BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on September 29, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Incidence and factors associated with neonatal hypothermia in Northern Uganda: a community-based cohort study

Journal: BMJ Open ManuscriptFor ID peerbmjopen-2020-041723 review only Article Type: Original research

Date Submitted by the 16-Jun-2020 Author:

Complete List of Authors: Mukunya, David ; University of Bergen Department of Medicine, Center for Intervention Science in Maternal and Child Health, Center for International Health; Sanyu Africa Research Institute Tumwine, James; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Nankabirwa, Victoria; Makerere University College of Health Sciences, Department of Epidemiology and Biostatistics; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Odongkara, Beatrice; Gulu University, Department of Paediatrics and Child Health Tongun, Justin; University of Juba, Department of Paediatrics and Child Health Arach, Agnes; Lira University, Department of Nursing and Midwifery http://bmjopen.bmj.com/ Tumuhamye, Josephine; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Napyo, Agnes; Busitema University, Department of Public Health Zalwango, Vivian; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Achora, Vicentina; Gulu University, Department of Obstetrics and Gynaecology Musaba, Milton; Busitema University, Department of Obstetrics and Gynaecology Ndeezi, Grace; Makerere University College of Health Sciences, on September 29, 2021 by guest. Protected copyright. Department of Paediatrics and Child Health Tylleskär, Thorkild; Universitetet i Bergen, Centre for International health

Epidemiology < TROPICAL MEDICINE, PAEDIATRICS, Public health < Keywords: INFECTIOUS DISEASES

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 24 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 2 of 24

1 2

3 Original research BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 Incidence and factors associated with neonatal hypothermia in Northern 7 Uganda: a community-based cohort study 8 9 David Mukunya1,3*, James K. Tumwine4, Victoria Nankabirwa2,5,6, Beatrice 10 Odongkara7, Justin B. Tongun8, Agnes A. Arach9, Josephine Tumuhamye3, Agnes 11 Napyo12, Vivian Zalwango4, Vicentina Achora10, Milton W. Musaba11, Grace 12 Ndeezi4, Thorkild Tylleskar6 13 14 15 1Sanyu Africa Research Institute, Mbale, Uganda 16 17 2Department of Epidemiology and Biostatistics, School of Public Health, Makerere 18 University CollegeFor of Health peer Sciences, review Kampala, Uganda only 19 20 3 21 Centre for Intervention Science in Maternal and Child Health (CISMAC), Centre for 22 International health, University of Bergen. 23 24 4Department of Paediatrics and Child Health, School of Medicine, Makerere 25 University College of Health Sciences, Kampala, Uganda 26 27 5Department of Epidemiology and Biostatistics, School of Public Health, Makerere 28 29 University College of Health Sciences, Kampala, Uganda 30 31 6Centre for Intervention Science and Maternal Child health (CISMAC), Centre for 32 International health, University of Bergen. 33 34 7Department of Paediatrics and Child Health, Gulu University 35 36 8Department of Paediatrics and Child Health, Juba University 37 http://bmjopen.bmj.com/ 38 39 9Department of Nursing and Midwifery, Lira University 40 41 10Department of Obstetrics and Gynaecology, Gulu University 42 43 11 44 Department of Obstetrics and Gynaecology, Busitema University Faculty of Health

45 Sciences on September 29, 2021 by guest. Protected copyright. 46 47 12Department of Public Health, Busitema University Faculty of Health Sciences 48 49 50 *Corresponding author 51 52 David Mukunya, P.O Box 2190, Mbale, Uganda 53 Email; [email protected], Mob; +256775152316 54 55 Key words: newborn care, neonatal care, kangaroo mother care 56 57 Word count: 2892 58 59 60

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 24 BMJ Open

1 2

3 Abstract BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 7 Objective: To determine the incidence and risk factors for hypothermia among 8 neonates in Lira district, Northern Uganda. 9 10 Setting: Three sub counties of Lira district in Northern Uganda 11 12 Design: This was a community-based prospective cohort study nested in a cluster 13 14 randomized controlled trial. 15 Participants: Mother – baby pairs enrolled in a cluster randomised controlled trial, 16 designed to promote health facility births and newborn care practices, in Lira district, 17 Northern Uganda. An axillary temperature was taken during a home visit using a 18 lithium battery-operatedFor peerdigital thermometer review only 19 20 21 Primary and secondary outcomes: The primary outcome measure was the incidence 22 of hypothermia. The secondary outcome measures were socioeconomic, 23 environmental, birth and newborn related factors associated with moderate to severe 24 hypothermia determined using a generalized estimating equation model for the 25 Poisson family, with a log link, taking clustering into account, and assuming an 26 exchangeable correlation. 27 28 29 Results: We recruited 1527 participants. Of these, 1330 had a temperature taken 30 within the first 3 days of life. The incidence of hypothermia (temperature less than 31 36.5°C) was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Of these, 32% (429/1330), 32 95%CI (29.5-35.2)] had mild hypothermia (temperature 36.0°C - <36.5°C), whereas 33 18.7% (249/1330), 95% CI (15.8-22.0) had moderate hypothermia (temperature 34 32.0°C - <36.0°C). None had severe hypothermia (temperature less than 32.0°C). At 35 multivariable analysis, the factors associated with neonatal hypothermia included: 36 home birth [Adjusted Risk Ratio, ARR, 1.9, 95% CI (1.4-2.6)]; low birth weight 37 http://bmjopen.bmj.com/ 38 [ARR 1.7, 95%CI (1.3-2.3)]; and delayed breastfeeding initiation [ARR 1.2, 95%CI 39 (1.0-1.5]. 40 41 Conclusion: The incidence of neonatal hypothermia was very high, demonstrating 42 that communities in tropical climates should not ignore neonatal hypothermia. 43 44 Interventions designed to address neonatal hypothermia should consider ways of

45 reaching newborns born at home, as those with low birth weight are at greater risk of on September 29, 2021 by guest. Protected copyright. 46 hypothermia. Early breastfeeding initiation should also be promoted to address 47 neonatal hypothermia. 48 49 Article summary: 50 51 52 Strengths and limitations of this study; 53 54  This is the first purely community based assessment of neonatal hypothermia 55 in sub-Saharan Africa 56  Estimates obtained are generalizable to settings with a significant proportion 57 of home births unlike previous estimates from health facility based studies 58  We suggest that hypothermia is a significant problem, especially in cases of 59 60 home births, low birth weight, and delayed breastfeeding initiation

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 24

1 2

3  Our choice of using a digital thermometer, placed in the axilla could have BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 underestimated hypothermia, but this was the most socially acceptable option 5 6  We did not measure some risk factors such as deliver room temperature and 7 maternal body temperature 8 9 10 11 12 13 14 15 16 17 18 1 IntroductionFor peer review only 19 20 21 Neonatal mortality (death of newborns less than 28 days) in Uganda is unacceptably 22 high, at 22.3 deaths per 1,000 live births compared to 1.6 deaths per 1,000 live births 23 in high-income countries [1]. In order to attain the global target of reducing neonatal 24 mortality to under 12 deaths per 1,000 live births by 2030 [2], there is a need to 25 identify and quantify the predictors of neonatal mortality; especially those that are 26 preventable by available low-cost interventions [3,4]. One of the predictors of 27 neonatal mortality that can easily be solved by available low-cost interventions is 28 29 neonatal hypothermia [5]. 30 31 Neonatal hypothermia, defined as an axillary temperature less than 36.5°C [6,7], is 32 associated with increased neonatal morbidity and mortality [8-10]. Countries with 33 high neonatal mortality have high rates of neonatal hypothermia [11]. Hypothermia 34 mainly contributes to mortality by worsening outcomes of severe neonatal infections, 35 preterm birth, and birth asphyxia. It is estimated that 20% of deaths due to 36 prematurity and 10% of deaths in term babies could be prevented by improved 37 http://bmjopen.bmj.com/ 38 thermal care [12]. In addition, neonatal hypothermia results in reduced growth and 39 development [13]. 40 41 Newborns are unable to maintain their body temperature without thermal protection 42 [14]. They are susceptible to hypothermia due to physical and environmental factors. 43 44 Physical factors that predispose neonates to hypothermia include a large surface area

45 to volume ratio, thin skin, and low amounts of insulating fat [5,11,14,15]. on September 29, 2021 by guest. Protected copyright. 46 Environmental factors that predispose neonates to hypothermia include poor thermal 47 practices around the time of birth, such as keeping the neonate away from the mother 48 and bathing the newborn within 24 hours of birth [16], which are common practices in 49 sub-Saharan Africa [17,18]. The World Health Organization recommends a ten-step 50 warm chain to prevent neonatal hypothermia: a warm delivery room, immediate 51 52 drying, delayed bathing, skin to skin contact, early and exclusive breastfeeding, 53 appropriate clothing/bedding, keeping the baby with the mother, warm transportation 54 and resuscitation, and training/raising awareness on the dangers of hypothermia [6]. 55 However, these actions are often suboptimal in most communities in sub-Saharan 56 Africa [19], and disregarded with the misguided assumption that a warm climate 57 guarantees thermal protection to the newborns [20,21]. Newborns are at greatest risk 58 of hypothermia on the first day of life and this is mainly a result of evaporation of 59 60 amniotic fluid and the neonate’s limited ability to generate heat [15,22].

