4/ 57-62 4/ 57-62 16-20/ 16-20/ 3 NA NA

Checklist item Checklist BMJ Open http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

generalisability generalisability findings of the findings and how the fit current with knowledge. Describe any any Describe potentialfor conflictof interest of study contributors in accordance journal with policy. Int eh absence of journal policy,a recommend we comply authors Internationalwith Committee Journalof Medical Editors recommendations. Describe how Describe the study wasand the funded role ofthe funder in the identification, conduct, design, and reporting of Summarise key studySummarise key findings and describehow supportthey conclusions the reached. Discusslimitations and the If applicable, reportIf differencesapplicable, costs, outcomes, in cost-effectivenessor can that explained be by variations between subgroupsof patients with differentbaseline characteristics otheror observed variability in effectsnot are that more reducible by information. For peeranalysis. the Describe othersources non-monetary of support. review only

22 24 23 #

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist

Section/topic

Characterising Characterising heterogeneity 21

Conflictsof interest Source Source offunding Other Other Study findings, Study limiations, generalisability,and current knowledge DISCUSSION DISCUSSION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 41 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 Supplemental Appendix: 4 5 2 The Cost‐Effectiveness of HPV‐Based Screening in the Public Health System in Nicaragua 6 7 3 8 4 MODEL CALIBRATION 9 5 Overview of the calibration process 10 6 Details of model development for the U.S. and Nicaragua models, including initial parameterization and 11 7 calibration, have been published elsewhere [1, 2]. Baseline HPV incidence rates, as a function of 12 8 genotype and age, were derived from a published prospective study of sexually active women aged 15‐ 13 14 9 85 years in Bogota, Colombia [3]. Because HPV incidence is not necessarily similar in Colombia and 15 10 Nicaragua due to Fordifferences inpeer sexual behavior, review we considered age ‐specificonly HPV incidence and natural 16 11 immunity following initial infection as candidate parameters for calibration. Type‐specific data on CIN2 17 12 and CIN3 regression and progression are limited [4‐9], so these highly uncertain parameters were also 18 13 candidates for calibration. Time‐dependent rates of HPV clearance and progression by genotype were 19 14 informed by primary longitudinal data from the control arm of the Costa Rica Vaccine Trial [10], 20 21 15 assuming that rates of HPV clearance and progression to precancer are similar across populations and 22 16 vary only by time since infection and HPV genotype. 23 17 24 18 To calibrate the model, we selected empirical epidemiologic data from Nicaragua, including age‐specific 25 19 prevalence of high‐risk HPV and age‐specific cancer incidence (see below for further detail). We then set 26 20 plausible search ranges around baseline model input values for age‐ and type‐specific HPV incidence; 27 21 natural immunity following initial infection; and progression and regression of CIN. We performed 28 29 22 repeated model simulations in the absence of any preventive intervention (i.e., natural history 30 23 simulations). For each simulation, we randomly selected a single value for each of the uncertain 31 24 parameters from the identified search range, creating a unique vector of parameter values (i.e., 32 25 parameter “set”). From over 1,475,000 repeated samplings, we identified the parameter sets with the

33 26 highest correspondence to the empirical calibration target data from Nicaragua by calculating and http://bmjopen.bmj.com/ 34 27 aggregating the log‐likelihood of model‐projected outcomes. We used the 50 parameter sets with the 35 36 28 highest likelihood scores (i.e., best overall fit to the empirical data from Nicaragua) for analysis to 37 29 capture uncertainty in the model parameters as a form of probabilistic sensitivity analysis. We report 38 30 results as a mean of outcomes across these top 50 parameter sets; incremental cost‐effectiveness ratios 39 31 are reported as the ratio of the mean costs divided by the mean effects of one strategy versus another 40 32 across sets. 41 33 on September 26, 2021 by guest. Protected copyright. 42 34 Calibration targets 43 44 35 We assessed model fit by comparing projected model outcomes of age‐specific prevalence of high‐risk 45 36 HPV and age‐specific cancer incidence relative to empirical data from Nicaragua. The scoring algorithm 46 37 included age‐specific prevalence of high‐risk HPV and age‐specific cervical cancer incidence. 47 38 48 39 Age‐specific prevalence of high‐risk HPV was drawn from the Screening Technologies to Advance Rapid 49 40 Testing for Cervical Cancer Prevention–Utility and Program Planning (START‐UP) project data on 50 41 careHPV positivity using a cut‐off ratio cut‐point of 0.5 relative light units (Table A.1). For each age 51 52 42 group, we derived a 95% binomial confidence interval around the point prevalence, which comprised 53 43 the calibration target. The likelihood function for each age group was assumed to follow a binomial 54 44 distribution. 55 45 56 46 Age‐specific cancer incidence was drawn from Globocan due to the lack of a local cancer registry [11] 57 47 (Table A.2). The likelihood function for each age group was assumed to follow a normal distribution. 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 48 4 5 49 Composite goodness‐of‐fit scores for each input parameter set were generated by summing the log 6 50 likelihood of each model outcome (i.e., age‐specific HPV prevalence, age‐specific cancer incidence). The 7 51 50 input parameter sets with the highest goodness‐of‐fit scores yielded the model outputs that were 8 52 simultaneously closest to all calibration targets, and these were selected for analysis. Figures A.1 and 9 53 A.2 display model fit to epidemiologic data on age‐specific prevalence of high‐risk HPV and age‐specific 10 54 cancer incidence. 11 55 12 13 56 Transition probabilities for the calibrated natural history model are presented in Table A.3. 14 57 15 58 COST DATA For peer review only 16 59 17 60 Direct Medical Costs: Screening, Diagnosis, and Treatment of Precancerous Lesions 18 61 The direct medical costs of screening, diagnosis, and treatment of precancerous lesions were drawn 19 20 62 from the START‐UP demonstration study in Masaya department and from the Scale‐Up Nicaragua 21 63 Project in the departments of Carazo, Chontales, and Chinandega. Direct medical costs included clinical 22 64 staff time, clinical supplies, drugs, clinical equipment, laboratory staff time, laboratory supplies, and 23 65 laboratory equipment. Costs from the START‐UP study were collected in 2010 local currency units and 24 66 inflated to year 2015 levels using Nicaragua GDP deflators and then converted to US$ using the official 25 67 exchange rate [12]; the exception was for equipment, which was generally procured in the United States 26 68 and was inflated to year 2015 levels using US GDP deflators. Costs from the Scale‐Up Nicaragua project 27 28 69 (including equipment) were collected in 2015 local currency units and converted to US$ using the official 29 70 exchange rate [12]. 30 71 31 72 Costs are reported in Table A.4 (screening procedures) and Table A.5 (precancer diagnostic and 32 73 treatment procedures).

33 74 http://bmjopen.bmj.com/ 34 75 Women’s Time and Transportation Costs 35 36 76 To account for the opportunity cost of women’s time spent traveling to, waiting for, or receiving care, 37 77 we valued women’s time using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 US$. 38 78 While the minimum wage may be a conservative estimate for valuation of women’s time if they are 39 79 formally employed, it can serve as a proxy for the societal value of women’s time spent working in the 40 80 informal sector or at home. We converted this to an hourly wage rate assuming 171 hours of work per 41 81 month (40 hours per week). on September 26, 2021 by guest. Protected copyright. 42 43 82 44 83 Estimates for time spent traveling to, waiting for, and receiving care were dependent upon the facility 45 84 level where care was assumed to take place (Table A.6). While screening and delivery of primary 46 85 screening results were assumed to take place at basic primary health facilities, we assumed cryotherapy 47 86 and LEEP treatment were only provided at referral facilities. Estimates of women’s time spent receiving 48 87 a procedure were based on site‐specific data from the START‐UP demonstration projects, with staff time 49 88 spent on the procedure (excluding preparation and registration time, which we assumed were built into 50 51 89 patient waiting time) used as a proxy for women’s procedure time. Estimates of women’s wait time for 52 90 screening and delivery of primary screening results was based on the average wait time for screening at 53 91 facilities in the Chinandega department; wait time for cryotherapy, colposcopy, and LEEP was based on 54 92 the average wait time at referral centers in the Chontales department. Round‐trip transportation time 55 93 was based on Scale‐Up project staff estimates of the typical amount of time women spent traveling to 56 94 screening or referral facilities, using a sample of 9 facilities the Carazo department and 10 facilities in the 57 95 Chontales and Chinandega departments. We averaged across screening and referral facilities within 58 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 96 each department to obtain a department average, and then averaged across departments to obtain 4 5 97 average round‐trip transportation times based on facility level. Estimates of women’s time and time 6 98 costs are presented in Table A.7. 7 99 8 100 Women’s transportation costs were based on Scale‐Up project staff reports of the typical mode of 9 101 transportation to a particular facility by women in the catchment area, using a sample of 9 facilities in 10 102 Carazo and 10 facilities each in Chontales or Chinandega. We used the most commonly reported mode 11 103 of transportation for each department (for screening facilities: walking for Carazo; walking for 12 13 104 Chontales; bus for Chinandega; for referral facilities: bus in all three departments). The cost of a bus trip 14 105 was based on typical bus fare for the distance to the health facility, as reported by Scale‐Up project staff. 15 106 We averaged the Fortypical transportation peer costs acrossreview departments to obtainonly an estimate of women’s 16 107 round‐trip transportation costs to and from a health facility, assuming the same mode of transport was 17 108 used for the trip to and from the health facility. Women’s transportation costs are presented in Table 18 109 A.8. 19 20 110 21 111 Programmatic Costs 22 112 We included the programmatic cost of the following types of training sessions that were conducted 23 113 during the Scale‐Up project: 1) for health care personnel to offer careHPV screening and counsel women 24 114 on results; 2) for laboratory technicians to run the careHPV assay; 3) for community outreach workers 25 115 and auxiliary nurses to promote careHPV screening and offer self‐collection of HPV samples; and 4) for 26 116 health care providers to offer VIA and cryotherapy. We included the cost of trainer time and trainee 27 28 117 time (valued using average monthly salaries converted to an hourly rate; volunteer outreach worker 29 118 time was valued using the minimum wage), as well as the cost of supplies, materials, transportation, and 30 119 refreshments. To derive an average health care personnel training cost per woman screened, we divided 31 120 the total training cost for health care personnel by 100,000, corresponding to the number of women 32 121 screened during the Scale‐Up project. To derive an average laboratory technician training cost per

33 122 woman screened, we divided the total training cost for laboratory technicians by 100,000. To derive an http://bmjopen.bmj.com/ 34 123 average community outreach worker training cost per woman screened, we divided the totalg trainin 35 36 124 cost for outreach workers by 80,000, corresponding to the number of women who participated in self‐ 37 125 collection of HPV samples. To derive an average health care provider training cost per woman receiving 38 126 VIA and cryotherapy, we divided the total VIA/cryotherapy training cost for health care providers by 39 127 16,000, corresponding to the approximate number of women who tested HPV‐positive in Scale‐Up and 40 128 would thus require visual assessment (to determine eligibility for cryotherapy) and, if eligible, 41 129 cryotherapy. on September 26, 2021 by guest. Protected copyright. 42 43 130 44 131 All women screened with HPV testing incurred the per‐woman training costs for health care personnel 45 132 and laboratory technicians. Only women screened with HPV self‐collection incurred the per‐woman cost 46 133 of training outreach workers to offer self‐collection (due to lack of data on self‐collection in a 47 134 community setting, in the base case analysis we assumed self‐collection took place at the clinic; 48 135 however, we assumed outreach workers would be involved in mobilizing and educating women about 49 136 self‐collection, and conservatively included the cost of training so as to represent a known 50 51 137 programmatic cost as self‐collection efforts shift to also include community settings). Only women who 52 138 received either VIA (as triage) or screen‐and‐treat cryotherapy incurred the per‐woman cost of training 53 139 health providers to perform VIA and cryotherapy. Training costs are presented in Table A.9. 54 140 55 141 Cost of Cancer Care by Stage 56 142 Costs associated with cancer care by stage (Local versus Regional or Distant), including direct medical 57 143 costs, direct non‐medical costs (i.e., transport to a cancer center, and meals and temporary housing 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 144 during the course of treatment), and patient and support person time were derived from a previous 4 5 145 analysis in El Salvador [14]. Based on expert opinion in El Salvador, we assumed 50% of local cancer 6 146 patients presented with FIGO stage 1a, requiring radical hysterectomy, and 50% of local cancer patients 7 147 presented with FIGO stage 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions 8 148 followed by brachytherapy. We assumed all women presenting with FIGO stage 2b or higher received 9 149 cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. 10 150 11 151 To adjust the direct medical costs of cancer treatment in El Salvador to the setting of Nicaragua, we 12 13 152 multiplied by the ratio of WHO‐CHOICE inpatient bed‐day costs at a teaching hospital in Nicaragua 14 153 relative to El Salvador. We assumed each patient and a support person spent the same amount of time 15 154 traveling, waiting Forfor, and receiving peer care as in Elreview Salvador, and valued only this time at the 2013 minimum 16 155 wage rate in Nicaragua (inflated and converted to 2015 US$). To adjust temporary housing and 17 156 transportation costs to the setting of Nicaragua, we multiplied the Salvadoran costs by the ratio of GDP 18 157 per capita in Nicaragua relative to El Salvador. Costs are reported in Table A.10. 19 20 158 21 159 SUPPLEMENTARY RESULTS 22 160 Additional results described in the main manuscript are presented in Figures A3 to A6. 23 161 24 162 25 163 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 164 Table A.1. Age‐specific prevalence of high‐risk HPV, Nicaragua [15].a 4 5 Age group Number of women Number of women Prevalence (95% CI) 6 with high‐risk HPV 7 30 – 34 years 1,693 310 0.18 (0.17, 0.20) 8 35 – 39 years 1,141 184 0.16 (0.14, 0.18) 9 40 – 44 years 933 125 0.13 (0.11, 0.16) 10 45 – 49 years 878 121 0.14 (0.12, 0.16) 11 a 12 165 HPV positivity was based on a cut‐off of 0.5 relative light units. 13 14 166 15 167 Table A.2. Age‐specificFor cervical peer cancer incidence, review Nicaragua (GLOBOCAN only 2012) [11]. 16 Age group Cases Rate per 100,000 women (95% 17 CI) 18 40 – 44 years 123 78.7 (64.8, 92.6) 19 20 45 – 49 years 112 85.4 (69.6, 101.2) 21 50 – 54 years 102 88.4 (71.2, 105.6) 22 55 – 59 years 85 88.1 (69.4, 106.8) 23 60 – 64 years 51 84.0 (61.0, 107.1) 24 65 – 69 years 37 80.8 (54.8, 106.8) 25 70 – 74 years 30 74.6 (47.9, 101.3) 26 27 ≥75 years 45 70.3 (49.8, 90.8) 28 168 29 169 30 170 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 171 Table A.3. Monthly transition probabilities in the calibrated natural history model of HPV infection 4 5 172 and cervical carcinogenesis. 6 7 HPV Infection 8 HPV Type and Age Range of Multiplier Values Among Top 9 Baseline Value a 10 Group, years 50 Parameter Sets 11 Min Max Min Max 12 HPV 16 13 <21 0 0.00186 3.15278 7.90861 14 21‐24 0.000903 0.00125 0.670187 6.75291 15 For peer review only 16 25‐29 0.000782 0.00087 0.604518 7.39154 17 30‐49 0.000602 0.00078 0.677505 7.86645 18 >= 50 0.000217 0.00059 0.363714 8.9736 19 20 21 22 HPV 18 23 <21 0 0.00116 0.365305 9.94514 24 25 21‐24 0.0011 0.001173 0.304444 8.18583 26 25‐29 0.0006 0.001 0.305103 8.23571 27 30‐49 0.000301 0.00058 0.336658 9.81842 28 >= 50 0.000109 0.000295 0.347223 9.86684 29 30 31 HPV 31 32 <21 0 0.001428 0.432486 7.54208