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 24 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 Despite a significant proportion of births and deaths taking place at home in sub- 5 6 Saharan Africa, there is little to no data on hypothermia obtained from community 7 studies [5,23]. Previous estimates of hypothermia in sub-Saharan Africa have mostly 8 been obtained from health facility studies [9,10,20,21,24,25] and may therefore not be 9 representative of populations with poor health-seeking behaviors. Researchers 10 conducting community-based studies have been encouraged to incorporate axillary 11 temperatures with standard inexpensive digital thermometers in their study protocols 12 in order to enrich the literature on community estimates of neonatal hypothermia [5]. 13 14 This information is necessary when advocating for the scale-up of existing 15 interventions known to reduce hypothermia [23]. Therefore, in this community-based 16 study, we determined the incidence and risk factors for neonatal hypothermia in Lira 17 district, Northern Uganda. 18 For peer review only 19 20 21 2 Materials and Methods 22 23 24 Study setting 25 This study was conducted in Lira district is a post-conflict area in Northern Uganda in 26 the sub-counties of Aromo, Agweng, and Ogur between January 2018 and March 27 2019. The trial was conducted in the sub-counties of Aromo, Agweng and Ogur, as 28 29 they had the poorest maternal and newborn health indicators in the district. About 30 400,000 people live in Lira; the majority live in rural areas and practice subsistence 31 farming [26]. In the region covering Lira district, approximately 29 out of every 1,000 32 newborns died in the first 28 days of life [27]. During the period of this study (Jan 33 2018 - March 2019), the average monthly temperatures ranged from 27.8°C to 35.0°C 34 (Ngeta weather station, Lira district). 35 36 Study design 37 http://bmjopen.bmj.com/ 38 This was a prospective cohort study, where pregnant women were recruited during 39 pregnancy and followed up until they gave birth. This study was nested within a 40 cluster randomized controlled trial designed to promote health facility birth, newborn 41 care practices (early and exclusive breastfeeding, skin to skin care), and timely 42 postnatal health facility visits. 43 44

45 Study participants on September 29, 2021 by guest. Protected copyright. 46 Newborns born to mothers participating in the cluster randomized controlled trial 47 promoting health facility births and newborn care practices. 48 49 Power and sample size 50 We had 1330 neonates in our study. The participants were initially enrolled in a 51 52 cluster randomized controlled study which had a neonatal hypothermia intra cluster 53 correlation coefficient of 0.044, and average cluster sample size of 65, giving us a 54 design effect of 3.8, and effective sample size of 350, resulting in absolute precision 55 of 1.5% to 5.2%, i.e. the difference between the point estimate and the 95% 56 confidence interval (CI) for incidence values ranging from 2% to 50%. Since we were 57 studying a very common outcome, we deemed this precision adequate. 58 59 60 Inclusion criteria

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 24

1 2

3 Newborns born alive to mothers participating in the cluster randomized controlled BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 trial. 5 6 7 Exclusion criteria 8 Newborns whose mothers were too sick to participate in the interview, and newborns 9 that died before we visited. 10 11 Main variables 12 13 14 Outcome variable 15 The outcome variable in this study was hypothermia, which was defined as mild 16 hypothermia if the axillary temperature was between 36.0°C and less than 36.5°C, 17 moderate if the temperature was between 32.0°C and less than 36.0°C, and severe 18 hypothermia if theFor temperature peer was lessreview than 32.0°C. only 19 20 21 Exposure variables/risk factors 22 23 Data was collected on several risk factors during pregnancy and immediately after 24 birth. These included: maternal age, parity, maternal education, paternal education, 25 wealth, singleton or multiple birth, sex of the newborn, place of birth, birth weight, 26 early breastfeeding initiation, bathing of the newborn, and the place the newborn was 27 placed immediately after birth. We classified the season as wet if the average monthly 28 29 precipitation was 60 mm or more (Koppen-Geiger climate classification) [28]. The 30 average monthly precipitation and temperature for the study period were obtained 31 from the Ngeta weather station in Lira district. Wealth quintiles were calculated from 32 an asset-based index using principal component analysis. The following assets and 33 house characteristics were considered: cupboard, bicycle, radio, mobile phone, 34 motorcycle, cement floor, iron sheets, burnt bricks, and land ownership. We defined 35 36 early breastfeeding initiation as the initiation of breastfeeding within one hour of birth. 37 http://bmjopen.bmj.com/ 38 39 40 41 Data collection 42 43 44 As part of the trial in which this study was nested, a team of 42 research assistants

45 collected data and conducted measurements on the first day of birth, or as soon as on September 29, 2021 by guest. Protected copyright. 46 possible after birth at the mother’s home. A temperature was taken high in the axilla 47 during the study visit. We used a lithium battery-operated digital thermometer: Model 48 TM01 (manufactured by Cotronic Manufacturing, Shenzhen). The research assistants 49 were trained on how to measure temperature and supervised by a team consisting of 50 51 three paediatricians, one obstetrician, two general practitioners, two nurses, and one 52 data analyst. Temperature measurements were mostly conducted before taking the 53 baby's anthropometric measurements, with emphasis placed on minimizing the time 54 the babies were exposed to the cold. Measurements involved putting the tip of the 55 thermometer high up in the apex of the axilla, halfway between the anterior and 56 posterior margins, and holding the arm in place until an automatic audible beep was 57 heard. Two measurement readings in degrees Celsius were taken and the average of 58 59 these used. Thermometers were cleaned with cotton wool soaked in 70% alcohol after 60 the examination.

5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 24 BMJ Open

1 2

3 Recruitment and follow-up BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 All villages had a recruiter who was elected during the community sensitization 5 6 meetings of the trial. The recruiter was a female resident in the cluster. Recruiters 7 identified pregnant women and accompanied research assistants to the home of the 8 women during recruitment. They were trained during a one-day workshop, which 9 emphasized ethics, confidentiality, and good record keeping. Recruiters were also 10 given a cell phone to contact the team (site supervisor/research assistants) whenever 11 they identified a pregnant woman or whenever a pregnant woman had given birth. 12 They were paid Uganda Shillings 5,000 (USD 1.4) whenever they identified an 13 14 eligible participant and whenever they informed the team within 24 hours of a mother 15 giving birth. Approximately 250 recruiters were trained. After a recruiter informed the 16 team of an eligible participant, a research assistant would be sent to the participant to 17 ascertain eligibility, obtain consent to participate in the study and conduct the 18 interview. To ensureFor that peer recruiters were review reporting all pregnantonly women, we employed 19 community health workers (village health team members) to conduct a census of all 20 21 pregnant women in the area. Pregnant mothers and their relatives were also 22 encouraged to contact the study team immediately after giving birth. Research 23 assistants also obtained phone numbers of pregnant women and their relatives and 24 periodically conducted follow phone calls and visits to ensure that mothers were 25 visited immediately after birth. 26 27 Patient and Public Involvement 28 29 The public was not involved in the design and conceptualisation of the study but they 30 were involved in the recruitment of participants. We held community meetings in 31 each village during which a recruiter was elected from among the village members. 32 The recruiter was responsible for recruitment in their village. The results of this study 33 will be disseminated to the wider community through community dialogue meetings 34 at parish level in each participating village. 35 36 Statistical analysis 37 http://bmjopen.bmj.com/ 38 Data were analyzed using Stata version 14.0 (StataCorp; College Station, TX, USA). 39 Study characteristics were compared across the exposure status and summarized as 40 proportions for categorical data and means for continuous data. Hypothermia was 41 categorized using both the WHO classification [6], and a classification suggested by 42 Mullany et al [23], and presented as proportions with corresponding 95% Confidence 43 44 intervals adjusted for clustering. Factors associated with moderate to severe

45 hypothermia were determined using a generalized estimating equation model for the on September 29, 2021 by guest. Protected copyright. 46 Poisson family, with a log link, allowing for the clustering and assuming an 47 exchangeable correlation. Risk factors in the model were determined a priori during a 48 review of the literature on the subject. Factors included as risk factors in our model 49 included: mother’s age, mother’s education, mode of birth, place of birth, low birth 50 weight, wealth, parity, season, place baby put immediately after birth, cleaning/drying 51 52 the baby immediately after birth, bathing the baby, delayed initiation of breastfeeding. 53 All variables included in the model were assessed for collinearity and considered 54 collinear if they had a variance inflation factor greater than 10. In the case of 55 collinearity, we retained the variable with greater biological plausibility and/or 56 measure of association. The multivariable analyses were based on a complete case 57 analysis. However, we conducted sensitivity analyses of best case, worst case and 58 most realistic scenarios to assess the potential effect of the missing data. 59 60

6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 24

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 3 Results 7 8 Participant characteristics 9 On our visits to the participants, we were able to take the temperature measurements 10 of 1527 neonates; of these we used the data of 1330 for whom temperatures were 11 taken within the first 72 hours after birth (Figure 1). The mean age of mothers in the 12 study was 24.6 years (Standard Deviation (SD) 6.8) and their median education was 5 13 14 years (Inter Quartile Range (IQR) 3-6). The mean weight of children in the study was 15 3.2 kg (SD 0.5). The rest of the study characteristics are presented in table 1. 16 17 Hypothermia 18 The mean temperatureFor was peer 36.4°C (SD review 0.7), and the median only temperature was 36.4°C 19 (IQR 36.1°C -36.8°C). The minimum temperature recorded was 32.0 and the 20 21 maximum temperature recorded was 39.4 °C. The incidence of hypothermia 22 (temperature less than 36.5°C) was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Of these, 23 32% (429/1330), 95%CI (29.5-35.2)] had mild hypothermia (temperature 36.0°C - 24 <36.5°C), whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate hypothermia 25 (temperature 32.0°C - <36.0°C). No child had severe hypothermia (temperature less 26 than 32.0°C) (Table 2a). We also graded hypothermia according to a classification 27 proposed by Mullany et al and present the results in Table 2b. Sensitivity analyses 28 29 conducted suggested that we might have underestimated the burden (Table 2c,d). 30 31 Mortality 32 The risk of death among newborns with moderate hypothermia was 3/249 (1.2%, 33 95%CI 0.38-3.7), compared to 6/1023 (0.59%, 95% CI 0.28-1.2) among newborn 34 with normal temperature, resulting in a crude risk ratio of 2.0 (95% CI 0.60-6.9). 35 36 Factors associated with hypothermia 37 http://bmjopen.bmj.com/ 38 Using multivariable analysis, the factors associated with hypothermia with neonatal 39 hypothermia included: home birth [Adjusted Risk Ratio, ARR, 1.9, 95% CI (1.4-2.6)], 40 low birth weight [ARR 1.7, 95%CI (1.3-2.3)], and delayed breastfeeding initiation 41 [ARR 1.2, 95%CI (1.0-1.5] (Table 3). 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 24 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 4. Discussion 7 8 The incidence of hypothermia in this study was high. Half of the newborns developed 9 hypothermia; 33% developed mild hypothermia; 19% developed moderate 10 hypothermia. Similar findings were observed in a community-based study in Nepal, 11 where 59% of neonates developed hypothermia on the first day [29], and in another 12 community-based study in India where the incidence of hypothermia was 45% [30]. 13 14 However, the incidence of hypothermia observed in our study was much higher than 15 that observed in two other studies in India, which observed an incidence of 11% [31] 16 and 17% [32]. The difference could be explained by the different definitions of 17 hypothermia used in the studies. We defined hypothermia as a temperature less than 18 36.5 °C in accordanceFor with peer recommendations review from the onlyWorld Health Organization 19 [6], whereas Kumar et al defined hypothermia as temperature less than 35.6 °C, and 20 21 Bang et al defined hypothermia as a temperature less than 35°C. 22 23 Newborns who had low birth weight were more likely to be hypothermic compared to 24 newborns with normal birth weight. This finding is not surprising. Similar findings 25 were observed in a community-based study conducted in Nepal [33] and in many 26 other hospital-based studies in Uganda, Ethiopia [24,25], and other countries [20,34]. 27 Low birth weight infants have less capability to conserve and generate heat. This is 28 29 mainly because of physiological factors such as the reduced amount of brown fat and 30 a poor shivering reflex [15,35]. These thermo-protective mechanisms are needed to 31 maintain normal temperature in newborns who are exposed to hypothermic situations. 32 33 Babies born at home were more likely to be hypothermic compared to babies born in 34 health facilities. This finding has also been reported in other settings [35]. A study in 35 Uganda found that mothers who gave birth at home were more likely to practice sub- 36 optimal thermal care practices [36]. Mothers who give birth at home are more likely 37 http://bmjopen.bmj.com/ 38 to bathe their babies soon after birth [37,38], which could explain the increased risk of 39 hypothermia observed in babies born at home. The main reason for bathing newborns 40 early is the belief that newborns are dirty, having come into contact with maternal 41 fluids and the vernix caseosa [17,39-41]. Bathing newborns is also perceived as a 42 prerequisite to good neonatal rest and sleep after birth [39]. 43 44

45 Despite the generally impoverished nature of the study area, belonging to a relatively on September 29, 2021 by guest. Protected copyright. 46 lower social economic status was also a risk factor for hypothermia in this population. 47 Mothers with low socioeconomic status often lack resources to buy materials that can 48 keep the baby warm[42] and may have limited access to health information [43]. This 49 should not be a big problem if the mother practices skin to skin care. Unfortunately, 50 many mothers in Uganda and other countries in sub-Saharan Africa do not practice 51 52 skin-to-skin care[41,44-46]. Reasons for not practicing adequate newborn care 53 include beliefs that skin-to-skin care could result in the transmission of diseases to the 54 baby and could hurt the umbilical cord of the baby [41,45]. 55 56 Mothers who delayed putting their babies to the breast were more likely to have 57 hypothermic babies. This finding was also observed in the community-based study in 58 Nepal [33]. Newborns who are breastfed early receive warmth from their mothers and 59 60 this explains the reduction in hypothermia [23,47]. Mothers who had higher education

8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 24

1 2

3 were less likely to have hypothermic babies, although this finding was imprecise. BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 There was also no difference between mothers in the intervention group and the 5 6 control group, meaning the intervention did not prevent the newborns from becoming 7 hypothermic. 8 9 Methodological considerations; 10 We did not measure some risk factors such as delivery room temperature and 11 maternal temperature. We could also have underestimated hypothermia by using a 12 digital thermometer, placed in the axilla. Digital thermometers might slightly over- or 13 14 under-estimate temperature readings as compared to mercury thermometers [48-50]. 15 We used these because they are inexpensive, locally available, and easy to use by 16 community workers [29]. In addition, digital thermometers are easier to use in poorly 17 lit rural homes [29]. We used axillary measurements because they were easier to do, 18 safer, and more Foracceptable peer than rectal review measurements [29]. only In a systematic review 19 studying differences between rectal and axillary temperatures, the pooled mean 20 21 difference of rectal minus axillary temperature was estimated to be 0.17°C, ranging 22 from -0.15°C to 0.5°C [51]. Our study could have suffered from a selection bias since 23 only 75% of eligible participants were recruited. From our sensitivity analysis, we 24 believe hypothermia is still a big challenge, and that this selection bias might have 25 slightly underestimated the burden, since it was possibly the very sick who were not 26 visited within 72 hours of birth. We believe this selection bias also greatly 27 underestimated the mortality attributed to hypothermia since many more children died 28 29 in the unmeasured group. This is understandable since the majority of newborn deaths 30 in the study, as would be expected, occurred in the first few hours after birth, before 31 our team were able to reach the scene. We believe that our findings are generalizable 32 to rural areas in tropical low-income countries with similar newborn care practices. 33 34 35 6 Conclusion 36 The incidence of neonatal hypothermia was very high, demonstrating that 37 http://bmjopen.bmj.com/ 38 communities in tropical climates should not ignore neonatal hypothermia. 39 Interventions designed to address neonatal hypothermia should consider ways of 40 reaching newborns born at home, as these are at greater risk of hypothermia. Low 41 birth weight newborns, and newborns born to mothers in the poorest socioeconomic 42 status, should also be prioritized. 43 44

45 Ethics approval and consent to participate on September 29, 2021 by guest. Protected copyright. 46 Ethical approval to conduct the study was obtained from the following bodies: 1) 47 Research and Ethics committee School of Medicine, Makerere University (SOMREC: 48 REF 2015-121); 2) Uganda National Council of Science and Technology (UNCST: 49 SS 3954); 3) Regional Committees for Medical and Health Research Ethics (REK 50 VEST 2017/2079) and the trial was registered at ClinicalTrial.gov as NCT02605369. 51 52 We also obtained permission from the Ministry of Health and Lira Local Government. 53 Written informed consent was obtained from the respondents in the study. Research 54 assistants were trained in confidentiality and the right of the respondent to withdraw 55 their participation at any time during the study. At the community level, we obtained 56 permission to include clusters during community sensitization meetings, after which 57 the community members democratically elected recruiters, and peer buddies when 58 applicable, from amongst themselves. 59 60

9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 24 BMJ Open

1 2

3 Consent for publication BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 Not applicable 5 6 7 Availability of data and materials 8 The datasets used and/or analysed during the current study are available from the 9 corresponding author on reasonable request. 10 11 Funding 12 Funding was obtained from the Survival Pluss project; grant number UGA-13-0030 at 13 14 Makerere University. Survival Pluss project is funded by The Norwegian Program for 15 Capacity Development in Higher Education and Research for Development 16 (NORHED) under The Norwegian Agency for Development Cooperation (NORAD). 17 18 Author contributionsFor peer review only 19 David Mukunya (D.M), James K. Tumwine (J.K.T), Victoria Nankabirwa (V.N), 20 21 Grace Ndeezi (G.N), and Thorkild Tylleskar (T.T) conceived, designed, supervised 22 the study, analyzed the data, and wrote the first draft of manuscript. Milton W. 23 Musaba (M.W.M), Josephine Tumuhamye (J.T), Justin. B. Tongun (J.B.T), Agnes 24 Napyo (A.N), Vivian Zalwango (V.Z), Vicentina Achora (V.A), Beatrice Odongkara 25 (B.O), and Agnes Anna Arach (A.A.A) were instrumental in the design and 26 supervision of the study, and in drafting of the manuscript. All authors read and 27 approved the final version to be published. 28 29 30 Conflict of interest 31 All authors declare no conflict of interest. 32 33 Acknowledgments 34 In a special way, we acknowledge the District Health Office of Lira district, and the 35 various district, sub-county, parish, and village leaders for their assistance in this 36 study. We thank the study participants for accepting to be part of the study and 37 http://bmjopen.bmj.com/ 38 research assistants for working tirelessly to make this work a reality. In a special way, 39 we acknowledge the excellent work performed by our recruiters in making this study 40 possible. Finally, we extend heartfelt appreciation to Ms. Jo Weeks for the excellent 41 English editing. 42 43 44

45 References on September 29, 2021 by guest. Protected copyright. 46 47 1 GBD 2016 Mortality Collaborators. Global, regional, and national under-5 48 mortality, adult mortality, age-specific mortality, and life expectancy, 1970-2016: a 49 systematic analysis for the Global Burden of Disease Study 2016. Lancet (London, 50 England) 2017;390(10100):1084-150. 51 52 53 2 United Nations. Sustainable Development Goals. Secondary Sustainable 54 Development Goals 2015. http://www.un.org/sustainabledevelopment/summit/. 55 56 3 Darmstadt GL, Bhutta ZA, Cousens S, et al. Evidence-based, cost-effective 57 interventions: how many newborn babies can we save? Lancet (London, England) 58 2005;365(9463):977-88. 59 60