33 http://bmjopen.bmj.com/ 34 21‐24 0.0009 0.001364 0.332414 5.86815 35 25‐29 0.0006 0.000805 0.312983 6.63163 36 30‐49 0.000301 0.00055 0.381602 7.86096 37 >= 50 0.000109 0.000295 0.445801 7.91394 38 39 40 HPV 33 41 <21

0 0.000786 0.337263 5.98918 on September 26, 2021 by guest. Protected copyright. 42 21‐24 0.000412 0.00069 0.321392 5.21972 43 44 25‐29 0.00025 0.00036 0.321163 4.52998 45 30‐49 0.000131 0.00022 0.315971 5.91849 46 >= 50 0.000054 0.000128 0.341468 5.73381 47 48 49 HPV 45 50 <21 0 0.001 0.838185 7.90205 51 21‐24 0.00041 0.0008 0.303553 6.97879 52 53 25‐29 0.000315 0.00037 0.373551 7.22544 54 30‐49 0.000161 0.000305 0.320699 7.65339 55 >= 50 0.000054 0.00015 0.313646 7.97911 56 57 58 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 HPV 52 4 5 <21 0 0.001186 0.339479 5.92479 6 21‐24 0.00027 0.0008 0.328541 4.11892 7 25‐29 0.000184 0.000254 0.301478 4.73227 8 9 30‐49 0.00015 0.000173 0.365365 5.61621 10 >= 50 0.000054 0.000148 0.333456 11.406 11 12 HPV 58 13 14 <21 0 0.001189 0.586648 7.78762 15 21‐24 For0.00059 peer review0.0011 only0.30793 7.02402 16 25‐29 0.000465 0.00055 0.376855 6.55248 17 18 30‐49 0.000244 0.00045 0.300592 7.20446 19 >= 50 0.000054 0.000239 0.42182 7.99349 20 21 Other Oncogenic Types 22 23 <21 0 0.00247 0.324069 5.82435 24 21‐24 0.001804 0.0023 0.301534 5.30746 25 25‐29 0.0012 0.0017 0.311474 4.99753 26 27 30‐49 0.000602 0.0011 0.318774 5.86776 28 >= 50 0.000217 0.00059 0.313403 5.86636 29 30 Nononcogenic Types 31 32 <21 0 0.00261 0.361451 14.721

33 21‐24 0.00255 0.00262 0.327572 10.6641 http://bmjopen.bmj.com/ 34 25‐29 0.002 0.0025 0.327932 13.399 35 36 30‐49 0.00085 0.00186 0.326549 14.4844 37 >= 50 0.000217 0.0008 0.324229 14.05 38 39 40 HPV clearance 41 HPV Type and time since Multiplier Value Among Top 50 Baseline Value on September 26, 2021 by guest. Protected copyright. 42 infection, months Parameter Setsa 43 HPV 16 44 45 1‐15 0.041886 1 46 16‐27 0.040754 1 47 28‐39 0.033905 1 48 40‐51 0.031888 1 49 50 52+ 0.019846 1 51 52 HPV 18 53 54 1‐15 0.073342 1 55 16‐27 0.063235 1 56 28‐39 0.053605 1 57 40‐51 0.020616 1 58 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 52+ 0.020616 1 4 5 6 HPV 31 7 1‐15 0.063447 1 8 9 16‐27 0.033826 1 10 28‐39 0.033826 1 11 40‐51 0.033826 1 12 52+ 0.033826 1 13 14 15 HPV 33 For peer review only 16 1‐15 0.083452 1 17 18 16‐27 0.044955 1 19 28‐39 0.036156 1 20 40‐51 0.036156 1 21 52+ 0.036156 1 22 23 24 HPV 45 25 1‐15 0.078517 1 26 27 16‐27 0.042579 1 28 28‐39 0.041675 1 29 40‐51 0.030133 1 30 52+ 0.030133 1 31 32

33 HPV 52 http://bmjopen.bmj.com/ 34 1‐15 0.062999 1 35 36 16‐27 0.044401 1 37 28‐39 0.044401 1 38 40‐51 0.039325 1 39 52+ 0.039325 1 40 41 on September 26, 2021 by guest. Protected copyright. 42 HPV 58 43 1‐15 0.065572 1 44 45 16‐27 0.05443 1 46 28‐39 0.053968 1 47 40‐51 0.033319 1 48 52+ 0.033319 1 49 50 51 High Risk HPV 52 1‐15 0.080766 1 53 54 16‐27 0.066633 1 55 28‐39 0.053972 1 56 40‐51 0.049229 1 57 52+ 0.005094 1 58 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 Low Risk HPV 6 1‐15 0.051888 1 7 16‐27 0.050005 1 8 9 28‐39 0.034649 1 10 40‐51 0.034649 1 11 52‐63 0.034649 1 12 64+ 0.028608 1 13 14 15 HPV ProgressionFor to CIN2 peer review only 16 HPV Type and time since Multiplier Value Among Top 50 17 Baseline Value a 18 infection, months Parameter Sets 19 HPV 16 20 1‐15 0.001707 1 21 16‐27 0.002422 1 22 23 28‐39 0.002577 1 24 40‐51 0.005518 1 25 52+ 0.014998 1 26 27 28 HPV 18 29 1‐15 4.27E‐05 1 30 16‐27 0.000189 1 31 32 28‐39 0.000189 1

33 40‐51 0.007733 1 http://bmjopen.bmj.com/ 34 52+ 0.007733 1 35 36 37 HPV 31 38 1‐15 0.000262 1 39 16‐27 0.00278 1 40 41 28‐39 0.003091 1 on September 26, 2021 by guest. Protected copyright. 42 40‐51 0.006931 1 43 52+ 0.006931 1 44 45 46 HPV 33 47 1‐15 0.000719 1 48 16‐27 0.000719 1 49 50 28‐39 0.004939 1 51 40‐51 0.004939 1 52 52+ 0.004939 1 53 54 55 HPV 45 56 1‐15 0 1 57 16‐27 0 1 58 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 28‐39 0.002258 1 4 5 40‐51 0.005333 1 6 52+ 0.005333 1 7 8 9 HPV 52 10 1‐15 0.000884 1 11 16‐27 0.001681 1 12 28‐39 0.001681 1 13 14 40‐51 0.001978 1 15 52+ For peer review0.005678 only 1 16 17 18 HPV 58 19 1‐15 0.000587 1 20 16‐27 0.002468 1 21 28‐39 0.002468 1 22 23 40‐51 0.004617 1 24 52+ 0.010246 1 25 26 27 High Risk HPV 28 1‐15 0.000126 1 29 16‐27 0.000373 1 30 28‐39 0.00196 1 31 32 40‐51 0.00196 1

33 52+ 0.00196 1 http://bmjopen.bmj.com/ 34 35 36 Low Risk HPV 37 1‐15 0.000205 1 38 16‐27 0.000291 1 39 28‐39 0.00031 1 40 41 40‐51 0.000663 1 on September 26, 2021 by guest. Protected copyright. 42 52+ 0.000663 1 43 44 45 HPV progression to CIN3 46 HPV Type and time since Multiplier Value Among Top 50 Baseline Value 47 infection, months Parameter Setsa 48 HPV 16 49 50 1‐15 0.000569 1 51 16‐27 0.000808 1 52 28‐39 0.00086 1 53 54 40‐51 0.001843 1 55 52+ 0.005024 1 56 57 HPV 18 58 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1‐15 1.07E‐05 1 4 5 16‐27 4.72E‐05 1 6 28‐39 4.72E‐05 1 7 40‐51 0.001939 1 8 9 52+ 0.001939 1 10 11 HPV 31 12 1‐15 6.55E‐05 1 13 14 16‐27 0.000696 1 15 28‐39 For peer review0.000774 only 1 16 40‐51 0.001737 1 17 18 52+ 0.001737 1 19 20 HPV 33 21 1‐15 0.00018 1 22 23 16‐27 0.00018 1 24 28‐39 0.001237 1 25 40‐51 0.001237 1 26 27 52+ 0.001237 1 28 29 HPV 45 30 1‐15 0 1 31 32 16‐27 0 1

33 28‐39 0.000565 1 http://bmjopen.bmj.com/ 34 40‐51 0.001336 1 35 36 52+ 0.001336 1 37 38 HPV 52 39 1‐15 0.000221 1 40 41 16‐27 0.000421 1 on September 26, 2021 by guest. Protected copyright. 42 28‐39 0.000421 1 43 40‐51 0.000495 1 44 45 52+ 0.001423 1 46 47 HPV 58 48 1‐15 0.000147 1 49 50 16‐27 0.000618 1 51 28‐39 0.000618 1 52 40‐51 0.001156 1 53 54 52+ 0.002571 1 55 56 High Risk HPV 57 1‐15 3.16E‐05 1 58 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 16‐27 9.32E‐05 1 4 5 28‐39 0.00049 1 6 40‐51 0.00049 1 7 52+ 0.00049 1 8 9 10 Low Risk HPV 11 1‐15 2.28E‐05 1 12 16‐27 3.23E‐05 1 13 14 28‐39 3.44E‐05 1 15 40‐51 For peer review7.37E‐05 only 1 16 52+ 7.37E‐05 1 17 18 19 Regression of CIN2 to NL 20 HPV Type and 21 Range of Multiplier Values Among Top time since Baseline Value 22 50 Parameter Setsa 23 infection, years 24 Min Max Min Max 25 HPV 16 26 1‐5 27 0.05 0.523578 0.500863 0.687389 28 6‐10 0.035 0.523578 0.500863 0.687389 29 11‐20 0.013 0.523578 0.500863 0.687389 30 21‐29 0.0005 0.523578 0.500863 0.687389 31 32 30‐39 0.0001 0.523578 0.500863 0.687389

33 40+ 0.00005 0.523578 0.500863 0.687389 http://bmjopen.bmj.com/ 34 35 All Other HPV 36 37 Types 38 1‐5 0.05 0.523578 0.527319 3.70898 39 6‐10 0.035 0.523578 0.527319 3.70898 40 11‐20 0.013 0.523578 0.527319 3.70898 41 on September 26, 2021 by guest. Protected copyright. 42 21‐29 0.0005 0.523578 0.527319 3.70898 43 30‐39 0.0001 0.523578 0.527319 3.70898 44 40+ 0.00005 0.523578 0.527319 3.70898 45 46 47 Regression of CIN3 to NL 48 HPV Type and 49 Range of Multiplier Values Among Top time since Baseline Value 50 50 Parameter Setsa 51 infection, years 52 Min Max Min Max 53 HPV 16 54 55 1‐5 0.025 0.639221 0.50664 1.67845 56 6‐10 0.0175 0.639221 0.50664 1.67845 57 11‐20 0.0065 0.639221 0.50664 1.67845 58 59 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 21‐29 0.00025 0.639221 0.50664 1.67845 4 5 30‐39 0.00005 0.639221 0.50664 1.67845 6 40+ 0.000025 0.639221 0.50664 1.67845 7 8 9 All Other HPV 10 Types 11 1‐5 0.025 0.639221 0.577736 3.9781 12 6‐10 0.0175 0.639221 0.577736 3.9781 13 11‐20 14 0.0065 0.639221 0.577736 3.9781 15 21‐29 For0.00025 peer review0.639221 only0.577736 3.9781 16 30‐39 0.00005 0.639221 0.577736 3.9781 17 40+ 0.000025 0.639221 0.577736 3.9781 18 19 20 Progression of CIN2 to CA 21 HPV Type and time Range of Multiplier Values Among Top 50 22 Baseline Value a 23 since infection, years Parameter Sets 24 Min Max 25 HPV16 26 1‐5 3.29E‐05 1.00901 1.49572 27 28 6‐10 3.56E‐05 1.00901 1.49572 29 11‐20 0.000857 1.00901 1.49572 30 21‐29 0.002506 1.00901 1.49572 31 32 30‐34 0.005011 1.00901 1.49572 35‐39 33 0.005429 1.00901 1.49572 http://bmjopen.bmj.com/ 34 40‐44 0.011693 1.00901 1.49572 35 45‐49 0.012528 1.00901 1.49572 36 37 50+ 0.33408 1.00901 1.49572 38 39 HPV 18 40 41 1‐5 3.29E‐05 1.00901 1.49572 on September 26, 2021 by guest. Protected copyright. 42 6‐10 3.56E‐05 1.00901 1.49572 43 11‐20 0.000571 1.00901 1.49572 44 21‐29 0.002506 1.00901 1.49572 45 46 30‐34 0.005011 1.00901 1.49572 47 35‐39 0.005429 1.00901 1.49572 48 40‐44 0.011693 1.00901 1.49572 49 50 45‐49 0.012528 1.00901 1.49572 51 50+ 0.33408 1.00901 1.49572 52 53 HPV 31 54 55 1‐5 2.2E‐05 1.00901 1.49572 56 6‐10 2.38E‐05 1.00901 1.49572 57 11‐20 0.000857 1.00901 1.49572 58 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 54 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 21‐29 0.00167 1.00901 1.49572 4 5 30‐34 0.003341 1.00901 1.49572 6 35‐39 0.003619 1.00901 1.49572 7 40‐44 0.007795 1.00901 1.49572 8 9 45‐49 0.008352 1.00901 1.49572 10 50+ 0.22272 1.00901 1.49572 11 12 HPV 33 13 14 1‐5 3.29E‐05 1.00901 1.49572 15 6‐10 For peer3.56E‐05 review1.00901 only 1.49572 16 11‐20 0.000571 1.00901 1.49572 17 18 21‐29 0.002506 1.00901 1.49572 19 30‐34 0.005011 1.00901 1.49572 20 35‐39 0.005429 1.00901 1.49572 21 40‐44 0.011693 1.00901 1.49572 22 23 45‐49 0.012528 1.00901 1.49572 24 50+ 0.33408 1.00901 1.49572 25 26 27 HPV 45 28 1‐5 2.2E‐05 1.00901 1.49572 29 6‐10 2.38E‐05 1.00901 1.49572 30 11‐20 0.000571 1.00901 1.49572 31 32 21‐29 0.00167 1.00901 1.49572