10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 24

1 2

3 4 Knippenberg R, Lawn JE, Darmstadt GL, et al. Systematic scaling up of BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 neonatal care in countries. Lancet (London, England) 2005;365(9464):1087-98. 5 6 7 5 Lunze K, Bloom DE, Jamison DT, et al. The global burden of neonatal 8 hypothermia: systematic review of a major challenge for newborn survival. BMC 9 medicine 2013;11:24. 10 11 6 World Health Organization. Thermal Protection of the Newborn: a practical 12 guide. Geneva: World Health Organization, 1997. 13 14 7 Lunze K, Yeboah-Antwi K, Marsh DR, et al. Prevention and management of 15 neonatal hypothermia in rural Zambia. PloS one 2014;9(4):e92006. 16 17 18 8 ChristenssonFor K, Bhatpeer GJ, Eriksson review B, et al. The only effect of routine hospital care 19 on the health of hypothermic newborn infants in Zambia. Journal of tropical 20 pediatrics 1995;41(4):210-4. 21 22 9 Kambarami R, Chidede O. Neonatal hypothermia levels and risk factors for 23 mortality in a tropical country. The Central African journal of medicine 2003;49(9- 24 10):103-6. 25 26 27 10 Sodemann M, Nielsen J, Veirum J, et al. Hypothermia of newborns is 28 associated with excess mortality in the first 2 months of life in Guinea-Bissau, West 29 Africa. Tropical medicine & international health : TM & IH 2008;13(8):980-6. 30 31 11 Kumar V, Shearer JC, Kumar A, et al. Neonatal hypothermia in low resource 32 settings: a review. Journal of perinatology : official journal of the California 33 Perinatal Association 2009;29(6):401-12. 34 35 12 Bhutta ZA, Das JK, Bahl R, et al. Can available interventions end preventable 36 deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet (London, 37 http://bmjopen.bmj.com/ 38 England) 2014;384(9940):347-70. 39 40 13 Glass L, Silverman WA, Sinclair JC. Relationship of thermal environment and 41 caloric intake to growth and resting metabolism in the late neonatal period. Biologia 42 neonatorum Neo-natal studies 1969;14(5):324-40. 43 44 14 Lunze K, Hamer DH. Thermal protection of the newborn in resource-limited 45 environments. Journal of perinatology : official journal of the California Perinatal on September 29, 2021 by guest. Protected copyright. 46 47 Association 2012;32(5):317-24. 48 49 15 Adamson SK, Jr., Towell ME. THERMAL HOMEOSTASIS IN THE FETUS 50 AND NEWBORN. Anesthesiology 1965;26:531-48. 51 52 16 Bergstrom A, Byaruhanga R, Okong P. The impact of newborn bathing on the 53 prevalence of neonatal hypothermia in Uganda: a randomized, controlled trial. Acta 54 paediatrica (Oslo, Norway : 1992) 2005;94(10):1462-7. 55 56 57 17 Waiswa P, Kemigisa M, Kiguli J, et al. Acceptability of evidence-based 58 neonatal care practices in rural Uganda - implications for programming. BMC 59 pregnancy and childbirth 2008;8:21. 60

11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 24 BMJ Open

1 2

3 18 Hill Z, Tawiah-Agyemang C, Manu A, et al. Keeping newborns warm: beliefs, BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 practices and potential for behaviour change in rural Ghana. Tropical medicine & 5 6 international health : TM & IH 2010;15(10):1118-24. 7 8 19 Coalter WS, Patterson SL. Sociocultural factors affecting uptake of home- 9 based neonatal thermal care practices in Africa: A qualitative review. Journal of child 10 health care : for professionals working with children in the hospital and community 11 2017;21(2):132-41. 12 13 20 Manji KP, Kisenge R. Neonatal hypothermia on admission to a special care 14 unit in Dar-es-Salaam, Tanzania: a cause for concern. The Central African journal of 15 16 medicine 2003;49(3-4):23-7. 17 18 21 ByaruhangaFor R, Bergstrom peer A, reviewOkong P. Neonatal only hypothermia in Uganda: 19 prevalence and risk factors. Journal of tropical pediatrics 2005;51(4):212-5. 20 21 22 Smales OR, Kime R. Thermoregulation in babies immediately after birth. 22 Archives of disease in childhood 1978;53(1):58-61. 23 24 23 Mullany LC. Neonatal hypothermia in low-resource settings. Seminars in 25 26 perinatology 2010;34(6):426-33. 27 28 24 Tasew H, Gebrekristos K, Kidanu K, et al. Determinants of hypothermia on 29 neonates admitted to the intensive care unit of public hospitals of Central Zone, 30 Tigray, Ethiopia 2017: unmatched case-control study. BMC research notes 31 2018;11(1):576. 32 33 25 Demissie BW, Abera BB, Chichiabellu TY, et al. Neonatal hypothermia and 34 associated factors among neonates admitted to neonatal intensive care unit of public 35 36 hospitals in Addis Ababa, Ethiopia. BMC pediatrics 2018;18(1):263.

37 http://bmjopen.bmj.com/ 38 26 Uganda Bureau of Statistics. The National Population and Housing Census 39 2014 – Area 40 Specific Profile Series, Kampala, Uganda. Secondary The National Population and 41 Housing Census 2014 – Area 42 Specific Profile Series, Kampala, Uganda 2017. 43 http://www.ubos.org/onlinefiles/uploads/ubos/2014CensusProfiles/MUKONO.pdf. 44

45 on September 29, 2021 by guest. Protected copyright. 46 27 Uganda Bureau of Statistics (UBOS) and ICF. Uganda Demographic and 47 Health Survey 2016. Kampala, Uganda and Rockville, Maryland, USA: UBOS and 48 ICF, 2018. 49 50 28 Peel MC, Finlayson BL, McMahon TA. Updated world map of the Koppen- 51 Geiger climate classification. Hydrol Earth Syst Sci 2007;11:1633–44. 52 53 29 Mullany LC, Katz J, Khatry SK, et al. Incidence and seasonality of 54 55 hypothermia among newborns in southern Nepal. Archives of pediatrics & adolescent 56 medicine 2010;164(1):71-7. 57 58 30 Darmstadt GL, Kumar V, Yadav R, et al. Introduction of community-based 59 skin-to-skin care in rural Uttar Pradesh, India. Journal of perinatology : official 60 journal of the California Perinatal Association 2006;26(10):597-604.

12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 24

1 2

3 31 Kumar R, Aggarwal AK. Body temperatures of home delivered newborns in BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 north India. Tropical doctor 1998;28(3):134-6. 5 6 7 32 Bang AT, Reddy HM, Baitule SB, et al. The incidence of morbidities in a 8 cohort of neonates in rural Gadchiroli, India: seasonal and temporal variation and a 9 hypothesis about prevention. Journal of perinatology : official journal of the 10 California Perinatal Association 2005;25 Suppl 1:S18-28. 11 12 33 Mullany LC, Katz J, Khatry SK, et al. Neonatal hypothermia and associated 13 risk factors among newborns of southern Nepal. BMC medicine 2010;8:43. 14 15 34 Zayeri F, Kazemnejad A, Ganjali M, et al. Hypothermia in Iranian newborns. 16 17 Incidence, risk factors and related complications. Saudi medical journal 18 2005;26(9):1367-71.For peer review only 19 20 35 Onalo R. Neonatal hypothermia in sub-Saharan Africa: a review. Niger J Clin 21 Pract 2013;16(2):129-38. 22 23 36 Kabwijamu L, Waiswa P, Kawooya V, et al. Newborn Care Practices among 24 Adolescent Mothers in Hoima District, Western Uganda. PloS one 25 26 2016;11(11):e0166405. 27 28 37 Mrisho M, Schellenberg JA, Mushi AK, et al. Understanding home-based 29 neonatal care practice in rural southern Tanzania. Transactions of the Royal Society of 30 Tropical Medicine and Hygiene 2008;102(7):669-78. 31 32 38 Salasibew MM, Filteau S, Marchant T. A qualitative study exploring newborn 33 care behaviours after home births in rural Ethiopia: implications for adoption of 34 essential interventions for saving newborn lives. BMC pregnancy and childbirth 35 36 2014;14:412.

37 http://bmjopen.bmj.com/ 38 39 Adejuyigbe EA, Bee MH, Amare Y, et al. "Why not bathe the baby today?": 39 A qualitative study of thermal care beliefs and practices in four African sites. BMC 40 pediatrics 2015;15:156. 41 42 40 Shamba D, Schellenberg J, Hildon ZJ, et al. Thermal care for newborn babies 43 in rural southern Tanzania: a mixed-method study of barriers, facilitators and 44 potential for behaviour change. BMC pregnancy and childbirth 2014;14:267. 45 on September 29, 2021 by guest. Protected copyright. 46 47 41 Byaruhanga RN, Nsungwa-Sabiiti J, Kiguli J, et al. Hurdles and opportunities 48 for newborn care in rural Uganda. Midwifery 2011;27(6):775-80. 49 50 42 Lunze K, Dawkins R, Tapia A, et al. Market mechanisms for newborn health 51 in Nepal. BMC pregnancy and childbirth 2017;17(1):428. 52 53 43 Owor MO, Matovu JKB, Murokora D, et al. Factors associated with adoption 54 of beneficial newborn care practices in rural Eastern Uganda: a cross-sectional study. 55 56 BMC pregnancy and childbirth 2016;16:83. 57 58 44 Bee M, Shiroor A, Hill Z. Neonatal care practices in sub-Saharan Africa: a 59 systematic review of quantitative and qualitative data. Journal of health, population, 60 and nutrition 2018;37(1):9.

13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 24 BMJ Open

1 2

3 45 Byaruhanga RN, Bergstrom A, Tibemanya J, et al. Perceptions among post- BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 delivery mothers of skin-to-skin contact and newborn baby care in a periurban 5 6 hospital in Uganda. Midwifery 2008;24(2):183-9. 7 8 46 Waiswa P, Peterson S, Tomson G, et al. Poor newborn care practices - a 9 population based survey in eastern Uganda. BMC pregnancy and childbirth 10 2010;10:9. 11 12 47 Huffman SL, Zehner ER, Victora C. Can improvements in breast-feeding 13 practices reduce neonatal mortality in developing countries? Midwifery 14 2001;17(2):80-92. 15 16 17 48 Smith J. Are electronic thermometry techniques suitable alternatives to 18 traditional mercuryFor in glass peer thermometry review techniques in only the paediatric setting? Journal 19 of advanced nursing 1998;28(5):1030-9. 20 21 49 Jones HL, Kleber CB, Eckert GJ, et al. Comparison of rectal temperature 22 measured by digital vs. mercury glass thermometer in infants under two months old. 23 Clinical pediatrics 2003;42(4):357-9. 24 25 26 50 Latman NS, Hans P, Nicholson L, et al. Evaluation of clinical thermometers 27 for accuracy and reliability. Biomedical instrumentation & technology 28 2001;35(4):259-65. 29 30 51 Craig JV, Lancaster GA, Williamson PR, et al. Temperature measured at the 31 axilla compared with rectum in children and young people: systematic review. BMJ 32 (Clinical research ed) 2000;320(7243):1174-8. 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 16 of 24

1 2 3 4 5 6 7 Table 1: Participant characteristics of newborns assessed for hypothermia in Northern Uganda 8 9 † 10 All participants Late participants* Missed participants 11 No No hypothermia hypothermia Unknown 12 hypothermia For peerhypothermia review only 13 N=652 N=678 N=88 N=109 N=241 14 15 n (%) n (%) n (%) n (%) n (%) 16

Age of mother http://bmjopen.bmj.com/ 17 18 <=19 148 (22.7) 201 (29.7) 28 (31.8) 33 (30.3) 66 (27.4) 19 20-30 367 (56.3) 347 (51.2) 48 (54.6) 56 (51.4) 121 (50.2) 20 >30 137 (21.0) 130 (19.2) 12 (13.6) 20 (18.4) 54 (22.4) 21 Mother’s 22 23 education 24

None 74 (11.4) 105 (15.5) 6 (06.8) 12 (11.0) 34 (14.1) on September 29, 2021 by guest. Protected copyright. 25 Primary 513 (78.7) 519 (76.6) 73 (83.0) 85 (78.0) 190 (78.8) 26 27 Secondary 51 (07.8) 47 (06.9) 7 (08.0) 9 (08.3) 17 (07.1) 28 Tertiary 14 (02.2) 07 (01.0) 2 (02.3) 3 (02.8) - 29 Father’s 30 31 education 32 None 14 (02.2) 11 (01.6) 1 (01.1) 1 (0.92) 6 (02.5) 33 Primary 377 (57.8) 416 (61.4) 52 (59.1) 55 (50.5) 151 (62.7) 34 35 Secondary 177 (27.2) 147 (21.7) 23 (26.1) 28 (25.7) 51 (21.2) 36 Tertiary 41 (06.3) 38 (05.6) 4 (04.6) 8 (07.3) 12 (05.0) 37 Missing 43 (06.6) 66 (09.7) 8 (09.1) 17 (15.6) 21 (08.7) 38 39 40 41 42 15 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 17 of 24 BMJ Open

1 2 3 4 5 Parity 6 <=1 286 (43.9) 325 (47.9) 44 (50.0) 55 (50.5) 101 (41.9) 7 8 2-4 219 (33.6) 218 (32.2) 36 (40.9) 28 (25.7) 86 (35.7) 9 >4 147 (22.6) 135 (19.9) 8 (09.1) 26 (23.9) 54 (22.4) 10 Place of birth 11 12 Home 157 (24.1) For254 (37.5) peer26 (29.6) review40 (36.7) only100 (41.5) 13 Health facility 495 (75.9) 424 (62.5) 62 (70.5) 69 (63.3) 141 (58.5) 14 Caesarean

15 section 16 17 No 641 (98.3) 670 (98.8) 79 (89.8) 94 (86.2) 232 (96.3) http://bmjopen.bmj.com/ 18 Yes 11(01.7) 8 (01.2) 9 (10.2) 15 (13.8) 9 (03.7) 19 20 Marital status 21 Married 609 (93.4) 612 (90.3) 80 (90.9) 92 (84.4) 220 (91.3) 22 Single 43 (06.6) 66 (09.7) 8 (09.1) 17 (15.6) 21 (08.7) 23 24 Electricity 25 Yes 71 (10.9) 86 (12.7) 4 (04.6) 6 (05.5) 24 (10.0) on September 29, 2021 by guest. Protected copyright. 26 No 581 (89.1) 592 (87.3) 84 (95.5) 103 (94.5) 217 (90.0) 27 28 Presence of 29 mobile phone in 30 the household 31 Yes 346 (53.1) 363 (53.5) 42 (47.7) 53 (48.6) 159 (66.0) 32 33 No 306 (46.9) 315 (46.5) 46 (52.3) 56 (51.4) 82 (34.0) 34 Source of

35 drinking water 36 37 Borehole 319 (48.9) 340 (50.2) 54 (61.4) 58 (53.2) 138 (57.3) 38 Tap/piped water 88 (13.5) 84 (12.4) 9 (10.2) 10 (09.2) 20 (08.3) 39 40 41 42 16 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 18 of 24

1 2 3 4 5 Protected natural 131 (20.1) 150 (22.1) 13 (14.8) 20 (18.4) 43 (17.8) 6 spring 7 Unprotected 8 114 (17.5) 104 (15.3) 12 (13.6) 21 (19.3) 40 (16.6) 9 water source 10 Twin 11 No 648 (99.4) 668 (98.5) 87 (98.9) 107 (98.2) 237 (98.3) 12 For peer review only 13 Yes 4 (0.61) 10 (01.5) 1 (01.1) 2 (01.8) 4 (1.7) 14 Low birth

15 weight 16 17 No 613 (94.0) 622 (91.7) 83 (94.3) 101 (92.7) 15 (06.2) http://bmjopen.bmj.com/ 18 Yes 35 (05.4) 45 (06.6) 4 (04.6) 6 (05.5) 1 (0.41) 19 Missing 4 (0.6) 11 (01.6) 1 (01.1) 2 (01.8) 225 (93.4) 20 21 Wealth quintiles 22 1 (Poorest) 146 (22.4) 140 (20.7) 19 (21.6) 23 (21.1) 35 (14.5) 23 2 143 (21.9) 185 (27.3) 20 (22.7) 22 (20.2) 63 (26.1) 24 25 3 123 (18.9) 121 (17.9) 19 (21.6) 18 (16.5) 45 (18.7) on September 29, 2021 by guest. Protected copyright. 26 4 105 (16.1) 114 (16.8) 10 (11.4) 20 (18.4) 49 (20.3) 27 5 (Richest) 135 (20.7) 118 (17.4) 20 (22.7) 26 (23.9) 49 (20.3) 28 29 Season 30 Wet 589 (90.3) 579 (85.4) 74 (84.1) 87 (79.8) 47 (19.5) 31 Dry 63 (09.7) 99 (14.6) 14 (15.9) 22 (20.2) 194 (80.5) 32 33 Place baby 34 immediately 35 after birth 36 Mother chest or 37 547 (83.9) 548 (80.8) 68 (77.3) 76 (69.7) 163 (67.6) 38 abdomen 39 40 41 42 17 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 19 of 24 BMJ Open

1 2 3 4 5 Other place 105 (16.1) 130 (19.2) 20 (22.7) 33 (30.3) 78 (32.4) 6 7 Clean and dry 8 baby 9 immediately 10 No 68 (10.4) 104 (15.3) 10 (11.4) 21 (19.3) 41 (17.0) 11 12 Yes 584 (89.6) For574 (84.7) peer78 (88.6) review88 (80.7) only200 (83.0) 13 Bathed baby

14 before visit 15 16 No 326 (50.0) 274 (40.4) 1 (01.1) 2 (01.8) 81 (34.3) 17 Yes 326 (50.0) 404 (59.6) 87 (98.9) 107 (98.2) 155 (65.7) http://bmjopen.bmj.com/ 18 Died in first 19 20 month 21 No 643 (98.6) 675 (99.6) 88 (100.0) 108 (99.1) 227 (94.2) 22 Yes 9 (1.4) 3 (0.44) 0 (0.0) 1 (0.92) 14 (5.8) 23 24 Early 25 breastfeeding on September 29, 2021 by guest. Protected copyright. 26 initiation 27 28 No 208 (31.9) 257 (37.9) 35 (39.8) 58 (53.2) 110 (48.0) 29 30 Yes 444 (68.1) 421 (62.1) 53 (60.2) 51 (46.8) 119 (52.0) 31 32 *Participants whose temperature was measured after 3 days 33 34 Missed participants†: Eligible participants whose temperature was not measured 35 36 37 38 39 40 41 42 18 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 20 of 24

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 Table 2a: Prevalence of hypothermia in Lira district, Northern Uganda, defined by the 7 World Health Organization classification 8 9 Hypothermia n/N % 10 Mild (36.0-36.5) 429/1330 32.3 (29.5-35.2) 11 Moderate (32.0-35.9) 249/1330 18.7 (15.8-22.0) 12 13 Severe (<32.0) 0/1330 0 14 Any 678/1330 51.0 (46.9-55.1) 15 16 17 Table 2b: Prevalence of hypothermia in Lira district, Northern Uganda, defined by the 18 Mullany classificationFor peer review only 19 20 21 Hypothermia n/N % 22 Grade 1 (36.0-36.5) 429/1330 32.3 (29.5-35.2) 23 Grade 2 (35.0-35.99) 218/1330 16.4 (14.0-19.1) 24 Grade 3 (34.0-34.99) 26/1330 2.0 (1.2-3.1) 25 26 Grade 4 (less than 34.0) 5/1330 0.38 (0.16-0.90) 27 28 Table 2c: Sensitivity analyses assuming all unmeasured temperatures in the first three 29 days were normo-thermic 30 31 Hypothermia Best Case Scenario* Worst case scenario** 32 33 n/N % 34 Mild (36.0-36.5) 429/1768 24.3 (22.3-26.3) 867/1768 49.0 (46.7-51.4) 35 Moderate (32.0-35.9) 249/1768 14.1 (12.5-15.8) 687/1768 38.9 (36.6-41.2) 36 Severe (<32.0) 0/1768 0 0/1768 0 37 Any hypothermia 678/1768 38.3 (36.1- 40.7) 1554/1768 87.9 (86.3-89.4) http://bmjopen.bmj.com/ 38 39 *: Assuming all unmeasured temperatures were normothermic 40 **: Assuming all unmeasured temperatures were hypothermic 41 Confidence intervals calculated by the exact method 42 43 Table 2d: Sensitivity analyses assuming all unmeasured temperatures in the first three 44 days had similar distribution of hypothermia as observed, based on place of birth