33 30‐34 0.003341 1.00901 1.49572 http://bmjopen.bmj.com/ 34 35‐39 0.003619 1.00901 1.49572 35 36 40‐44 0.007795 1.00901 1.49572 37 45‐49 0.008352 1.00901 1.49572 38 50+ 0.22272 1.00901 1.49572 39 40 41 HPV 52 on September 26, 2021 by guest. Protected copyright. 42 1‐5 2.2E‐05 1.00901 1.49572 43 6‐10 2.38E‐05 1.00901 1.49572 44 45 11‐20 0.000571 1.00901 1.49572 46 21‐29 0.00167 1.00901 1.49572 47 30‐34 0.003341 1.00901 1.49572 48 35‐39 0.003619 1.00901 1.49572 49 50 40‐44 0.007795 1.00901 1.49572 51 45‐49 0.008352 1.00901 1.49572 52 50+ 0.22272 1.00901 1.49572 53 54 55 HPV 58 56 1‐5 2.2E‐05 1.00901 1.49572 57 6‐10 2.38E‐05 1.00901 1.49572 58 59 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 55 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 11‐20 0 1.00901 1.49572 4 5 21‐29 0.00167 1.00901 1.49572 6 30‐34 0.003341 1.00901 1.49572 7 35‐39 0.003619 1.00901 1.49572 8 9 40‐44 0.007795 1.00901 1.49572 10 45‐49 0.008352 1.00901 1.49572 11 50+ 0.22272 1.00901 1.49572 12 13 14 High Risk HPV 15 1‐5 For peer2.2E‐05 review1.00901 only 1.49572 16 6‐10 2.38E‐05 1.00901 1.49572 17 18 11‐20 0.000571 1.00901 1.49572 19 21‐29 0.00167 1.00901 1.49572 20 30‐34 0.003341 1.00901 1.49572 21 35‐39 0.003619 1.00901 1.49572 22 23 40‐44 0.007795 1.00901 1.49572 24 45‐49 0.008352 1.00901 1.49572 25 50+ 0.22272 1.00901 1.49572 26 27 28 Progression of CIN3 to CA 29 HPV Type and time Range of Multiplier Values Among Top 50 30 Baseline Value since infection, years Parameter Setsa 31 32 Min Max

33 HPV 16 http://bmjopen.bmj.com/ 34 1‐5 35 0.000165 1.00199 1.48007 36 6‐10 0.000178 1.00199 1.48007 37 11‐20 0.004284 1.00199 1.48007 38 21‐29 0.012528 1.00199 1.48007 39 40 30‐34 0.025056 1.00199 1.48007 41 35‐39 0.027144 1.00199 1.48007 on September 26, 2021 by guest. Protected copyright. 42 40‐44 0.058464 1.00199 1.48007 43 45‐49 44 0.06264 1.00199 1.48007 45 50+ 0.33408 1.00199 1.48007 46 47 HPV 18 48 49 1‐5 0.000165 1.00199 1.48007 50 6‐10 0.000178 1.00199 1.48007 51 11‐20 0.004284 1.00199 1.48007 52 21‐29 53 0.012528 1.00199 1.48007 54 30‐34 0.025056 1.00199 1.48007 55 35‐39 0.027144 1.00199 1.48007 56 40‐44 0.058464 1.00199 1.48007 57 58 45‐49 0.06264 1.00199 1.48007 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 50+ 0.33408 1.00199 1.48007 4 5 6 HPV 31 7 1‐5 0.00011 1.00199 1.48007 8 9 6‐10 0.000119 1.00199 1.48007 10 11‐20 0.002856 1.00199 1.48007 11 21‐29 0.008352 1.00199 1.48007 12 30‐34 0.016704 1.00199 1.48007 13 14 35‐39 0.018096 1.00199 1.48007 15 40‐44 For peer0.038976 review1.00199 only 1.48007 16 45‐49 0.04176 1.00199 1.48007 17 18 50+ 0.22272 1.00199 1.48007 19 20 HPV 33 21 1‐5 0.000165 1.00199 1.48007 22 23 6‐10 0.000178 1.00199 1.48007 24 11‐20 0.004284 1.00199 1.48007 25 21‐29 0.012528 1.00199 1.48007 26 27 30‐34 0.025056 1.00199 1.48007 28 35‐39 0.027144 1.00199 1.48007 29 40‐44 0.058464 1.00199 1.48007 30 45‐49 0.06264 1.00199 1.48007 31 32 50+ 0.33408 1.00199 1.48007

33 http://bmjopen.bmj.com/ 34 HPV 45 35 36 1‐5 0.00011 1.00199 1.48007 37 6‐10 0.000119 1.00199 1.48007 38 11‐20 0.002856 1.00199 1.48007 39 21‐29 0.008352 1.00199 1.48007 40 41 30‐34 0.016704 1.00199 1.48007 on September 26, 2021 by guest. Protected copyright. 42 35‐39 0.018096 1.00199 1.48007 43 40‐44 0.038976 1.00199 1.48007 44 45 45‐49 0.04176 1.00199 1.48007 46 50+ 0.22272 1.00199 1.48007 47 48 HPV 52 49 50 1‐5 0.00011 1.00199 1.48007 51 6‐10 0.000119 1.00199 1.48007 52 11‐20 0.002856 1.00199 1.48007 53 54 21‐29 0.008352 1.00199 1.48007 55 30‐34 0.016704 1.00199 1.48007 56 35‐39 0.018096 1.00199 1.48007 57 40‐44 0.038976 1.00199 1.48007 58 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 57 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 45‐49 0.04176 1.00199 1.48007 4 5 50+ 0.22272 1.00199 1.48007 6 7 HPV 58 8 9 1‐5 0.00011 1.00199 1.48007 10 6‐10 0.000119 1.00199 1.48007 11 11‐20 0.002856 1.00199 1.48007 12 21‐29 0.008352 1.00199 1.48007 13 14 30‐34 0.016704 1.00199 1.48007 15 35‐39 For peer0.018096 review1.00199 only 1.48007 16 40‐44 0.038976 1.00199 1.48007 17 18 45‐49 0.04176 1.00199 1.48007 19 50+ 0.22272 1.00199 1.48007 20 21 High Risk HPV 22 23 1‐5 0.00011 1.00199 1.48007 24 6‐10 0.000119 1.00199 1.48007 25 11‐20 0.002856 1.00199 1.48007 26 27 21‐29 0.008352 1.00199 1.48007 28 30‐34 0.016704 1.00199 1.48007 29 35‐39 0.018096 1.00199 1.48007 30 40‐44 0.038976 1.00199 1.48007 31 32 45‐49 0.04176 1.00199 1.48007

33 50+ 0.22272 1.00199 1.48007 http://bmjopen.bmj.com/ 34 35 36 Progression of invasive cancer stages 37 Baseline Value 38 39 40 Local to regional 0.02 41 Regional to distant 0.025 on September 26, 2021 by guest. Protected copyright. 42 43 44 Invasive cancer mortality 45 Local 0.005876 46 47 Regional 0.015137 48 Distant 0.065519 49 50 51 Probability of symptom detection 52 Local 0.0039 53 Regional 0.1333 54 55 Distant 0.1746 56 173 57 58 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 58 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 174 a Multiplier values were drawn from the identified plausible search ranges and applied to baseline input 4 5 175 probabilities; the values presented represent the multiplier values that yielded the 50 best fits to the 6 176 epidemiologic calibration targets. 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

‐

for and ‐ and

The

the (other) START

is type

c with

[16] UP UP on

UP START ‐ ‐

‐ sample clinic. the

staff (transport (transport Testing based

sample.

usual

which to (HPV);

Source (other) per the Up Up types

‐ ‐ START START based per START (Cytology/VIA) was

specimens),

the

collection

at e ‐

Rapid

inspection cost project UP

Scale Scale and

personnel); Scale fuel); samples vehicle

sample. place valued ‐ Up

transport, brush, project

minutes visual

cytology

7 Advance

and Up

average own take

‐ Scale

major

to and

VIA:

an to 5

her sealers,

Scale triage

department,

HPV Using 3.81 0.36 0.02 0.00 0.00 0.00 4.19 laboratory

transporting

from

the VIA dollars; plate

for

derived

each and in

assumed

‐ collects economy kit, departments, in

We fuel sample.

Technologies

ranged self States was

of fuel test

per

three

equivalent

transport the preparing

which facilities

fuel the United

woman

for samples was Screening

of a of

average

3.81 0.82 0.01 1.37 0.32 0.92 7.26

exception department.

Cytology

US$: the 9 ‐ UP: includes cost

each the

when

vaginal each sample

in samples

exception and

of from in using

Chinandega),

same with (2010); START

per

time

the lab

the

and

b staff time facilities project the

sample with

preparing

samples project (2015);

10 to Up supplies,

‐ collection

a staff per test

the to

clinical assumed of self 2.88 6.72 0.01 0.41 0.21 0.81 9 BMJ Open

spent

11.04

collection) ng Chontales, http://bmjopen.bmj.com/ Scale ‐ project, HPV project

cost dry.

US$).

we clinical

and cost distance

UP time (self ‐ the

of ‐

and fuel across

reduced the the

(2015

demonstration

(Carazo, Staff

transporti from testing,

START

samples

decant load,

for

average category due

the

averaged to

b demonstration is per drawn

project. Planning the the

(assuming

procedure cytology

Up cervical was in

from

was epartments

on September 26, 2021 by guest. Protected copyright. departments. of [12]. d towels per

and

test

responsible Scale ‐ collection) derived

‐

Program Scale ‐ and 3.81 6.72 0.01 0.41 0.21 0.81 modalities

drawn

11.96

HPV three costs samples three

testing

bundled

HPV transported we

Up:

department the

and of

was the

the For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml are

Nicaragua both

HPV ‐ collection

gloves,

(provider of in Scale ‐

each personnel

from

for time medical For

collection

samples

ter across for

For peer review onlythe li each

transport supplies types), provider staff

department, in

drawn

Direct per (3

costs

lab trip

supplies ‐ supplies. cost between

both

was

each all

wage

tips

price mber Prevention–Utility

lab in

these

papillomavirus; transported round

with

costs

the clinical laboratory estimates

gas

which medical

in Screening:

for of

used of of and

testing,

pipette

under

monthly Cancer

staff

staff supplies equipment human

acid.

testing A.4.

cost cost direct time averaged HPV

category

estimate number staff supplies equipment

LEEP: We To

Lab Lab Lab Total Table Cost Clinical Clinical a b c Clinical Drugs annual Rapid the purchase number

economic 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 61 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.6. Location of service delivery for screening, diagnosis, and treatment of precancerous lesions 4 a 5 and cancer. b 6 Procedure Location of services 7 HPV DNA test Primary facility 8 Cytology test Primary facility 9 VIA test Primary facility 10 Colposcopy/biopsy Referral facility 11 12 Cryotherapy Referral facility 13 LEEP Referral facility 14 Follow‐up visits (after cryotherapy or LEEP) Primary facility 15 Cancer treatmentFor peer reviewTertiary facility only 16 a HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection with 17 acetic acid. 18

19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 62 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Tabe A.7. Women’s time estimates.a 4 b 5 Visit Minutes Cost (2015 US$) Source c 6 Screening 7 Wait time 33 0.48 Scale‐Up 8 Procedure time 20 0.29 START‐UP 9 Transport time (round‐ 57 0.82 Scale‐Up 10 trip) 11

12 c 13 Results 14 Wait time 33 0.48 Scale‐Up 15 Procedure time For peer2 review0.03 only START‐UP 16 Transport time (round‐ 57 0.82 Scale‐Up 17 trip) 18

19 c 20 Cryotherapy 21 Wait time 120 1.75 Scale‐Up 22 Procedure time 35 0.51 START‐UP 23 Transport time (round‐ 213 3.10 Scale‐Up 24 trip) 25 26 27 Colposcopy/biopsy 28 Wait time 120 1.75 Scale‐Up 29 Procedure time 37 0.54 START‐UP 30 Transport time (round‐ 213 3.10 Scale‐Up 31 trip) 32 33 LEEP http://bmjopen.bmj.com/ 34 35 Wait time 120 1.75 Scale‐Up 36 Procedure time 25 0.36 START‐UP 37 Transport time (round‐ 213 3.10 Scale‐Up 38 trip) 39 a LEEP: loop electrosurgical excision procedure; Scale‐Up: Scale‐Up demonstration project (2015); START‐ 40 UP: Screening Technologies to Advance Rapid Testing for Cervical Cancer Prevention–Utility and 41 Program Planning demonstration project (2010); US$: United States dollars. on September 26, 2021 by guest. Protected copyright. 42 b 43 Women’s time was valued using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 44 US$. We converted this to an hourly wage rate assuming 171 hours of work per month (40 hours per 45 week). 46 c Wait and transport time costs were accrued for each visit to a health facility, but we did not double‐ 47 count wait and transport times when multiple procedures were performed in the same visit (e.g., VIA 48 triage of screen‐positive women and cryotherapy of VIA+ women accrued the wait time and transport 49 50 time costs associated with a single visit to a referral facility, but accrued the time costs of both 51 procedures.) 52 53 54 55 56 57 58 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 63 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.8. Women’s transportation costs.a 4 b 5 Facility Round‐trip transportation cost Source 6 (2015 US$) 7 Screening facility 0.41 Scale‐Up 8 Referral facility 2.81 Scale‐Up 9 a Scale‐Up: Scale‐Up demonstration project (2015); US$: United States dollar. 10 b Screening with cytology or HPV testing was assumed to take place at a screening facility (i.e., a primary 11 12 level health facility). Triage testing with visual inspection with acetic acid (VIA), cryotherapy, 13 colposcopy/biopsy, and loop electrosurgical excision procedure (LEEP) were assumed to take place at 14 referral facilities. 15 For peer review only 16 17 18

19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 64 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.9. Programmatic training costs for HPV‐based screening.a 4 d 5 Type of training Total cost of Total cost of personnel Per woman cost b c 6 supplies/materials (2015 time (2015 US$) (2015 US$) 7 US$) 8 Health care personnel 2,758 5,784 0.09 9 offering careHPV 10 Laboratory technicians 3,503 526 0.04 11 12 processing careHPV 13 assays 14 Community outreach 2,662 3,618 0.08 15 workers promotingFor and peer review only 16 offering self‐collection 17 Health care providers 8,103 16,069 1.51 18 offering VIA and 19 20 cryotherapy a 21 HPV: human papillomavirus; US$: United States dollars; VIA: visual inspection with acetic acid. 22 b The total cost of supplies, materials, transportation, and refreshments for all training sessions of each 23 type was converted from 2015 local currency units using the official exchange rate. 24 c The total cost of personnel time includes time costs for both the trainer and the trainees, valued using 25 average monthly salary rates of each type of personnel present (and the proportion of trainees of each 26 personnel type) and based on the length of the training session. 27 d 28 To derive an average health care personnel training cost per woman screened, we divided the total 29 training cost for health care personnel by 100,000, corresponding to the number of women screened 30 during the Scale‐Up project. To derive an average laboratory technician training cost per woman 31 screened, we divided the total gtrainin cost for laboratory technicians by 100,000. To derive an average 32 community outreach worker training cost per woman screened with self‐collection, we divided the total