45 on September 29, 2021 by guest. Protected copyright. 46 47 Hypothermia n/N % 48 Any hypothermia in 678/1330 51.0 (46.9-55.1) 49 measured infants 50 Any hypothermia in un 228/438 34.2 (30.6-38.0) 51 measured infants 52 53 54 Please note: 272 of the unmeasured infants were delivered at a health facility 55 (incidence of hypothermia for health facility births is 46.1% in first 3 days) and 56 166 of the unmeasured infants were delivered at home (incidence of hypothermia for 57 home births is 61.8% in first 3 days). Confidence intervals of unobserved calculated 58 by the exact method. 59 60

19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 24 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 Table 3: Factors associated with moderate to severe hypothermia among newborn in 7 Lira district Northern Uganda 8 9 10 Bi-variable P value Multivariable 11 N=1330 RR N=1315 12 13 RR (95% C.I) RR (95% C.I) 14 15 Intervention group 16 1 1 17 Control 18 Intervention 0.85 (0.62-1.2)For peer0.331 review1.0 (0.79-1.4) only 19 20 Age of mother 21 <=19 1 1 22 0.71 (0.58-0.88) 0.002 0.81 (0.59-1.1) 23 20-30 24 >30 0.70 (0.50-0.96) 0.029 0.75 (0.43-1.3) 25 26 Mother’s education 27 None 1 1 1 28 0.93 (0.69-1.2) 0.601 0.94 (0.70-1.3) 29 Primary 30 >=Secondary 0.53 (0.31-0.88) 0.014 0.63 (0.39-1.0) 31 32 Father’s education 33 None 1 34 35 Primary 1.2 (0.58-2.6) 0.596 36 Secondary 0.81 (0.35-1.9) 0.615 -

37 http://bmjopen.bmj.com/ 38 Tertiary 0.73 (0.27-2.0) 0.531 39 Parity 40 41 <=1 1 1 42 2-4 0.75 (0.57-0.99) 0.044 0.85 (0.57-1.3) 43 44 >4 0.77 (0.55-1.1) 0.120 0.84 (0.50-1.4)

45 Place of birth on September 29, 2021 by guest. Protected copyright. 46 47 Health Facility 1 1 48 Home 2.0 (1.5-2.6) <0.001 1.9 (1.4-2.6) 49 50 Caesarean section 51 No 1 1 52 53 Yes 0.94 (0.44-2.0) 0.866 0.82 (0.31-2.1) 54 Marital status 55 56 Single 1 - 57 Married 0.77 (0.55-1.1) 0.125 58 59 Low birth weight* 60 (less than 2.5)

20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 24

1 2

3 No 1 1 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 Yes 1.9 (1.4-2.6) <0.001 1.7 (1.3-2.3) 6 Wealth quintiles 7 8 1 (Poorest) 1 1 9 2 1.1 (0.82-1.6) 0.453 1.3 (0.91-1.7) 10 11 3 0.81 (0.57-1.1) 0.227 0.93 (0.67-1.3) 12 4 0.71 (0.46-1.1) 0.134 0.87 (0.59-1.3) 13 14 5 (Richest) 0.59 (0.40-0.87) 0.008 0.79 (0.53-1.2) 15 Season 16 17 Wet 1 1 18 Dry 1.3 (0.92-1.8)For peer0.142 review1.4 (1.0-1.9) only 19 20 Place baby 21 OtherPPUTimmediately place 1 1 22 after birth 23 Mother chest or 0.78 (0.61-0.99) 0.039 0.98 (0.76-1.3) 24 abdomen 25 Clean and dry 26 Nobaby immediately 1 1 27 28 Yes 0.87 (0.59-1.3) 0.479 0.96 (0.65-1.4) 29 Bathed baby before 30 visit 31 No 1 1 32 Yes 1.2 (0.98-1.5) 0.079 1.0 (0.81-1.2) 33 34 Breastfeeding 35 initiation 36 Early 1 1 37 http://bmjopen.bmj.com/ 38 Late 1.4 (1.1-1.8) 0.002 1.2 (1.0-1.5) 39 40 Child’s sex 41 Male 1 42 43 Female 1.1 (0.95-1.3) 0.184 - 44

45 on September 29, 2021 by guest. Protected copyright. 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 23 of 24 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 Figure 1: Study profile of neonates assessed for hypothermia in Lira district, Northern 27 Uganda. 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 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 24 of 24

1 STROBE Statement—Checklist of items that should be included in reports of cohort studies 2 Item Page 3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 No Recommendation number 5 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the 1 6 abstract 7 (b) Provide in the abstract an informative and balanced summary of what was 2 8 9 done and what was found 10 Introduction 11 Background/rationale 2 Explain the scientific background and rationale for the investigation being 3-4 12 13 reported 14 Objectives 3 State specific objectives, including any prespecified hypotheses 4 15 16 Methods 17 Study design 4 Present key elements of study design early in the paper 4 18 Setting For5 Describe peer the setting, review locations, and relevant only dates, including periods of 4 19 recruitment, exposure, follow-up, and data collection 20 21 Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of 5 22 participants. Describe methods of follow-up 23 (b) For matched studies, give matching criteria and number of exposed and NA 24 unexposed 25 26 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and 5 27 effect modifiers. Give diagnostic criteria, if applicable 28 Data sources/ 8* For each variable of interest, give sources of data and details of methods of 5-6 29 measurement assessment (measurement). Describe comparability of assessment methods if 30 31 there is more than one group 32 Bias 9 Describe any efforts to address potential sources of bias 5-6 33 Study size 10 Explain how the study size was arrived at 5 34 35 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, 6 36 describe which groupings were chosen and why

37 Statistical methods 12 (a) Describe all statistical methods, including those used to control for 6 http://bmjopen.bmj.com/ 38 confounding 39 40 (b) Describe any methods used to examine subgroups and interactions NA 41 (c) Explain how missing data were addressed 7 42 (d) If applicable, explain how loss to follow-up was addressed 7 43 44 (e) Describe any sensitivity analyses 7

45 Results on September 29, 2021 by guest. Protected copyright. 46 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially 23 47 48 eligible, examined for eligibility, confirmed eligible, included in the study, 49 completing follow-up, and analysed 50 (b) Give reasons for non-participation at each stage 23 51 (c) Consider use of a flow diagram 23 52 53 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) 7 54 and information on exposures and potential confounders 55 (b) Indicate number of participants with missing data for each variable of interest 7 56 57 (c) Summarise follow-up time (eg, average and total amount) NA 58 Outcome data 15* Report numbers of outcome events or summary measures over time 59 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates 6 60 and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 24 BMJ Open

1 (b) Report category boundaries when continuous variables were categorized 2 (c) If relevant, consider translating estimates of relative risk into absolute risk for NA

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 a meaningful time period 5 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and 7 6 sensitivity analyses 7 8 Discussion 9 Key results 18 Summarise key results with reference to study objectives 8 10 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or 9 11 12 imprecision. Discuss both direction and magnitude of any potential bias 13 Interpretation 20 Give a cautious overall interpretation of results considering objectives, 9 14 limitations, multiplicity of analyses, results from similar studies, and other 15 relevant evidence 16 17 Generalisability 21 Discuss the generalisability (external validity) of the study results 9 18 Other information For peer review only 19 Funding 22 Give the source of funding and the role of the funders for the present study and, if 10 20 21 applicable, for the original study on which the present article is based 22 23 *Give information separately for exposed and unexposed groups. 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Neonatal hypothermia in Northern Uganda: a community- based cross-sectional study

Journal: BMJ Open ManuscriptFor ID peerbmjopen-2020-041723.R1 review only Article Type: Original research

Date Submitted by the 17-Dec-2020 Author:

Primary Subject Public health Heading:

Secondary Subject Heading: Paediatrics

Epidemiology < TROPICAL MEDICINE, PAEDIATRICS, Public health < Keywords: INFECTIOUS DISEASES

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 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 29, 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 2 of 26

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 3 of 26 BMJ Open

1 2

3 1 Original research BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 2 5 6 3 Neonatal hypothermia in Northern Uganda: a community-based cross-sectional 7 4 study 8 5 9 6 David Mukunya1,3,12*, James K. Tumwine4, Victoria Nankabirwa2,3,5, Beatrice 10 7 Odongkara7, Justin B. Tongun8, Agnes A. Arach9, Josephine Tumuhamye3, Agnes 11 8 Napyo12, Vivian Zalwango4, Vicentina Achora10, Milton W. Musaba11, Grace 12 Ndeezi4, Thorkild Tylleskar6 13 9 14 10 15 11 1Sanyu Africa Research Institute, Mbale, Uganda 16 12 17 13 2Department of Epidemiology and Biostatistics, School of Public Health, Makerere 18 14 University CollegeFor of Health peer Sciences, review Kampala, Uganda only 19 15 20 3 21 16 Centre for Intervention Science in Maternal and Child Health (CISMAC), Centre for 22 17 International health, University of Bergen. 23 18 24 19 4Department of Paediatrics and Child Health, School of Medicine, Makerere 25 20 University College of Health Sciences, Kampala, Uganda 26 21 27 22 5Department of Epidemiology and Biostatistics, School of Public Health, Makerere 28 29 23 University College of Health Sciences, Kampala, Uganda 30 24 31 25 7Department of Paediatrics and Child Health, Gulu University 32 26 33 27 8Department of Paediatrics and Child Health, Juba University 34 28 35 29 9Department of Nursing and Midwifery, Lira University 36 30 37 http://bmjopen.bmj.com/ 38 31 10Department of Obstetrics and Gynaecology, Gulu University 39 32 40 33 11Department of Obstetrics and Gynaecology, Busitema University Faculty of Health 41 34 Sciences 42 35 43 12 44 36 Department of Public Health, Busitema University Faculty of Health Sciences

45 37 on September 29, 2021 by guest. Protected copyright. 46 38 47 39 *Corresponding author 48 40 David Mukunya, P.O Box 2190, Mbale, Uganda 49 41 Email; [email protected], Mob; +256775152316 50 42 51 52 43 Key words: newborn care, neonatal care, kangaroo mother care 53 44 54 45 Word count: 2892 55 46 56 47 57 48 58 49 59 60 50

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 26

1 2

3 51 Abstract BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 52 5 6 53 7 54 Objective: To determine the prevalence, predictors, and case fatality risk of 8 55 hypothermia among neonates in Lira district, Northern Uganda. 9 56 10 57 Setting: Three sub counties of Lira district in Northern Uganda. 11 58 12 Design: This was a community-based cross-sectional study nested in a cluster 13 59 14 60 randomized controlled trial. 15 61 16 62 Participants: Mother – baby pairs enrolled in a cluster randomised controlled trial. 17 63 An axillary temperature was taken during a home visit using a lithium battery- 18 64 operated digitalFor thermometer. peer review only 19 65 20 21 66 Primary and secondary outcomes: The primary outcome measure was the 22 67 prevalence of hypothermia. Hypothermia was defined as mild if the axillary 23 68 temperature was 36.0 °C - <36.5 °C, moderate if the temperature was 32.0 °C - <36.0 24 69 °C, and severe hypothermia if the temperature was < 32.0 °C. The secondary outcome 25 70 measure was the case-fatality risk of neonatal hypothermia. Predictors of moderate to 26 71 severe hypothermia were determined using a generalized estimating equation model 27 72 for the Poisson family. 28 29 73 30 74 Results: We recruited 1330 participants. The incidence of hypothermia (<36.5°C) 31 75 was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Of these, 32% (429/1330), 95%CI (29.5- 32 76 35.2)] had mild hypothermia, whereas 18.7% (249/1330), 95% CI (15.8-22.0) had 33 77 moderate hypothermia. None had severe hypothermia. At multivariable analysis, 34 78 predictors of neonatal hypothermia included: home birth [adjusted prevalence ratio, 35 79 aPR, 1.9, 95% CI (1.4-2.6)]; low birth weight [aPR 1.7, 95%CI (1.3-2.3)]; and 36 80 delayed breastfeeding initiation [aPR 1.2, 95%CI (1.0-1.5)]. The case fatality risk 37 http://bmjopen.bmj.com/ 38 81 ratio of hypothermic compared to normothermic neonates was 2.0 (95% CI 0.60-6.9). 39 82 40 83 Conclusion: The incidence of neonatal hypothermia was very high, demonstrating 41 84 that communities in tropical climates should not ignore neonatal hypothermia. 42 85 Interventions designed to address neonatal hypothermia should consider ways of 43 44 86 reaching newborns born at home, as those with low birth weight are at greater risk of

45 87 hypothermia. The promotion of early breastfeeding initiation and skin-to-skin care on September 29, 2021 by guest. Protected copyright. 46 88 could reduce the effects of neonatal hypothermia. 47 89 48 90 Article summary: 49 91 50 92 Strengths and limitations of this study; 51 52 93 53 94  This is the first purely community based assessment of neonatal hypothermia 54 95 in sub-Saharan Africa 55 96  Estimates obtained are generalizable to settings with a significant proportion 56 97 of home births unlike previous estimates from health facility based studies 57 98  We suggest that hypothermia is a significant problem, especially in cases of 58 99 home births, low birth weight, and delayed breastfeeding initiation 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 26 BMJ Open

1 2

3 100  Our choice of using a digital thermometer, placed in the axilla could have BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 101 underestimated hypothermia, but this was the most socially acceptable option 5 6 102  We did not measure some risk factors such as delivery temperature and 7 103 maternal body temperature 8 104 9 105 10 106 11 107 12 13 108 14 109 15 110 1 Introduction 16 111 17 112 Neonatal mortality (death of newborns less than 28 days) in Uganda is unacceptably 18 113 high, at 22.3 deathsFor per 1,000peer live births review compared to 1.6only deaths per 1,000 live births 19 114 in high-income countries [1]. In order to attain the global target of reducing neonatal 20 21 115 mortality to under 12 deaths per 1,000 live births by 2030 [2], there is a need to 22 116 identify and quantify the predictors of neonatal mortality; especially those that are 23 117 preventable by available low-cost interventions [3,4]. One of the predictors of 24 118 neonatal mortality that can easily be solved by available low-cost interventions is 25 119 neonatal hypothermia [5]. 26 120 27 121 Neonatal hypothermia, defined as an axillary temperature less than 36.5 °C [6,7], is 28 29 122 associated with increased neonatal morbidity and mortality [8-10]. Countries with 30 123 high neonatal mortality have high rates of neonatal hypothermia [11]. Hypothermia 31 124 mainly contributes to mortality by worsening outcomes of severe neonatal infections, 32 125 preterm birth, and birth asphyxia. It is estimated that 20% of deaths due to 33 126 prematurity and 10% of deaths in term babies could be prevented by improved 34 127 thermal care [12]. In addition, neonatal hypothermia results in reduced growth and 35 128 development [13]. 36 129 37 http://bmjopen.bmj.com/ 38 130 Newborns are unable to maintain their body temperature without thermal protection 39 131 [14]. They are susceptible to hypothermia due to physical and environmental factors. 40 132 Physical factors that predispose neonates to hypothermia include a large surface area 41 133 to volume ratio, thin skin, and low amounts of insulating fat [5,11,14,15]. 42 134 Environmental factors that predispose neonates to hypothermia include poor thermal 43 44 135 practices around the time of birth, such as keeping the neonate away from the mother

45 136 and bathing the newborn within 24 hours of birth [16], which are common practices in on September 29, 2021 by guest. Protected copyright. 46 137 sub-Saharan Africa [17,18]. The World Health Organization recommends a ten-step 47 138 warm chain to prevent neonatal hypothermia: a warm delivery room, immediate 48 139 drying, delayed bathing, skin to skin contact, early and exclusive breastfeeding, 49 140 appropriate clothing/bedding, keeping the baby with the mother, warm transportation 50 141 and resuscitation, and training/raising awareness on the dangers of hypothermia [6]. 51 52 142 However, these actions are often suboptimal in most communities in sub-Saharan 53 143 Africa [19], and disregarded with the misguided assumption that a warm climate 54 144 guarantees thermal protection to the newborns [20,21]. Newborns are at greatest risk 55 145 of hypothermia on the first day of life and this is mainly a result of evaporation of 56 146 amniotic fluid and the neonate’s limited ability to generate heat [15,22]. 57 147 58 148 Despite a significant proportion of births and deaths taking place at home in sub- 59 60 149 Saharan Africa, there is little to no data on hypothermia obtained from community

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 26

1 2

3 150 studies [5,23]. Previous estimates of hypothermia in sub-Saharan Africa have mostly BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 151 been obtained from health facility studies [9,10,20,21,24,25] and may therefore not be 5 6 152 representative of populations with poor health-seeking behaviors. Researchers 7 153 conducting community-based studies have been encouraged to incorporate axillary 8 154 temperatures with standard inexpensive digital thermometers in their study protocols 9 155 in order to enrich the literature on community estimates of neonatal hypothermia [5]. 10 156 This information is necessary when advocating for the scale-up of existing 11 157 interventions known to reduce hypothermia [23]. Therefore, in this study, we 12 determined the prevalence, predictors, and case fatality risk of hypothermia among 13 158 14 159 neonates in Lira district, Northern Uganda. 15 160 16 161 2 Materials and Methods 17 162 18 163 For peer review only 19 164 Study setting 20 21 165 This study was conducted in Lira district is a post-conflict area in Northern Uganda in 22 166 the sub-counties of Aromo, Agweng, and Ogur between January 2018 and March 23 167 2019. The trial was conducted in the sub-counties of Aromo, Agweng and Ogur, as 24 168 they had the poorest maternal and newborn health indicators in the district. About 25 169 400,000 people live in Lira; the majority live in rural areas and practice subsistence 26 170 farming [26]. In the region covering Lira district, Over 90% of pregnant women 27 171 attend at least one antenatal visit, only 66% of births take place in a health facility, 28 29 172 and approximately 29 out of every 1,000 newborns died in the first 28 days of life 30 173 [27]. During the period of this study (Jan 2018 - March 2019), the average monthly 31 174 temperatures ranged from 27.8 °C to 35.0 °C (Ngeta weather station, Lira district). 32 175 Women who give birth vaginally are discharged from health facilities within 24 hours 33 176 and those who give birth by caesarean section are discharged within 72 hours, unless 34 177 complications occur. 35 178 36 179 Study design 37 http://bmjopen.bmj.com/ 38 180 This was a cross sectional study conducted between January 2018 and March 2019. 39 181 The study was nested in a cluster randomized controlled trial designed to promote 40 182 health facility birth, newborn care practices (early and exclusive breastfeeding, skin to 41 183 skin care), and timely postnatal health facility visits (Survival Pluss study registered 42 184 on ClinicalTrial.gov as NCT02605369). 43 44 185

45 186 Study participants on September 29, 2021 by guest. Protected copyright. 46 187 All newborns born to mothers participating in the cluster randomized controlled trial 47 188 were eligible for this study. We excluded newborns whose mothers were too sick to 48 189 participate in the interview, and newborns that died before we visited. 49 190 50 191 Power and sample size 51 52 192 A total of 1330 neonates participated in our study. The participants were initially 53 193 enrolled in a cluster randomized controlled study which had a neonatal hypothermia 54 194 intra cluster correlation coefficient of 0.044, and average cluster sample size of 65, 55 195 giving us a design effect of 3.8, and effective sample size of 350, resulting in absolute 56 196 precision of 1.5% to 5.2%, i.e. the difference between the point estimate and the 95% 57 197 confidence interval (CI) for incidence values ranging from 2% to 50%. Since we were 58 198 studying a very common outcome, we deemed this precision adequate. 59 60 199