33 http://bmjopen.bmj.com/ training cost for outreach workers by 80,000, corresponding to the number of women who participated 34 35 in self‐collection of HPV samples (due to lack of data on self‐collection in a community setting, in the 36 base case analysis we assumed self‐collection took place at the clinic; however, we assumed outreach 37 workers would be involved in mobilizing and educating women about self‐collection, and conservatively 38 included the cost of training so as to represent a known programmatic cost as self‐collection efforts shift 39 to also include community settings). To derive an average health care provider training cost per woman 40 receiving VIA or cryotherapy, we divided the total VIA/cryotherapy training cost for health care 41 on September 26, 2021 by guest. Protected copyright. 42 providers by 16,000, corresponding to the approximate number of women who tested HPV‐positive in 43 Scale‐Up and would thus require visual assessment and cryotherapy. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 65 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.10. Cancer treatment costs (2015 US$). 4 5 Cancer stage Local cancer Regional cancer Distant cancer a 6 Direct medical costs 943.62 917.83 917.83 7 Woman and support 345.77 639.14 639.14 8 person time 9 Transportation, temporary 196.78 389.50 389.50 10 housing, and meals 11 12 Total cancer treatment 1486 1946 1946 13 cost 14 a Direct medical costs are based on the premise that 50% of local cancer patients presented with FIGO 15 stage 1a, requiringFor radical hysterectomy, peer and 50%review of local cancer patients only presented with FIGO stage 16 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. We 17 assumed all women presenting with FIGO stage 2b or higher received cobalt therapy followed by 5 18 chemotherapy sessions followed by brachytherapy. 19 b 20 Patient and support person time costs value time using the minimum wage in Nicaragua, as described 21 in the text. 22 c Direct non‐medical costs include round‐trip transportation for visits to a cancer center and meals and 23 temporary housing during the course of treatment. 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 66 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 Prevalence of High-Risk HPV, Nicaragua 7 8 0.50 9 10 0.45 11 12 0.40 13 14 15 0.35 For peer review only 16 17 0.30 18 19 0.25 20 21 0.20 22 23 Prevalence of HR HPV 24 0.15 25 26 0.10 27 28 0.05 29 30 0.00 31 30-34 35-39 40-44 45-49 32 Age group

33 http://bmjopen.bmj.com/ 34 35 36 Figure A1. Selected model output from the top 50 input parameter sets compared with empirical data 37 (i.e., calibration targets) on age‐specific prevalence of high‐risk HPV in Nicaragua, based on a relative 38 light unit cut‐off value of 0.5 in the START‐UP studies [2, 15]. Bold lines represent the 95% confidence 39 40 intervals around the empirical data, and gray circles represent model output from each of the top 50 41 input parameter sets. on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 67 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 Cancer Incidence, Nicaragua 7 300 8 9 10 11 250 12 13 14 15 200 For peer review only 16 17 18 19 150 20 21 22 23 100 24 25 per 100,000 Women Incidence 26 50 27 28 29 30 0 31 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ 32 Age group

33 http://bmjopen.bmj.com/ 34 35 36 Figure A2. Selected model output from the top 50 input parameter sets compared with empirical data 37 (i.e., calibration targets) on age‐specific cancer incidence in Nicaragua [2, 11]. Bold lines represent the 38 95% confidence intervals around the empirical data, and gray circles represent model output from each 39 40 of the top 50 input parameter sets. 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 68 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 50% Screening coverage: 50% Screening coverage: 70% Screening coverage: 80% 5 6 45% 7 8 % 9 40% 10 cancer,

11 35% 12 13 30% cervical 14 of 15 For peer review only 25% 16 risk

17 18 20% lifetime

19 in 20 21 15% 22 23 10% 24 Reduction 25 5% 26 27 28 0% 29 Pap (3 yrs) HPV‐Cryo (5 yrs) HPV‐VIA (5 yrs) HPV‐Pap (5 yrs) 30 Screening strategy 31 32

33 http://bmjopen.bmj.com/ 34 Figure A3. Reduction in lifetime risk of cervical cancer, by screening coverage level. Bars indicate the 35 percent reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 36 years; careHPV with cryotherapy for HPV‐positive women every 5 years [HPV‐Cryo]; careHPV every 5 37 38 years with visual inspection with acetic acid [VIA] triage of HPV‐positive women [HPV‐VIA]; and careHPV 39 every 5 years with Pap triage of HPV‐positive women [HPV‐Pap]) as screening coverage of the target 40 population increases. 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 28

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 69 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 Figure A4. Cost‐effectiveness analysis: Payer cost only. The graph displays the discounted lifetime costs 36 (x‐axis; in 2015 US$) and life expectancy (y‐axis) associated with each screening strategy (Pap testing 37 every 3 years; careHPV with cryotherapy for HPV‐positive women every 5 years [HPV‐Cryo]; careHPV 38 every 5 years with visual inspection with acetic acid [VIA] triage of HPV‐positive women [HPV‐VIA]; and 39 40 careHPV every 5 years with Pap triage of HPV‐positive women [HPV‐Pap]), when only costs to the payer 41

are included (i.e., women’s time and transportation costs are excluded). The cost‐effectiveness on September 26, 2021 by guest. Protected copyright. 42 associated with a change from one strategy to a more costly alternative is represented by the difference 43 44 in cost divided by the difference in life expectancy associated with the two strategies. The curve 45 indicates the strategies that are efficient because they are more effective and either 1) cost less; or 2) 46 have a more attractive cost‐effectiveness ratio than less effective options. The incremental cost‐ 47 48 effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two strategies under 49 comparison. In this sensitivity analysis focused on costs to the payer, HPV‐Cryo every 5 years was less 50 costly and more effective than other screening strategies considered, and was thus a dominant strategy 51 52 with an ICER of US$270 per year of life saved. 53 54 55 56 57 58 59 60 29

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 70 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 Figure A5. Cost‐effectiveness analysis: Direct medical cost of Pap testing, US$3. The graph displays the 36 discounted lifetime costs (x‐axis; in 2015 US$) and life expectancy (y‐axis) associated with each 37 screening strategy (Pap testing every 3 years; careHPV with cryotherapy for HPV‐positive women every 38 5 years [HPV‐Cryo]; careHPV every 5 years with visual inspection with acetic acid [VIA] triage of HPV‐ 39 40 positive women [HPV‐VIA]; and careHPV every 5 years with Pap triage of HPV‐positive women [HPV‐ 41

Pap]), when the direct medical cost of Pap testing was assumed to be as low as US$3 (base case: on September 26, 2021 by guest. Protected copyright. 42 US$7.26). The cost‐effectiveness associated with a change from one strategy to a more costly alternative 43 44 is represented by the difference in cost divided by the difference in life expectancy associated with the 45 two strategies. The curve indicates the strategies that are efficient because they are more effective and 46 either 1) cost less; or 2) have a more attractive cost‐effectiveness ratio than less effective options. The 47 48 incremental cost‐effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two 49 strategies under comparison. In this sensitivity analysis examining the impact of a lower Pap test cost, 50 HPV‐Cryo every 5 years was more effective and had a more attractive cost‐effectiveness ratio than Pap 51 52 testing every 3 years; furthermore, HPV‐Cryo was more effective and less costly than HPV‐VIA and HPV‐ 53 Pap, and was thus a dominant strategy with an ICER of US$320 per year of life saved. 54 55 56 57 58 59 60 30

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

“very 2000 VIA range ‐

demarcated

the HPV

sensitivity

US$)

and

effects

ratios, (2015

1500 represent case VIA

‐

interventions

mean

HPV

bars base

the

Ratio,

the by

effectiveness ‐ 1000 designates orange for

cost Effectiveness The

that divided

saved) Cost ‐

costs

life

500 of

available. threshold

Incremental mean Incremental

are year

the

per

of possible

a 0 US$ HPV ‐ Pap

ICER

d 0% 5% unavailable: 85% 75% visit visit visit visit Case

Only Only Case Case Case Case Case Case Case

Case Case Ca an the

2015

per per per per

Reduced Rate Base Base Base se Cost 0.40/0.85 0.90/0.50 0.70/0.65 % Screening Screening Base Base Base

Base Base Base Base

Base Costs to

85 40% 55% 70% Cryotherapy of of of of of of of of of of Test with

Cure ‐ axis, US$2,090,

Payer Colposcopy 75% 50% 75% women (x 35% women women

125% 125% 150% 150% Medical Routine Routine is 170%

HPV ‐ VIA,

BMJ Open sets, Receive

http://bmjopen.bmj.com/ 50% 80% HPV+ HPV+ HPV+ Direct assumed / /

in in in

Following (GDP) is Cryotherapy CIN2/3

alone,

Colposcopy, Screening/Triage

Rate Screen Screen presented Cryotherapy, treat of

Pap of

‐ Cryo parameter

Cure and ‐ are 50% and transportation, Cost ‐

Never Never product Positive confirmed

but Cost

HPV time

50% 20% input Screen Medical sensitivity/specificity wait Treatment (ICERs) sensitivity/specificity sensitivity/specificity

50 Medical

Following

when

VIA Pap Pap domestic on September 26, 2021 by guest. Protected copyright. Direct available Histologically the

Direct ratios women's

not

of gross

analysis Cryotherapy across Strategies: Cost for

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Pap ‐ Cryo capita

years

per HPV Eligibility effectiveness

For peer review only ‐

sensitivity Compliance Rate tion LEEP ng ment age

Cover Visit unt collec of Traini

Treat every cost

and ation Disco ‐ self Cost of r

al Cost Cance Popul g,

caragua’s

assuming

Testin Medic am of case

HPV Direct Progr Cost

HPV ‐ VIA

‐ axis),

Base line.

Cost for

Incremental

A6. Medic

black

‐ effective”. ICERs Direct a

Figure the strategy. cost by analyses Page 71 of 72 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 72 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 References 4 5 6 2 1. Campos, N.G., et al., An updated natural history model of cervical cancer: derivation of model 7 3 parameters. Am J Epidemiol, 2014. 180(5): p. 545‐55. 8 4 2. Campos, N.G., Tsu, V., Jeronimo, J., Mvundura, M., Lee, K., Kim, J.J., When and how often to 9 5 screen for cervical cancer in three low‐ and middle‐income countries: A cost‐effectiveness 10 6 analysis. Papillomavirus Research, 2015. 11 7 3. Munoz, N., et al., Incidence, duration, and determinants of cervical human papillomavirus 12 8 infection in a cohort of Colombian women with normal cytological results. J Infect Dis, 2004. 13 14 9 190(12): p. 2077‐87. 15 10 4. McCredie,For M.R., et al., peer Natural history reviewof cervical neoplasia andonly risk of invasive cancer in women 16 11 with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol, 2008. 9(5): 17 12 p. 425‐34. 18 13 5. Meyskens, F.L., Jr., et al., Enhancement of regression ofl cervica intraepithelial neoplasia II 19 14 (moderate dysplasia) with topically applied all‐trans‐retinoic acid: a randomized trial. J Natl 20 21 15 Cancer Inst, 1994. 86(7): p. 539‐43. 22 16 6. Keefe, K.A., et al., A randomized, double blind, Phase III trial using oral beta‐carotene 23 17 supplementation for women with high‐grade cervical intraepithelial neoplasia. Cancer Epidemiol 24 18 Biomarkers Prev, 2001. 10(10): p. 1029‐35. 25 19 7. Castle, P.E., et al., Evidence for frequent regression of cervical intraepithelial neoplasia‐grade 2. 26 20 Obstet Gynecol, 2009. 113(1): p. 18‐25. 27 21 8. Wang, S.M., et al., Six‐year regression and progression of cervical lesions of different human 28 29 22 papillomavirus viral loads in varied histological diagnoses. Int J Gynecol Cancer, 2013. 23(4): p. 30 23 716‐23. 31 24 9. Moscicki, A.B., et al., Rate of and risks for regression of cervical intraepithelial neoplasia 2 in 32 25 adolescents and young women. Obstet Gynecol, 2010. 116: (6) p. 1373‐80.

33 26 10. Herrero, R., et al., Rationale and design of a community‐based double‐blind randomized clinical http://bmjopen.bmj.com/ 34 27 trial of an HPV 16 and 18 vaccine in Guanacaste, Costa Rica. Vaccine, 2008. 26(37): p. 4795‐808. 35 36 28 11. Ferlay, J., et al., GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC 37 29 CancerBase No. 11 [Internet], 2013, International Agency for Research on Cancer: Lyon, France. 38 30 12. World Development Indicators, 2016, World Bank. 39 31 13. Database, I., Country Profiles: Nicaragua, 2016. 40 32 14. Campos, N.G., et al., The comparative and cost‐effectiveness of HPV‐based cervical cancer 41 33 screening algorithms in El Salvador. Int J Cancer, 2015. on September 26, 2021 by guest. Protected copyright. 42 34 15. Jeronimo, J., et al., A multicountry evaluation of careHPV testing, visual inspection with acetic 43 44 35 acid, and papanicolaou testing for the detection of cervical cancer. Int J Gynecol Cancer, 2014. 45 36 24(3): p. 576‐85. 46 37 16. Energy, U.S.D.o. Average Fuel Economy of Major Vehicle Categories. August 1, 2016]; Available 47 38 from: http://www.afdc.energy.gov/data/10310. 48 49 39 50 51 40 52 53 54 55 56 57 58 59 60 32

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

The Cost-Effectiveness of HPV-Based Cervical Cancer Screening in the Public Health System in Nicaragua

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2016-015048.R2

Article Type: Research

Date Submitted by the Author: 09-Mar-2017

Primary Subject Health economics Heading:

Global health, Health economics, Health policy, Obstetrics and Secondary Subject Heading: gynaecology, Public health

Keywords: HEALTH ECONOMICS, Epidemiology < ONCOLOGY, PUBLIC HEALTH http://bmjopen.bmj.com/

33 14 Word count (text): 4,527 http://bmjopen.bmj.com/ 34 35 36 15 37 38 16 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1

33 30 4) HPV testing every 5 years, with referral to Pap testing for HPV-positive women (“HPV-Pap”). http://bmjopen.bmj.com/ 34 35 31 Outcome measures: Reduction in lifetime risk of cancer; incremental cost-effectiveness ratios (ICER; 36 37 32 2015 US$ per year of life saved [YLS]). 38 39 40 33 Results: HPV-based screening strategies were more effective than Pap testing. HPV-Cryo was the least 41 on September 26, 2021 by guest. Protected copyright. 42 34 costly and most effective strategy, reducing lifetime cancer risk by 29.5% and outperforming HPV-VIA, 43 44 35 HPV-Pap, and Pap only, which reduced cancer risk by 19.4%, 12.2%, and 10.8%, respectively. With an 45 46 47 36 ICER of US$320/YLS, HPV-Cryo every 5 years would be very cost-effective using a threshold based on 48 49 37 Nicaragua’s per capita GDP of US$2,090. Findings were robust across sensitivity analyses on test 50 51 38 performance, coverage, compliance, and cost parameters. 52 53 54 55 56 57 58 59 60 2

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 43 Strengths and limitations of this study: 4 5 6 44 • Strengths of this study include the use of implementation data from the Scale-Up project to 7 8 45 estimate the cost-effectiveness of HPV testing in Nicaragua’s public health system. Findings 9 10 46 were robust across extensive sensitivity and scenario analyses. 11 12 13 47 • A limitation of this study is that modeled screening algorithms do not entirely reflect the 14 15 48 complex downstreamFor peermanagement describedreview by in-country guidelines.only While the modeled 16 17 49 algorithms reflect the prototypical structure of a screening episode and the type of facility at 18 19 20 50 which visits usually take place, these do not capture variation due to geography or health facility 21 22 51 capacity. 23 24 52 • An additional limitation is that while we adopted a micro-costing approach to leverage data 25 26 27 53 from implementation in Nicaragua, individual-level data for each woman were not available; 28 29 54 furthermore, we did not have cost data associated with HPV self-collection in community 30 31 55 settings, where most self-collection takes place. 32

33 56 http://bmjopen.bmj.com/ 34 35 36 57 Funding statement: This work was supported by the Bill & Melinda Gates Foundation. 37 38 39 58 Disclaimer: The findings and conclusions contained within are those of the authors and do not 40 41 on September 26, 2021 by guest. Protected copyright. 42 59 necessarily reflect positions or policies of the Bill & Melinda Gates Foundation. The funders had no role 43 44 60 in study design; data collection, analysis, and interpretation; preparation of the manuscript; or decision 45 46 61 to submit the article for publication. 47 48 49 62 Competing interests disclosed: All authors have completed the ICMJE uniform disclosure form at 50 51 52 63 www.icmje.org/coi_disclosure.pdf and declare financial support from the Bill and Melinda Gates 53 54 64 Foundation for the submitted work; JJ was the director of the START-UP demonstration projects and 55 56 57 58 59 60 4

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 20

33 http://bmjopen.bmj.com/ 535 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from Page 24 of 72

40% 85% 70% 24 24 0.41/0.94 Alternative sensitivity/specificity pair: 0.95/0.68 0.67/0.86 0.78/0.89 0.85/0.55 treatment treatment treatment CIN2: 85%CIN2: Base Case: BMJ Open http://bmjopen.bmj.com/

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

d For peer review only

c [14, 33, 35,[14, 38] 33, [14, 33, 34, 36, 37] b Test/Parameter

[6] [14, 17]

f e

d For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

For peer review only [30]

[43] d

[14] [30-32]

case) case) case) case) 918 - 2,920 - 918 (direct only;medical of 150% base 944 - 2,229 - 944 (direct only;medical of 150% base

918 390 640 944 197 346 1,946 1,486 26 26 BMJ Open http://bmjopen.bmj.com/

on September 26, 2021 by guest. Protected copyright. [29] For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

g g

h h

Direct non-medicalDirect Direct medicalDirect Women’s timeWomen’s timeWomen’s Direct medicalDirect Direct non-medicalDirect

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 29

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-STHPV-Cryo), which 36 requires 3 or more visits for screening and necessary treatment; c) HPV testing followed by visual inspection 37 with acetic acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive 38 women every 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. 39 ASCUS+: Pap result of atypical squamous cells of undetermined significance or worse; CIN: cervical 40 intraepithelial neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: 41 visual inspection with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Figure 1. Pathways of care, by screening strategy. Each diagram indicates the flow of screening-eligible

33 women (i.e., women aged 30 to 59 years) through each point of contact in a screening episode, conditional http://bmjopen.bmj.com/ 34 on visit compliance and test results, for a) Pap testing every 3 years (Pap), which requires 4 visits for 35 screening, diagnosis, and treatment; b) HPV screen-and-treat every 5 years (HPV-Cryo), which requires 3 or 36 more visits for screening and necessary treatment; c) HPV testing followed by visual inspection with acetic 37 acid (VIA) triage of HPV-positive women every 5 years (HPV-VIA), which requires 3 or more visits for screening and necessary treatment; and d) HPV testing followed by Pap triage of HPV-positive women every 38 5 years (HPV-Pap), which requires 5 or more visits for screening and necessary treatment. ASCUS+: Pap 39 result of atypical squamous cells of undetermined significance or worse; CIN: cervical intraepithelial 40 neoplasia; HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection 41 with acetic acid. on September 26, 2021 by guest. Protected copyright. 42 43 254x190mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 254x190mm (300 x 300 DPI) 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Supplemental Appendix: 4 5 The Cost-Effectiveness of HPV-Based Screening in the Public Health System in Nicaragua 6 7 8 MODEL CALIBRATION 9 Overview of the calibration process 10 Details of model development for the U.S. and Nicaragua models, including initial parameterization and 11 calibration, have been published elsewhere [1, 2]. Baseline HPV incidence rates, as a function of 12 13 genotype and age, were derived from a published prospective study of sexually active women aged 15- 14 85 years in Bogota, Colombia [3]. Because HPV incidence is not necessarily similar in Colombia and 15 Nicaragua due to Fordifferences inpeer sexual behavior review, we considered age -specificonly HPV incidence and natural 16 immunity following initial infection as candidate parameters for calibration. Type-specific data on CIN2 17 and CIN3 regression and progression are limited [4-9], so these highly uncertain parameters were also 18 candidates for calibration. Time-dependent rates of HPV clearance and progression by genotype were 19 informed by primary longitudinal data from the control arm of the Costa Rica Vaccine Trial [10], 20 21 assuming that rates of HPV clearance and progression to precancer are similar across populations and 22 vary only by time since infection and HPV genotype. 23 24 To calibrate the model, we selected empirical epidemiologic data from Nicaragua, including age-specific 25 prevalence of high-risk HPV and age-specific cancer incidence (see below for further detail). We then set 26 plausible search ranges around baseline model input values for age- and type-specific HPV incidence; 27 natural immunity following initial infection; and progression and regression of CIN. We performed 28 29 repeated model simulations in the absence of any preventive intervention (i.e., natural history 30 simulations). For each simulation, we randomly selected a single value for each of the uncertain 31 parameters from the identified search range, creating a unique vector of parameter values (i.e., 32 parameter “set”). From over 1,475,000 repeated samplings, we identified the parameter sets with the

33 highest correspondence to the empirical calibration target data from Nicaragua by calculating and http://bmjopen.bmj.com/ 34 aggregating the log-likelihood of model-projected outcomes. We used the 50 parameter sets with the 35 36 highest likelihood scores (i.e., best overall fit to the empirical data from Nicaragua) for analysis to 37 capture uncertainty in the model parameters as a form of probabilistic sensitivity analysis. We report 38 results as a mean of outcomes across these top 50 parameter sets; incremental cost-effectiveness ratios 39 are reported as the ratio of the mean costs divided by the mean effects of one strategy versus another 40 across sets. 41 on September 26, 2021 by guest. Protected copyright. 42 Calibration targets 43 44 We assessed model fit by comparing projected model outcomes of age-specific prevalence of high-risk 45 HPV and age-specific cancer incidence relative to empirical data from Nicaragua. The scoring algorithm 46 included age-specific prevalence of high-risk HPV and age-specific cervical cancer incidence. 47 48 Age-specific prevalence of high-risk HPV was drawn from the Screening Technologies to Advance Rapid 49 Testing for Cervical Cancer Prevention–Utility and Program Planning (START-UP) project data on 50 care 51 HPV positivity using a cut-off ratio cut-point of 0.5 relative light units (Table A.1). For each age 52 group, we derived a 95% binomial confidence interval around the point prevalence, which comprised 53 the calibration target. The likelihood function for each age group was assumed to follow a binomial 54 distribution. 55 56 Age-specific cancer incidence was drawn from Globocan due to the lack of a local cancer registry [11] 57 (Table A.2). The likelihood function for each age group was assumed to follow a normal distribution. 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3

4 5 Composite goodness-of-fit scores for each input parameter set were generated by summing the log 6 likelihood of each model outcome (i.e., age-specific HPV prevalence, age-specific cancer incidence). The 7 50 input parameter sets with the highest goodness-of-fit scores yielded the model outputs that were 8 simultaneously closest to all calibration targets, and these were selected for analysis. Figures A.1 and 9 A.2 display model fit to epidemiologic data on age-specific prevalence of high-risk HPV and age-specific 10 cancer incidence. 11 12 13 Transition probabilities for the calibrated natural history model are presented in Table A.3. 14 15 COST DATA For peer review only 16 17 Direct Medical Costs: Screening, Diagnosis, and Treatment of Precancerous Lesions 18 The direct medical costs of screening, diagnosis, and treatment of precancerous lesions were drawn 19 20 from the START-UP demonstration study in Masaya department and from the Scale-Up Nicaragua 21 Project in the departments of Carazo, Chontales, and Chinandega. Direct medical costs included clinical 22 staff time, clinical supplies, drugs, clinical equipment, laboratory staff time, laboratory supplies, and 23 laboratory equipment. Costs from the START-UP study were collected in 2010 local currency units and 24 inflated to year 2015 levels using Nicaragua GDP deflators and then converted to US$ using the official 25 exchange rate [12]; the exception was for equipment, which was generally procured in the United States 26 and was inflated to year 2015 levels using US GDP deflators. Costs from the Scale-Up Nicaragua project 27 28 (including equipment) were collected in 2015 local currency units and converted to US$ using the official 29 exchange rate [12]. 30 31 Costs are reported in Table A.4 (screening procedures) and Table A.5 (precancer diagnostic and 32 treatment procedures).

33 http://bmjopen.bmj.com/ 34 35 Women’s Time and Transportation Costs 36 To account for the opportunity cost of women’s time spent traveling to, waiting for, or receiving care, 37 we valued women’s time using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 US$. 38 While the minimum wage may be a conservative estimate for valuation of women’s time if they are 39 formally employed, it can serve as a proxy for the societal value of women’s time spent working in the 40 informal sector or at home. We converted this to an hourly wage rate assuming 171 hours of work per 41 month (40 hours per week). on September 26, 2021 by guest. Protected copyright. 42 43 44 Estimates for time spent traveling to, waiting for, and receiving care were dependent upon the facility 45 level where care was assumed to take place (Table A.6). While screening and delivery of primary 46 screening results were assumed to take place at basic primary health facilities, we assumed cryotherapy 47 and LEEP treatment were only provided at referral facilities. Estimates of women’s time spent receiving 48 a procedure were based on site-specific data from the START-UP demonstration projects, with staff time 49 spent on the procedure (excluding preparation and registration time, which we assumed were built into 50 51 patient waiting time) used as a proxy for women’s procedure time. Estimates of women’s wait time for 52 screening and delivery of primary screening results was based on the average wait time for screening at 53 facilities in the Chinandega department; wait time for cryotherapy, colposcopy, and LEEP was based on 54 the average wait time at referral centers in the Chontales department. Round-trip transportation time 55 was based on Scale-Up project staff estimates of the typical amount of time women spent traveling to 56 screening or referral facilities, using a sample of 9 facilities the Carazo department and 10 facilities in the 57 58 Chontales and Chinandega departments. We averaged across screening and referral facilities within 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 each department to obtain a department average, and then averaged across departments to obtain 4 5 average round-trip transportation times based on facility level. Estimates of women’s time and time 6 costs are presented in Table A.7. 7 8 Women’s transportation costs were based on Scale-Up project staff reports of the typical mode of 9 transportation to a particular facility by women in the catchment area, using a sample of 9 facilities in 10 Carazo and 10 facilities each in Chontales or Chinandega. We used the most commonly reported mode 11 12 of transportation for each department (for screening facilities: walking for Carazo; walking for 13 Chontales; bus for Chinandega; for referral facilities: bus in all three departments). The cost of a bus trip 14 was based on typical bus fare for the distance to the health facility, as reported by Scale-Up project staff. 15 We averaged the Fortypical transportation peer costs acrossreview departments to obtainonly an estimate of women’s 16 round-trip transportation costs to and from a health facility, assuming the same mode of transport was 17 used for the trip to and from the health facility. Women’s transportation costs are presented in Table 18 A.8. 19 20 21 Programmatic Costs 22 We included the programmatic cost of the following types of training sessions that were conducted 23 during the Scale-Up project: 1) for health care personnel to offer careHPV screening and counsel women 24 on results; 2) for laboratory technicians to run the careHPV assay; 3) for community outreach workers 25 and auxiliary nurses to promote careHPV screening and offer self-collection of HPV samples; and 4) for 26 health care providers to offer VIA and cryotherapy. We included the cost of trainer time and trainee 27 28 time (valued using average monthly salaries converted to an hourly rate; volunteer outreach worker 29 time was valued using the minimum wage), as well as the cost of supplies, materials, transportation, and 30 refreshments. To derive an average health care personnel training cost per woman screened, we divided 31 the total training cost for health care personnel by 100,000, corresponding to the number of women 32 screened during the Scale-Up project. To derive an average laboratory technician training cost per

33 woman screened, we divided the total training cost for laboratory technicians by 100,000. To derive an http://bmjopen.bmj.com/ 34 35 average community outreach worker training cost per woman screened, we divided the total training 36 cost for outreach workers by 80,000, corresponding to the number of women who participated in self- 37 collection of HPV samples. To derive an average health care provider training cost per woman receiving 38 VIA and cryotherapy, we divided the total VIA/cryotherapy training cost for health care providers by 39 16,000, corresponding to the approximate number of women who tested HPV-positive in Scale-Up and 40 would thus require visual assessment (to determine eligibility for cryotherapy) and, if eligible, 41 cryotherapy. on September 26, 2021 by guest. Protected copyright. 42 43 44 All women screened with HPV testing incurred the per-woman training costs for health care personnel 45 and laboratory technicians. Only women screened with HPV self-collection incurred the per-woman cost 46 of training outreach workers to offer self-collection (due to lack of data on self-collection in a 47 community setting, in the base case analysis we assumed self-collection took place at the clinic; 48 however, we assumed outreach workers would be involved in mobilizing and educating women about 49 self-collection, and conservatively included the cost of training so as to represent a known 50 51 programmatic cost as self-collection efforts shift to also include community settings). Only women who 52 received either VIA (as triage) or screen-and-treat cryotherapy incurred the per-woman cost of training 53 health providers to perform VIA and cryotherapy. Training costs are presented in Table A.9. 54 55 Cost of Cancer Care by Stage 56 Costs associated with cancer care by stage (Local versus Regional or Distant), including direct medical 57 58 costs, direct non-medical costs (i.e., transport to a cancer center, and meals and temporary housing 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 during the course of treatment), and patient and support person time were derived from a previous 4 5 analysis in El Salvador [14]. Based on expert opinion in El Salvador, we assumed 50% of local cancer 6 patients presented with FIGO stage 1a, requiring radical hysterectomy, and 50% of local cancer patients 7 presented with FIGO stage 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions 8 followed by brachytherapy. We assumed all women presenting with FIGO stage 2b or higher received 9 cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. 10 11 12 To adjust the direct medical costs of cancer treatment in El Salvador to the setting of Nicaragua, we 13 multiplied by the ratio of WHO-CHOICE inpatient bed-day costs at a teaching hospital in Nicaragua 14 relative to El Salvador. We assumed each patient and a support person spent the same amount of time 15 traveling, waiting Forfor, and receiving peer care as in Elreview Salvador, and valued only this time at the 2013 minimum 16 wage rate in Nicaragua (inflated and converted to 2015 US$). To adjust temporary housing and 17 transportation costs to the setting of Nicaragua, we multiplied the Salvadoran costs by the ratio of GDP 18 per capita in Nicaragua relative to El Salvador. Costs are reported in Table A.10. 19 20 21 SUPPLEMENTARY RESULTS 22 Additional results described in the main manuscript are presented in Figures A3 to A6. 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.1. Age-specific prevalence of high-risk HPV, Nicaragua [15].a 4 5 Age group Number of women Number of women Prevalence (95% CI) 6 with high-risk HPV 7 30 – 34 years 1,693 310 0.18 (0.17, 0.20) 8 35 – 39 years 1,141 184 0.16 (0.14, 0.18) 9 40 – 44 years 933 125 0.13 (0.11, 0.16) 10 45 – 49 years 878 121 0.14 (0.12, 0.16) 11 a 12 HPV positivity was based on a cut-off of 0.5 relative light units. 13 14 15 Table A.2. Age-specificFor cervical peer cancer incidence, review Nicaragua (GLOBOCAN only 2012) [11]. 16 Age group Cases Rate per 100,000 women (95% 17 CI) 18 19 40 – 44 years 123 78.7 (64.8, 92.6) 20 45 – 49 years 112 85.4 (69.6, 101.2) 21 50 – 54 years 102 88.4 (71.2, 105.6) 22 55 – 59 years 85 88.1 (69.4, 106.8) 23 60 – 64 years 51 84.0 (61.0, 107.1) 24 65 – 69 years 37 80.8 (54.8, 106.8) 25 70 – 74 years 30 74.6 (47.9, 101.3) 26 27 ≥75 years 45 70.3 (49.8, 90.8) 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 1 Table A.3. Monthly transition probabilities in the calibrated natural history model of HPV infection 4 5 2 and cervical carcinogenesis. 6 7 HPV Infection 8 HPV Type and Age Range of Multiplier Values Among Top 9 Baseline Value a 10 Group, years 50 Parameter Sets 11 Min Max Min Max 12 HPV 16 13 <21 0 0.00186 3.15278 7.90861 14 21-24 0.000903 0.00125 0.670187 6.75291 15 For peer review only 16 25-29 0.000782 0.00087 0.604518 7.39154 17 30-49 0.000602 0.00078 0.677505 7.86645 18 >= 50 0.000217 0.00059 0.363714 8.9736 19 20 21 22 HPV 18 23 <21 0 0.00116 0.365305 9.94514 24 25 21-24 0.0011 0.001173 0.304444 8.18583 26 25-29 0.0006 0.001 0.305103 8.23571 27 30-49 0.000301 0.00058 0.336658 9.81842 28 >= 50 29 0.000109 0.000295 0.347223 9.86684

30 31 HPV 31 32 <21 0 0.001428 0.432486 7.54208

33 http://bmjopen.bmj.com/ 34 21-24 0.0009 0.001364 0.332414 5.86815 35 25-29 0.0006 0.000805 0.312983 6.63163 36 30-49 0.000301 0.00055 0.381602 7.86096 37 >= 50 38 0.000109 0.000295 0.445801 7.91394 39 40 HPV 33 41 <21 0 0.000786 0.337263 5.98918 on September 26, 2021 by guest. Protected copyright. 42 21-24 0.000412 0.00069 0.321392 5.21972 43 44 25-29 0.00025 0.00036 0.321163 4.52998 45 30-49 0.000131 0.00022 0.315971 5.91849 46 >= 50 0.000054 0.000128 0.341468 5.73381 47

48 49 HPV 45 50 <21 0 0.001 0.838185 7.90205 51 21-24 0.00041 0.0008 0.303553 6.97879 52 53 25-29 0.000315 0.00037 0.373551 7.22544 54 30-49 0.000161 0.000305 0.320699 7.65339 55 >= 50 0.000054 0.00015 0.313646 7.97911 56

57 58 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 HPV 52 4 5 <21 0 0.001186 0.339479 5.92479 6 21-24 0.00027 0.0008 0.328541 4.11892 7 25-29 0.000184 0.000254 0.301478 4.73227 8 9 30-49 0.00015 0.000173 0.365365 5.61621 10 >= 50 0.000054 0.000148 0.333456 11.406 11 12 HPV 58 13 14 <21 0 0.001189 0.586648 7.78762 15 21-24 For0.00059 peer review0.0011 only0.30793 7.02402 16 25-29 0.000465 0.00055 0.376855 6.55248 17 18 30-49 0.000244 0.00045 0.300592 7.20446 19 >= 50 0.000054 0.000239 0.42182 7.99349 20 21 Other Oncogenic Types 22 23 <21 0 0.00247 0.324069 5.82435 24 21-24 0.001804 0.0023 0.301534 5.30746 25 25-29 0.0012 0.0017 0.311474 4.99753 26 27 30-49 0.000602 0.0011 0.318774 5.86776 28 >= 50 0.000217 0.00059 0.313403 5.86636 29 30 Nononcogenic Types 31 32 <21 0 0.00261 0.361451 14.721

33 21-24 0.00255 0.00262 0.327572 10.6641 http://bmjopen.bmj.com/ 34 25-29 0.002 0.0025 0.327932 13.399 35 36 30-49 0.00085 0.00186 0.326549 14.4844 37 >= 50 0.000217 0.0008 0.324229 14.05 38 39 40 HPV clearance 41 HPV Type and time since Multiplier Value Among Top 50 Baseline Value on September 26, 2021 by guest. Protected copyright. 42 infection, months Parameter Setsa 43 HPV 16 44 45 1-15 0.041886 1 46 16-27 0.040754 1 47 28-39 0.033905 1 48 40-51 49 0.031888 1 50 52+ 0.019846 1 51

52 HPV 18 53 54 1-15 0.073342 1 55 16-27 0.063235 1 56 28-39 0.053605 1 57 40-51 58 0.020616 1 59 60 7

5 6 HPV 31

7 1-15 0.063447 1 8 9 16-27 0.033826 1 10 28-39 0.033826 1 11 40-51 0.033826 1 12 52+ 13 0.033826 1

14 15 HPV 33 For peer review only 16 1-15 0.083452 1 17 18 16-27 0.044955 1 19 28-39 0.036156 1 20 40-51 0.036156 1 21 52+ 22 0.036156 1

23 24 HPV 45

25 1-15 0.078517 1 26 27 16-27 0.042579 1 28 28-39 0.041675 1 29 40-51 0.030133 1 30 52+ 31 0.030133 1

33 HPV 52 http://bmjopen.bmj.com/

34 1-15 0.062999 1 35 36 16-27 0.044401 1 37 28-39 0.044401 1 38 40-51 0.039325 1 39 52+ 40 0.039325 1 41 on September 26, 2021 by guest. Protected copyright. 42 HPV 58

43 1-15 0.065572 1 44 45 16-27 0.05443 1 46 28-39 0.053968 1 47 40-51 0.033319 1 48 52+ 49 0.033319 1

50 51 High Risk HPV

52 1-15 0.080766 1 53 54 16-27 0.066633 1 55 28-39 0.053972 1 56 40-51 0.049229 1 57 52+ 58 0.005094 1 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3

4 Low Risk HPV 5 6 1-15 0.051888 1 7 16-27 0.050005 1 8 9 28-39 0.034649 1 10 40-51 0.034649 1 11 52-63 0.034649 1 12 64+ 13 0.028608 1 14 15 HPV ProgressionFor to CIN2 peer review only 16 17 HPV Type and time since Multiplier Value Among Top 50 Baseline Value a 18 infection, months Parameter Sets 19 HPV 16 20 1-15 0.001707 1 21 16-27 22 0.002422 1 23 28-39 0.002577 1 24 40-51 0.005518 1 25 52+ 0.014998 1 26

27 28 HPV 18

29 1-15 4.27E-05 1 30 16-27 31 0.000189 1 32 28-39 0.000189 1

33 40-51 0.007733 1 http://bmjopen.bmj.com/ 34 52+ 0.007733 1 35

36 37 HPV 31

38 1-15 0.000262 1 39 16-27 40 0.00278 1 41 28-39 0.003091 1 on September 26, 2021 by guest. Protected copyright. 42 40-51 0.006931 1 43 52+ 0.006931 1 44

45 46 HPV 33

47 1-15 0.000719 1 48 16-27 49 0.000719 1 50 28-39 0.004939 1 51 40-51 0.004939 1 52 52+ 0.004939 1 53

54 55 HPV 45

56 1-15 0 1 57 16-27 58 0 1 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 28-39 4 0.002258 1 5 40-51 0.005333 1 6 52+ 0.005333 1 7 8 HPV 52 9 10 1-15 0.000884 1 11 16-27 0.001681 1 12 28-39 13 0.001681 1 14 40-51 0.001978 1 15 52+ For peer review0.005678 only 1 16 17 HPV 58 18 19 1-15 0.000587 1 20 16-27 0.002468 1 21 28-39 22 0.002468 1 23 40-51 0.004617 1 24 52+ 0.010246 1 25 26 High Risk HPV 27 28 1-15 0.000126 1 29 16-27 0.000373 1 30 28-39 31 0.00196 1 32 40-51 0.00196 1

33 52+ 0.00196 1 http://bmjopen.bmj.com/ 34 35 Low Risk HPV 36 37 1-15 0.000205 1 38 16-27 0.000291 1 39 28-39 40 0.00031 1 41 40-51 0.000663 1 on September 26, 2021 by guest. Protected copyright. 42 52+ 0.000663 1 43 44 45 HPV progression to CIN3 46 HPV Type and time since Multiplier Value Among Top 50 Baseline Value 47 infection, months Parameter Setsa 48 HPV 16 49 50 1-15 0.000569 1 51 16-27 0.000808 1 52 28-39 0.00086 1 53 54 40-51 0.001843 1 55 52+ 0.005024 1 56 57 HPV 18 58 59 60 10

11 HPV 31 12 1-15 13 6.55E-05 1 14 16-27 0.000696 1 15 28-39 For peer review0.000774 only 1 16 40-51 0.001737 1 17 18 52+ 0.001737 1 19

20 HPV 33 21 1-15 22 0.00018 1 23 16-27 0.00018 1 24 28-39 0.001237 1 25 40-51 0.001237 1 26 27 52+ 0.001237 1 28

29 HPV 45 30 1-15 31 0 1 32 16-27 0 1

33 28-39 0.000565 1 http://bmjopen.bmj.com/ 34 40-51 0.001336 1 35 36 52+ 0.001336 1 37

38 HPV 52 39 1-15 40 0.000221 1 41 16-27 0.000421 1 on September 26, 2021 by guest. Protected copyright. 42 28-39 0.000421 1 43 40-51 0.000495 1 44 45 52+ 0.001423 1 46

47 HPV 58 48 1-15 49 0.000147 1 50 16-27 0.000618 1 51 28-39 0.000618 1 52 40-51 0.001156 1 53 54 52+ 0.002571 1 55

56 High Risk HPV 57 1-15 58 3.16E-05 1 59 60 11

9 10 Low Risk HPV

11 1-15 2.28E-05 1 12 16-27 13 3.23E-05 1 14 28-39 3.44E-05 1 15 40-51 For peer review7.37E-05 only 1 16 52+ 7.37E-05 1 17 18 19 Regression of CIN2 to NL 20 HPV Type and 21 Range of Multiplier Values Among Top time since Baseline Value 22 50 Parameter Setsa 23 infection, years 24 Min Max Min Max 25 HPV 16 26 27 1-5 0.05 0.523578 0.500863 0.687389 28 6-10 0.035 0.523578 0.500863 0.687389 29 11-20 0.013 0.523578 0.500863 0.687389 30 21-29 0.0005 0.523578 0.500863 0.687389 31 32 30-39 0.0001 0.523578 0.500863 0.687389

33 40+ 0.00005 0.523578 0.500863 0.687389 http://bmjopen.bmj.com/ 34 35 36 All Other HPV 37 Types 38 1-5 0.05 0.523578 0.527319 3.70898 39 6-10 0.035 0.523578 0.527319 3.70898 40 11-20 0.013 0.523578 0.527319 3.70898 41 on September 26, 2021 by guest. Protected copyright. 42 21-29 0.0005 0.523578 0.527319 3.70898 43 30-39 0.0001 0.523578 0.527319 3.70898 44 40+ 0.00005 0.523578 0.527319 3.70898 45 46 47 Regression of CIN3 to NL 48 HPV Type and 49 Range of Multiplier Values Among Top time since Baseline Value 50 50 Parameter Setsa 51 infection, years 52 Min Max Min Max 53 HPV 16 54 55 1-5 0.025 0.639221 0.50664 1.67845 56 6-10 0.0175 0.639221 0.50664 1.67845 57 11-20 0.0065 0.639221 0.50664 1.67845 58 59 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 21-29 4 0.00025 0.639221 0.50664 1.67845 5 30-39 0.00005 0.639221 0.50664 1.67845 6 40+ 0.000025 0.639221 0.50664 1.67845 7 8 9 All Other HPV 10 Types 11 1-5 0.025 0.639221 0.577736 3.9781 12 6-10 0.0175 0.639221 0.577736 3.9781 13 14 11-20 0.0065 0.639221 0.577736 3.9781 15 21-29 For0.00025 peer review0.639221 only0.577736 3.9781 16 30-39 0.00005 0.639221 0.577736 3.9781 17 40+ 0.000025 0.639221 0.577736 3.9781 18 19 20 Progression of CIN2 to CA 21 HPV Type and time Range of Multiplier Values Among Top 50 22 Baseline Value a 23 since infection, years Parameter Sets 24 Min Max

25 HPV16 26 1-5 3.29E-05 1.00901 1.49572 27 28 6-10 3.56E-05 1.00901 1.49572 29 11-20 0.000857 1.00901 1.49572 30 21-29 0.002506 1.00901 1.49572 31 32 30-34 0.005011 1.00901 1.49572

33 35-39 0.005429 1.00901 1.49572 http://bmjopen.bmj.com/ 34 40-44 0.011693 1.00901 1.49572 35 45-49 0.012528 1.00901 1.49572 36 37 50+ 0.33408 1.00901 1.49572 38

39 HPV 18 40 41 1-5 3.29E-05 1.00901 1.49572 on September 26, 2021 by guest. Protected copyright. 42 6-10 3.56E-05 1.00901 1.49572 43 11-20 0.000571 1.00901 1.49572 44 21-29 0.002506 1.00901 1.49572 45 46 30-34 0.005011 1.00901 1.49572 47 35-39 0.005429 1.00901 1.49572 48 40-44 0.011693 1.00901 1.49572 49 50 45-49 0.012528 1.00901 1.49572 51 50+ 0.33408 1.00901 1.49572 52

53 HPV 31 54 55 1-5 2.2E-05 1.00901 1.49572 56 6-10 2.38E-05 1.00901 1.49572 57 11-20 0.000857 1.00901 1.49572 58 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 21-29 4 0.00167 1.00901 1.49572 5 30-34 0.003341 1.00901 1.49572 6 35-39 0.003619 1.00901 1.49572 7 40-44 0.007795 1.00901 1.49572 8 9 45-49 0.008352 1.00901 1.49572 10 50+ 0.22272 1.00901 1.49572 11 12 HPV 33 13 14 1-5 3.29E-05 1.00901 1.49572 15 6-10 For peer3.56E-05 review1.00901 only 1.49572 16 11-20 0.000571 1.00901 1.49572 17 18 21-29 0.002506 1.00901 1.49572 19 30-34 0.005011 1.00901 1.49572 20 35-39 0.005429 1.00901 1.49572 21 40-44 22 0.011693 1.00901 1.49572 23 45-49 0.012528 1.00901 1.49572 24 50+ 0.33408 1.00901 1.49572 25 26 HPV 45 27 28 1-5 2.2E-05 1.00901 1.49572 29 6-10 2.38E-05 1.00901 1.49572 30 11-20 31 0.000571 1.00901 1.49572 32 21-29 0.00167 1.00901 1.49572

33 30-34 0.003341 1.00901 1.49572 http://bmjopen.bmj.com/ 34 35-39 0.003619 1.00901 1.49572 35 36 40-44 0.007795 1.00901 1.49572 37 45-49 0.008352 1.00901 1.49572 38 50+ 0.22272 1.00901 1.49572 39

40 41 HPV 52 on September 26, 2021 by guest. Protected copyright. 42 1-5 2.2E-05 1.00901 1.49572 43 6-10 2.38E-05 1.00901 1.49572 44 45 11-20 0.000571 1.00901 1.49572 46 21-29 0.00167 1.00901 1.49572 47 30-34 0.003341 1.00901 1.49572 48 35-39 49 0.003619 1.00901 1.49572 50 40-44 0.007795 1.00901 1.49572 51 45-49 0.008352 1.00901 1.49572 52 50+ 0.22272 1.00901 1.49572 53

54 55 HPV 58

56 1-5 2.2E-05 1.00901 1.49572 57 6-10 58 2.38E-05 1.00901 1.49572 59 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 11-20 4 0 1.00901 1.49572 5 21-29 0.00167 1.00901 1.49572 6 30-34 0.003341 1.00901 1.49572 7 35-39 0.003619 1.00901 1.49572 8 9 40-44 0.007795 1.00901 1.49572 10 45-49 0.008352 1.00901 1.49572 11 50+ 0.22272 1.00901 1.49572 12

13 High Risk HPV 14 15 1-5 For peer2.2E-05 review1.00901 only 1.49572 16 6-10 2.38E-05 1.00901 1.49572 17 18 11-20 0.000571 1.00901 1.49572 19 21-29 0.00167 1.00901 1.49572 20 30-34 0.003341 1.00901 1.49572 21 35-39 22 0.003619 1.00901 1.49572 23 40-44 0.007795 1.00901 1.49572 24 45-49 0.008352 1.00901 1.49572 25 50+ 0.22272 1.00901 1.49572 26

27 28 Progression of CIN3 to CA 29 HPV Type and time Range of Multiplier Values Among Top 50 30 Baseline Value a 31 since infection, years Parameter Sets 32 Min Max

33 HPV 16 http://bmjopen.bmj.com/ 34 35 1-5 0.000165 1.00199 1.48007 36 6-10 0.000178 1.00199 1.48007 37 11-20 0.004284 1.00199 1.48007 38 21-29 0.012528 1.00199 1.48007 39 40 30-34 0.025056 1.00199 1.48007 41 35-39 0.027144 1.00199 1.48007 on September 26, 2021 by guest. Protected copyright. 42 40-44 0.058464 1.00199 1.48007 43 44 45-49 0.06264 1.00199 1.48007 45 50+ 0.33408 1.00199 1.48007 46

47 HPV 18 48 49 1-5 0.000165 1.00199 1.48007 50 6-10 0.000178 1.00199 1.48007 51 11-20 0.004284 1.00199 1.48007 52 53 21-29 0.012528 1.00199 1.48007 54 30-34 0.025056 1.00199 1.48007 55 35-39 0.027144 1.00199 1.48007 56 40-44 0.058464 1.00199 1.48007 57 58 45-49 0.06264 1.00199 1.48007 59 60 15

5 6 HPV 31

7 1-5 0.00011 1.00199 1.48007 8 9 6-10 0.000119 1.00199 1.48007 10 11-20 0.002856 1.00199 1.48007 11 21-29 0.008352 1.00199 1.48007 12 30-34 13 0.016704 1.00199 1.48007 14 35-39 0.018096 1.00199 1.48007 15 40-44 For peer0.038976 review1.00199 only 1.48007 16 45-49 0.04176 1.00199 1.48007 17 18 50+ 0.22272 1.00199 1.48007 19

20 HPV 33 21 1-5 22 0.000165 1.00199 1.48007 23 6-10 0.000178 1.00199 1.48007 24 11-20 0.004284 1.00199 1.48007 25 21-29 0.012528 1.00199 1.48007 26 27 30-34 0.025056 1.00199 1.48007 28 35-39 0.027144 1.00199 1.48007 29 40-44 0.058464 1.00199 1.48007 30 45-49 31 0.06264 1.00199 1.48007 32 50+ 0.33408 1.00199 1.48007

33 http://bmjopen.bmj.com/

34 HPV 45 35 36 1-5 0.00011 1.00199 1.48007 37 6-10 0.000119 1.00199 1.48007 38 11-20 0.002856 1.00199 1.48007 39 21-29 40 0.008352 1.00199 1.48007 41 30-34 0.016704 1.00199 1.48007 on September 26, 2021 by guest. Protected copyright. 42 35-39 0.018096 1.00199 1.48007 43 40-44 0.038976 1.00199 1.48007 44 45 45-49 0.04176 1.00199 1.48007 46 50+ 0.22272 1.00199 1.48007 47 48 HPV 52 49 50 1-5 0.00011 1.00199 1.48007 51 6-10 0.000119 1.00199 1.48007 52 11-20 0.002856 1.00199 1.48007 53 54 21-29 0.008352 1.00199 1.48007 55 30-34 0.016704 1.00199 1.48007 56 35-39 0.018096 1.00199 1.48007 57 40-44 58 0.038976 1.00199 1.48007 59 60 16

7 HPV 58 8 9 1-5 0.00011 1.00199 1.48007 10 6-10 0.000119 1.00199 1.48007 11 11-20 0.002856 1.00199 1.48007 12 21-29 13 0.008352 1.00199 1.48007 14 30-34 0.016704 1.00199 1.48007 15 35-39 For peer0.018096 review1.00199 only 1.48007 16 40-44 0.038976 1.00199 1.48007 17 18 45-49 0.04176 1.00199 1.48007 19 50+ 0.22272 1.00199 1.48007 20 21 High Risk HPV 22 23 1-5 0.00011 1.00199 1.48007 24 6-10 0.000119 1.00199 1.48007 25 11-20 0.002856 1.00199 1.48007 26 27 21-29 0.008352 1.00199 1.48007 28 30-34 0.016704 1.00199 1.48007 29 35-39 0.018096 1.00199 1.48007 30 40-44 31 0.038976 1.00199 1.48007 32 45-49 0.04176 1.00199 1.48007

43 44 Invasive cancer mortality 45 Local 46 0.005876 47 Regional 0.015137 48 Distant 0.065519 49 50 51 Probability of symptom detection 52 Local 0.0039 53 Regional 0.1333 54 55 Distant 0.1746 56 3 57 58 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 55 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 a Multiplier values were drawn from the identified plausible search ranges and applied to baseline input 4 5 5 probabilities; the values presented represent the multiplier values that yielded the 50 best fits to the 6 6 epidemiologic calibration targets. 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

BMJ Open Page 56 of 72

1 2 3 Table A.4. Screening: Direct medical costs per procedure (2015 US$).a 4 Cost category HPV test HPV test Cytology VIA triage Source 5 b b 6 (provider-collection) (self-collection) 7 Clinical staff 3.81 2.88 3.81 3.81 START-UP c 8 Clinical supplies 6.72 6.72 0.82 0.36 Scale-Up (HPV); START- 9 UP (Cytology/VIA) 10 Clinical equipment 0.01 0.01 0.01 0.02 11 Lab staff For0.41 peer review0.41 1.37 only 0.00 Scale-Up (transport 12 personnel);d START-UP 13 (other) 14 Lab supplies 0.81 0.81 0.92 0.00 Scale-Up (transport 15 fuel);e START-UP (other) 16 Lab equipment 0.21 0.21 0.32 0.00 START-UP 17 http://bmjopen.bmj.com/ 18 Total direct medical cost 11.96 11.04 7.26 4.19 19 a HPV: human papillomavirus; Scale-Up: Scale-Up demonstration project (2015); START-UP: Screening Technologies to Advance Rapid Testing for 20 21 Cervical Cancer Prevention–Utility and Program Planning demonstration project (2010); US$: United States dollars; VIA: visual inspection with 22 acetic acid. 23 b HPV testing with both provider-collection of cervical samples and self-collection of vaginal samples was assumed to take place at the clinic. The 24 difference in costs between collection modalities is due to reduced clinical staff time when a woman self-collects her own sample. 25 c The cost of clinical supplies for HPV testing was drawn from the Scale-Up project and includes the test kit, plate sealers, brush, collection 26 medium, pipette tips (3 types), gloves, and towels to decant and dry. on September 26, 2021 by guest. Protected copyright. 27 d 28 The cost of laboratory staff time was drawn from the START-UP project, with the exception of preparing and transporting samples to the 29 laboratory, which was drawn from the Scale-Up project. Staff time spent preparing samples for transport in the Scale-Up project was based on 30 project staff estimates in each of the three departments (Carazo, Chontales, and Chinandega), which ranged from 5 to 7 minutes per sample. 31 Staff time for round-trip transport of samples was averaged across 9 to 10 facilities in each of the three departments, and valued based on the 32 average monthly wage of the personnel responsible for transporting the samples in each department. We derived an average cost per sample 33 prepared and transported for each department (assuming the cost of staff time per sample was equivalent for HPV and cytology specimens), and 34 then averaged these costs across the three departments. 35 e 36 For HPV testing, all lab supplies are bundled in the category of clinical supplies, with the exception of fuel for laboratory transport, which is 37 included under lab supplies. For both HPV and cytology testing, we assumed the same cost of fuel per sample. Using Scale-Up project staff 38 estimates of the number of samples transported per load, the distance to the lab from 9 to 10 facilities in each department, and the usual type 39 of vehicle used in each department, we derived the average fuel cost per sample using the average fuel economy of major vehicle types [16] and 40 the average gas price per liter in Nicaragua [12]. 41 42 43 44 19 45

46 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

Page 57 of 72 BMJ Open

1 2 3

4 a 5 Table A.5. Diagnosis and Treatment of Precancer: Direct medical costs per procedure (2015 US$). 6 Cost category Colposcopy Colposcopy and biopsy Cryotherapy LEEP Source 7 Clinical staff 5.92 6.29 7.54 7.34 START-UP 8 Clinical supplies 0.38 1.39 3.54 36.05 Scale-Up (cryotherapy 9 gas);b START-UP (other) 10 Drugs 0.13 0.13 0.30 0.10 START-UP 11 Clinical equipment For0.49 peer review0.53 9.13 only 13.29 Scale-Up (cryotherapy 12 unit);c START-UP (other) 13 Lab staff 0.00 4.09 0.00 4.09 START-UP 14 Lab supplies 0.00 3.35 0.00 3.35 START-UP 15 Lab equipment 0.00 4.12 0.00 4.12 START-UP 16 17 Total direct medical cost 6.91 19.91 18.16 68.36 http://bmjopen.bmj.com/ 18 a LEEP: loop electrosurgical excision procedure; Scale-Up: Scale-Up demonstration project (2015); START-UP: Screening Technologies to Advance 19 20 Rapid Testing for Cervical Cancer Prevention–Utility and Program Planning demonstration project (2010); US$: United States dollars. b 21 We used estimates from Scale-Up project staff to inform the cost of gas used per cryotherapy procedure, based on the cost of 1 cylinder and 22 the number of women treated per tank. We did not include the cost of the tank deposit. 23 c To estimate the cost per procedure of a cryotherapy unit, we assumed this equipment item was a tradable good, and accordingly converted the 24 purchase price from local currency units to US$ using the official exchange rate. We annualized the cost with a 3% interest rate and assumed an 25 economic life of 5 years. To derive the number of procedures per machine per year across the three departments considered, we divided the 26 on September 26, 2021 by guest. Protected copyright. annual number of cryotherapy procedures by the number of cryotherapy units. We then divided the annualized cryotherapy unit cost by the 27 28 number of procedures per machine per year. 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 20 45

46 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 58 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.6. Location of service delivery for screening, diagnosis, and treatment of precancerous lesions 4 a 5 and cancer. 6 Procedure Location of services 7 HPV DNA test Screening facility 8 Cytology test Screening facility 9 VIA test Referral facility 10 Colposcopy/biopsy Referral facility 11 12 Cryotherapy Referral facility 13 LEEP Referral facility 14 Follow-up visits (after cryotherapy or LEEP) Primary facility 15 Cancer treatmentFor peer reviewTertiary facility only 16 a HPV: human papillomavirus; LEEP: loop electrosurgical excision procedure; VIA: visual inspection with 17 acetic acid. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 59 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Tabe A.7. Women’s time estimates.a 4 b 5 Visit Minutes Cost (2015 US$) Source 6 Screeningc 7 Wait time 33 0.48 Scale-Up 8 Procedure time 20 0.29 START-UP 9 Transport time (round- 57 0.82 Scale-Up 10 trip) 11 12 c 13 Results 14 Wait time 33 0.48 Scale-Up 15 Procedure time For peer2 review0.03 only START-UP 16 Transport time (round- 57 0.82 Scale-Up 17 trip) 18

19 c 20 Cryotherapy

21 Wait time 120 1.75 Scale-Up 22 Procedure time 35 0.51 START-UP 23 Transport time (round- 213 3.10 Scale-Up 24 trip) 25 26 27 Colposcopy/biopsy 28 Wait time 120 1.75 Scale-Up 29 Procedure time 37 0.54 START-UP 30 Transport time (round- 213 3.10 Scale-Up 31 trip) 32

33 http://bmjopen.bmj.com/ LEEP 34 35 Wait time 120 1.75 Scale-Up 36 Procedure time 25 0.36 START-UP 37 Transport time (round- 213 3.10 Scale-Up 38 trip) 39 a LEEP: loop electrosurgical excision procedure; Scale-Up: Scale-Up demonstration project (2015); START- 40 UP: Screening Technologies to Advance Rapid Testing for Cervical Cancer Prevention–Utility and 41 on September 26, 2021 by guest. Protected copyright. 42 Program Planning demonstration project (2010); US$: United States dollars. b 43 Women’s time was valued using Nicaragua’s 2013 monthly minimum wage [13], converted to 2015 44 US$. We converted this to an hourly wage rate assuming 171 hours of work per month (40 hours per 45 week). 46 c Wait and transport time costs were accrued for each visit to a health facility, but we did not double- 47 count wait and transport times when multiple procedures were performed in the same visit (e.g., VIA 48 triage of screen-positive women and cryotherapy of VIA+ women accrued the wait time and transport 49 50 time costs associated with a single visit to a referral facility, but accrued the time costs of both 51 procedures.) 52 53 54 55 56 57 58 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 60 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.8. Women’s transportation costs.a 4 b 5 Facility Round-trip transportation cost Source 6 (2015 US$) 7 Screening facility 0.41 Scale-Up 8 Referral facility 2.81 Scale-Up 9 a Scale-Up: Scale-Up demonstration project (2015); US$: United States dollar. 10 b Screening with cytology or HPV testing was assumed to take place at a screening facility (i.e., a primary 11 12 level health facility). Triage testing with visual inspection with acetic acid (VIA), cryotherapy, 13 colposcopy/biopsy, and loop electrosurgical excision procedure (LEEP) were assumed to take place at 14 referral facilities. 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 61 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.9. Programmatic training costs for HPV-based screening.a 4 d 5 Type of training Total cost of Total cost of personnel Per woman cost b c 6 supplies/materials (2015 time (2015 US$) (2015 US$) 7 US$) 8 Health care personnel 2,758 5,784 0.09 9 offering careHPV 10 Laboratory technicians 3,503 526 0.04 11 12 processing careHPV 13 assays 14 Community outreach 2,662 3,618 0.08 15 workers promotingFor and peer review only 16 offering self-collection 17 Health care providers 8,103 16,069 1.51 18 19 offering VIA and 20 cryotherapy 21 a HPV: human papillomavirus; US$: United States dollars; VIA: visual inspection with acetic acid. 22 b The total cost of supplies, materials, transportation, and refreshments for all training sessions of each 23 type was converted from 2015 local currency units using the official exchange rate. 24 c The total cost of personnel time includes time costs for both the trainer and the trainees, valued using 25 average monthly salary rates of each type of personnel present (and the proportion of trainees of each 26 27 personnel type) and based on the length of the training session. d 28 To derive an average health care personnel training cost per woman screened, we divided the total 29 training cost for health care personnel by 100,000, corresponding to the number of women screened 30 during the Scale-Up project. To derive an average laboratory technician training cost per woman 31 screened, we divided the total training cost for laboratory technicians by 100,000. To derive an average 32 community outreach worker training cost per woman screened with self-collection, we divided the total

33 http://bmjopen.bmj.com/ 34 training cost for outreach workers by 80,000, corresponding to the number of women who participated 35 in self-collection of HPV samples (due to lack of data on self-collection in a community setting, in the 36 base case analysis we assumed self-collection took place at the clinic; however, we assumed outreach 37 workers would be involved in mobilizing and educating women about self-collection, and conservatively 38 included the cost of training so as to represent a known programmatic cost as self-collection efforts shift 39 to also include community settings). To derive an average health care provider training cost per woman 40 receiving VIA or cryotherapy, we divided the total VIA/cryotherapy training cost for health care 41 on September 26, 2021 by guest. Protected copyright. 42 providers by 16,000, corresponding to the approximate number of women who tested HPV-positive in 43 Scale-Up and would thus require visual assessment and cryotherapy. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 62 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 Table A.10. Cancer treatment costs (2015 US$). 4 5 Cancer stage Local cancer Regional cancer Distant cancer 6 Direct medical costsa 943.62 917.83 917.83 7 Woman and support 345.77 639.14 639.14 8 person time 9 Transportation, temporary 196.78 389.50 389.50 10 housing, and meals 11 12 Total cancer treatment 1486 1946 1946 13 cost 14 a Direct medical costs are based on the premise that 50% of local cancer patients presented with FIGO 15 stage 1a, requiringFor radical hysterectomy, peer and 50%review of local cancer patients only presented with FIGO stage 16 1b/2a, requiring cobalt therapy followed by 5 chemotherapy sessions followed by brachytherapy. We 17 assumed all women presenting with FIGO stage 2b or higher received cobalt therapy followed by 5 18 19 chemotherapy sessions followed by brachytherapy. b 20 Patient and support person time costs value time using the minimum wage in Nicaragua, as described 21 in the text. 22 c Direct non-medical costs include round-trip transportation for visits to a cancer center and meals and 23 temporary housing during the course of treatment. 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 63 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3

4 5 6 7 Prevalence of High-Risk HPV, Nicaragua 8 0.50 9 10 0.45 11 12 13 0.40 14 15 0.35 For peer review only 16 17 0.30 18 19 0.25 20 21 0.20 22 23 Prevalence of HR HPV 24 0.15 25 26 0.10 27 28 0.05 29 30 0.00 31 30-34 35-39 40-44 45-49 32 Age group

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 64 of 72 BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3

4 5 6 7 Cancer Incidence, Nicaragua 8 300 9 10 11 250 12 13 14 15 200 For peer review only 16 17 18 19 150 20 21 22 23 100 24 25 Incidence per 100,000 Women 26 50 27 28 29 30 0 31 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ 32 Age group

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 65 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 4 50% Screening coverage: 50% Screening coverage: 70% Screening coverage: 80% 5 6 45% 7 8 9 40% 10 11 35% 12 13 30% 14 15 For peer review only 16 25% 17 18 20% 19 20 21 15% 22 23 10% 24 Reduction in lifetime risk of cervical cancer, % 25 5% 26 27 28 0% 29 Pap (3 yrs) HPV-Cryo (5 yrs) HPV-VIA (5 yrs) HPV-Pap (5 yrs) 30 Screening strategy 31 32

33 http://bmjopen.bmj.com/ 34 Figure A3. Reduction in lifetime risk of cervical cancer, by screening coverage level. Bars indicate the 35 percent reduction in lifetime risk of cervical cancer for each screening strategy (Pap testing every 3 36 years; careHPV with cryotherapy for HPV-positive women every 5 years [HPV-Cryo]; careHPV every 5 37 38 years with visual inspection with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and careHPV 39 every 5 years with Pap triage of HPV-positive women [HPV-Pap]) as screening coverage of the target 40 population increases. 41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 28

33 http://bmjopen.bmj.com/ 34 35 Figure A4. Cost-effectiveness analysis: Payer cost only. The graph displays the discounted lifetime costs 36 (x-axis; in 2015 US$) and life expectancy (y-axis) associated with each screening strategy (Pap testing 37 every 3 years; careHPV with cryotherapy for HPV-positive women every 5 years [HPV-Cryo]; careHPV 38 39 every 5 years with visual inspection with acetic acid [VIA] triage of HPV-positive women [HPV-VIA]; and 40 careHPV every 5 years with Pap triage of HPV-positive women [HPV-Pap]), when only costs to the payer 41 are included (i.e., women’s time and transportation costs are excluded). The cost-effectiveness on September 26, 2021 by guest. Protected copyright. 42 43 associated with a change from one strategy to a more costly alternative is represented by the difference 44 in cost divided by the difference in life expectancy associated with the two strategies. The curve 45 indicates the strategies that are efficient because they are more effective and either 1) cost less; or 2) 46 have a more attractive cost-effectiveness ratio than less effective options. The incremental cost- 47 48 effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two strategies under 49 comparison. In this sensitivity analysis focused on costs to the payer, HPV-Cryo every 5 years was less 50 costly and more effective than other screening strategies considered, and was thus a dominant strategy 51 52 with an ICER of US$270 per year of life saved. 53 54 55 56 57 58 59 60 29

33 http://bmjopen.bmj.com/ 34 35 Figure A5. Cost-effectiveness analysis: Direct medical cost of Pap testing, US$3. The graph displays the 36 discounted lifetime costs (x-axis; in 2015 US$) and life expectancy (y-axis) associated with each 37 screening strategy (Pap testing every 3 years; careHPV with cryotherapy for HPV-positive women every 38 39 5 years [HPV-Cryo]; careHPV every 5 years with visual inspection with acetic acid [VIA] triage of HPV- 40 positive women [HPV-VIA]; and careHPV every 5 years with Pap triage of HPV-positive women [HPV- 41 Pap]), when the direct medical cost of Pap testing was assumed to be as low as US$3 (base case: on September 26, 2021 by guest. Protected copyright. 42 43 US$7.26). The cost-effectiveness associated with a change from one strategy to a more costly alternative 44 is represented by the difference in cost divided by the difference in life expectancy associated with the 45 two strategies. The curve indicates the strategies that are efficient because they are more effective and 46 either 1) cost less; or 2) have a more attractive cost-effectiveness ratio than less effective options. The 47 48 incremental cost-effectiveness ratio (ICER) is the reciprocal of the slope of the line connecting the two 49 strategies under comparison. In this sensitivity analysis examining the impact of a lower Pap test cost, 50 HPV-Cryo every 5 years was more effective and had a more attractive cost-effectiveness ratio than Pap 51 52 testing every 3 years; furthermore, HPV-Cryo was more effective and less costly than HPV-VIA and HPV- 53 Pap, and was thus a dominant strategy with an ICER of US$320 per year of life saved. 54 55 56 57 58 59 60 30

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

BMJ Open Page 68 of 72

1 2 3 4 Base Case 5 50% Never Screen / 50% Routine Screening age ation Popul 6 Cover 20% Never Screen / 80% Routine Screening 7 40% per visit 8 55% per visit 9 Visit 70% per visit 10 Compliance 85% per visit 11 For peerScreen-and-treat Cryotherapyreview Cure Rate 75% only Pap sensitivity/specificity in HPV+ women 0.40/0.85 12 Pap sensitivity/specificity in HPV+ women 0.90/0.50 13 VIA sensitivity/specificity in HPV+ women 0.70/0.65 14 Treatment Cure Rate Following Colposcopy 85% 15 Eligibility for Cryotherapy Following Positive Screening/Triage Test Reduced 16 Pap Strategies: Histologically confirmed CIN2/3 Receive Cryotherapy 17 0% http://bmjopen.bmj.com/ unt Rate 18 Disco 5% 19 Payer Costs Only 75% of Base Case al 20 g, of tion HPV self- Cost collec Direct Testin 125% of Base Case 21 Medic 22 Direct Medical Cost of Cryotherapy, 170% of Base Case Direct Medical Cost of Colposcopy, 35% of Base Case 23 75% of Base Case al 24 of Cost LEEP

Direct 125% of Base Case Medic 25 50% of Base Case of ng am on September 26, 2021 by guest. Protected copyright. Cost

26 Progr Traini 150% of Base Case 27 Cost of women's wait time and transportation, 50% of Base Case 28 Direct Medical Cost Only r of Cost Treat ment 29 Cance 150% of Base Case 30 0 500 1000 1500 2000 31 Incremental Cost-Effectiveness Ratio, HPV-VIA (2015 US$) 32 33 34 Figure A6. Base case and sensitivity analysis when HPV-Cryo is assumed to be unavailable: Incremental cost-effectiveness ratios, HPV-VIA 35 strategy. Incremental cost-effectiveness ratios (ICERs) are presented (x-axis, 2015 US$ per year of life saved) for the base case and sensitivity 36 37 analyses (y-axis), assuming HPV-Cryo is not available but Pap alone, HPV-VIA, and HPV-Pap are available. The orange bars represent the range of 38 the ICERs for HPV-VIA every 5 years across the 50 input parameter sets, with the ICER of the mean costs divided by the mean effects demarcated 39 by a black line. Nicaragua’s per capita gross domestic product (GDP) is US$2,090, as a possible threshold that designates interventions as “very 40 cost-effective”. 41 42 43 44 31 45

46 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 69 of 72 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 1 2 3 References 4 5 6 1. Campos, N.G., et al., An updated natural history model of cervical cancer: derivation of model 7 parameters. Am J Epidemiol, 2014. 180(5): p. 545-55. 8 2. Campos, N.G., Tsu, V., Jeronimo, J., Mvundura, M., Lee, K., Kim, J.J., When and how often to 9 screen for cervical cancer in three low- and middle-income countries: A cost-effectiveness 10 analysis. Papillomavirus Research, 2015. 11 3. Munoz, N., et al., Incidence, duration, and determinants of cervical human papillomavirus 12 13 infection in a cohort of Colombian women with normal cytological results. J Infect Dis, 2004. 14 190(12): p. 2077-87. 15 4. McCredie,For M.R., et al., peer Natural history reviewof cervical neoplasia andonly risk of invasive cancer in women 16 with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol, 2008. 9(5): 17 p. 425-34. 18 5. Meyskens, F.L., Jr., et al., Enhancement of regression of cervical intraepithelial neoplasia II 19 (moderate dysplasia) with topically applied all-trans-retinoic acid: a randomized trial. J Natl 20 21 Cancer Inst, 1994. 86(7): p. 539-43. 22 6. Keefe, K.A., et al., A randomized, double blind, Phase III trial using oral beta-carotene 23 supplementation for women with high-grade cervical intraepithelial neoplasia. Cancer Epidemiol 24 Biomarkers Prev, 2001. 10(10): p. 1029-35. 25 7. Castle, P.E., et al., Evidence for frequent regression of cervical intraepithelial neoplasia-grade 2. 26 Obstet Gynecol, 2009. 113(1): p. 18-25. 27 8. Wang, S.M., et al., Six-year regression and progression of cervical lesions of different human 28 29 papillomavirus viral loads in varied histological diagnoses. Int J Gynecol Cancer, 2013. 23(4): p. 30 716-23. 31 9. Moscicki, A.B., et al., Rate of and risks for regression of cervical intraepithelial neoplasia 2 in 32 adolescents and young women. Obstet Gynecol, 2010. 116(6): p. 1373-80.

33 10. Herrero, R., et al., Rationale and design of a community-based double-blind randomized clinical http://bmjopen.bmj.com/ 34 trial of an HPV 16 and 18 vaccine in Guanacaste, Costa Rica. Vaccine, 2008. 26(37): p. 4795-808. 35 36 11. Ferlay, J., et al., GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC 37 CancerBase No. 11 [Internet], 2013, International Agency for Research on Cancer: Lyon, France. 38 12. World Development Indicators, 2016, World Bank. 39 13. International Labour Organization. Country Profiles: Nicaragua, 2016. 40 14. Campos, N.G., et al., The comparative and cost-effectiveness of HPV-based cervical cancer 41 screening algorithms in El Salvador. Int J Cancer, 2015. on September 26, 2021 by guest. Protected copyright. 42 15. Jeronimo, J., et al., A multicountry evaluation of careHPV testing, visual inspection with acetic 43 44 acid, and papanicolaou testing for the detection of cervical cancer. Int J Gynecol Cancer, 2014. 45 24(3): p. 576-85. 46 16. United States Department of Energy. Average Fuel Economy of Major Vehicle Categories. 47 August 1, 2016]; Available from: http://www.afdc.energy.gov/data/10310. 48 49 50 51 52 53 54 55 56 57 58 59 60 32

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from 90 90 /126 /126 /126 /126 #/ line # line #/ on page page on Reported Reported Page 70 of 72 NA NA 24/527 24/527 24/527 24/527 8/129 8/129 8/139 8/139 8/129 8/129 10/170- 1 8/137 8/137 7/119; 8 7/119; 8 7/112 7/112 2/18-42 2/18-42 1/1 1/1

2 1 4 9 6 8 7 5 12 # # 11b 11a

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist AND ABSTRACT AND Section/topic Measurementand valuation of preference based

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

hecklist item hecklist C BMJ Open http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

19 14 18 16 15 17 # 20b 13b 13a

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist Section/topic outcomes

Characterising Characterising uncertainty 20a Incrementalcosts and outcomes Study Study parameters RESULTS RESULTS Choiceof model Currency, Currency, price anddate, conversion

Estimating resources Estimating and costs

Analyticalmodels Assumptions Page 71 of 72 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

BMJ Open: first published as 10.1136/bmjopen-2016-015048 on 15 June 2017. Downloaded from

12-407 12-407 Page 72 of Reported Reported on page # page on 4/ 63-67 4/ 63-67 4/ 57-62 4/ 57-62 16-20/ 16-20/ 3 NA NA

Checklist item Checklist BMJ Open http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

22 24 23 #

CHEERSChecklist CHEERSChecklist CHEERSChecklist CHEERSChecklist

Section/topic

Characterising Characterising heterogeneity 21