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 26 BMJ Open

1 2

3 200 Main variables BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 201 5 6 202 Outcome variable 7 203 The outcome variable in this study was hypothermia, which was defined as mild 8 204 hypothermia if the axillary temperature was between 36.0 °C and less than 36.5 °C, 9 205 moderate if the temperature was between 32.0 °C and less than 36.0 °C, and severe 10 206 hypothermia if the temperature was less than 32.0°C. We also graded hypothermia 11 207 according to a classification proposed by Mullany et al. [23]. Briefly, Mullany et al. 12 classified hypothermia as follows: Grade 1 (36.0 °C -36.5 °C), Grade 2 (35.0 °C - 13 208 14 209 <36.0 °C), Grade 3 (34.0 °C -<35.0 °C), and Grade 4 (<34.0 °C). 15 210 16 17 211 Exposure variables/risk factors 18 For peer review only 19 212 Data were collected on several risk factors during pregnancy and immediately after 20 21 213 birth. These included: maternal age, parity, maternal education, paternal education, 22 214 wealth, singleton or multiple birth, sex of the newborn, place of birth, birth weight, 23 215 early breastfeeding initiation, bathing of the newborn, and whether the baby was 24 216 placed on the mother’s chest or abdomen immediately after birth. We classified the 25 217 season as wet if the average monthly precipitation was 60 mm or more (Koppen- 26 218 Geiger climate classification) [28]. The average monthly precipitation and 27 219 temperature for the study period were obtained from the Ngeta weather station in Lira 28 29 220 district. Wealth quintiles were calculated from an asset-based index using principal 30 221 component analysis. The following assets and house characteristics were considered: 31 222 cupboard, bicycle, radio, mobile phone, motorcycle, cement floor, iron sheets, burnt 32 223 bricks, and land ownership. We defined early breastfeeding initiation as the initiation 33 224 of breastfeeding within one hour of birth. Education level was categorised into 34 225 primary, secondary and tertiary. The primary level corresponds to 1-7 years of 35 36 226 education, the secondary level to 8-13 years of education and the tertiary level to 227 more than 13 years of education. 37 http://bmjopen.bmj.com/ 38 39 228 40 41 229 Data collection 42 43 44 230 As part of the trial in which this study was nested, a team of 42 research assistants

45 231 collected data and conducted measurements on the first day of birth, or as soon as on September 29, 2021 by guest. Protected copyright. 46 232 possible after birth at the mother’s home. A temperature was taken high in the axilla 47 233 during the study visit. We used a lithium battery-operated digital thermometer: Model 48 234 TM01 (manufactured by Cotronic Manufacturing, Shenzhen). The research assistants 49 235 were trained on how to measure temperature and supervised by a team consisting of 50 51 236 three paediatricians, one obstetrician, two general practitioners, two nurses, and one 52 237 data analyst. Temperature measurements were mostly conducted before taking the 53 238 baby's anthropometric measurements, with emphasis placed on minimizing the time 54 239 the babies were exposed to the cold. Measurements involved putting the tip of the 55 240 thermometer high up in the apex of the axilla, halfway between the anterior and 56 241 posterior margins, and holding the arm in place until an automatic audible beep was 57 242 heard. Two measurement readings in degrees Celsius were taken and the average of 58 59 243 these used. Thermometers were cleaned with cotton wool soaked in 70% alcohol after 60 244 the examination.

5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 26

1 2

3 245 Recruitment and follow-up BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 246 All villages had a recruiter who was elected during the community sensitization 5 6 247 meetings of the trial. The recruiter was a female resident in the cluster. Recruiters 7 248 identified pregnant women and accompanied research assistants to the home of the 8 249 women during recruitment. They were trained during a one-day workshop, which 9 250 emphasized ethics, confidentiality, and good record keeping. Recruiters were also 10 251 given a cell phone to contact the team (site supervisor/research assistants) whenever 11 252 they identified a pregnant woman or whenever a pregnant woman had given birth. 12 They were paid Uganda Shillings 5,000 (USD 1.4) whenever they identified an 13 253 14 254 eligible participant and whenever they informed the team within 24 hours of a mother 15 255 giving birth. Approximately 250 recruiters were trained. After a recruiter informed the 16 256 team of an eligible participant, a research assistant would be sent to the participant to 17 257 ascertain eligibility, obtain consent to participate in the study and conduct the 18 258 interview. To ensureFor that peer recruiters were review reporting all pregnantonly women, we employed 19 259 community health workers (village health team members) to conduct a census of all 20 21 260 pregnant women in the area. Pregnant mothers and their relatives were also 22 261 encouraged to contact the study team immediately after giving birth. Research 23 262 assistants also obtained phone numbers of pregnant women and their relatives and 24 263 periodically conducted follow up phone calls and visits to ensure that mothers were 25 264 visited immediately after birth. The process of notification was similar between health 26 265 facility and home births. Data collectors also conducted follow up visits to assess vital 27 266 status of the babies at 7 days and 28 days. 28 29 267 30 268 Patient and Public Involvement 31 269 The public was not involved in the design and conceptualisation of the study but they 32 270 were involved in the recruitment of participants. We held community meetings in 33 271 each village during which a recruiter was elected from among the village members. 34 272 The recruiter was responsible for recruitment in their village. The results of this study 35 273 will be disseminated to the wider community through community dialogue meetings 36 274 at parish level in each participating village. 37 http://bmjopen.bmj.com/ 38 275 39 276 Statistical analysis 40 277 Data were analyzed using Stata version 14.0 (StataCorp; College Station, TX, USA). 41 278 Study characteristics were compared across the exposure status and summarized as 42 279 proportions for categorical data and means for continuous data. Hypothermia was 43 44 280 categorized using both the WHO classification [6], and a classification suggested by

45 281 Mullany et al [23], and presented as proportions with corresponding 95% confidence on September 29, 2021 by guest. Protected copyright. 46 282 intervals adjusted for clustering. Factors associated with moderate to severe 47 283 hypothermia were determined using a generalized estimating equation model for the 48 284 Poisson family, with a log link, allowing for the clustering and assuming an 49 285 exchangeable correlation. We used robust variance estimation in our model. Risk 50 286 factors included in our multivariable model were determined a priori during a review 51 52 287 of the literature on the subject. Factors included as risk factors in our model included: 53 288 mother’s age, mother’s education, mode of birth, place of birth, low birth weight, 54 289 wealth, parity, season, place baby put immediately after birth, cleaning/drying the 55 290 baby immediately after birth, bathing the baby, delayed initiation of breastfeeding. All 56 291 variables included in the model were assessed for collinearity and considered 57 292 collinear if they had a variance inflation factor greater than 10. In the case of 58 293 collinearity, we retained the variable with greater biological plausibility and/or 59 60 294 measure of association. The multivariable analyses were based on a complete case

6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 26 BMJ Open

1 2

3 295 analysis. However, we conducted sensitivity analyses of best case, worst case and BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 296 most realistic scenarios to assess the potential effect of the missing data. We also 5 6 297 conducted sub-group analysis of the prevalence of hypothermia by date of child on 7 298 examination and by place of birth. 8 299 9 300 10 301 11 302 3 Results 12 13 303 14 304 Participant characteristics 15 305 On our visits to the participants, we were able to take the temperature measurements 16 306 of 1527 neonates; of these we used the data of 1330 for whom temperatures were 17 307 taken within the first 72 hours after birth (figure 1). The mean age of mothers in the 18 308 study was 24.6 yearsFor (standard peer deviation review (sd) 6.8) and theironly median education was 5 19 309 years (inter quartile range (iqr) 3-6). The mean weight of children in the study was 3.2 20 21 310 kg (sd 0.5) (table 1). 22 311 23 312 Hypothermia 24 313 The mean temperature was 36.4 °C (sd 0.7), and the median temperature was 36.4 °C 25 314 (iqr 36.1 °C -36.8 °C). The minimum temperature recorded was 32.0 °C and the 26 315 maximum temperature recorded was 39.4 °C. The incidence of hypothermia 27 316 (temperature less than 36.5°C) was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Of these, 28 29 317 32% (429/1330), 95%CI (29.5-35.2)] had mild hypothermia (temperature 36.0 °C - 30 318 <36.5 °C), whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate hypothermia 31 319 (temperature 32.0 °C - <36.0 °C). No child had severe hypothermia (temperature less 32 320 than 32.0°C) (table 2a). We also graded hypothermia according to a classification 33 321 proposed by Mullany et al. [23] and present the results in table 2b. Sensitivity 34 322 analyses conducted suggested that we might have underestimated the burden (online 35 323 supplementary appendix table 1,2). Hypothermia was more common among home 36 324 births and on the first day of birth. Results of the third day of life were very imprecise 37 http://bmjopen.bmj.com/ 38 325 (online supplementary appendix table 3,4). 39 326 40 327 Factors associated with hypothermia 41 328 Using multivariable analysis, the factors associated with hypothermia with neonatal 42 329 hypothermia included: home birth [adjusted Risk Ratio, aPR, 1.9, 95% CI (1.4-2.6)], 43 44 330 low birth weight [aPR 1.7, 95%CI (1.3-2.3)], and delayed breastfeeding initiation

45 331 [aPR 1.2, 95%CI (1.0-1.5] as these infants (or neonates) are at greater risk (table 3). on September 29, 2021 by guest. Protected copyright. 46 332 47 333 Case fatality risk 48 334 The risk of death among newborns (first 28 days of life) with moderate hypothermia 49 335 was 3/249 (1.2%, 95%CI 0.38-3.7), compared to 6/1023 (0.59%, 95% CI 0.28-1.2) 50 336 among newborns (first 28 days of life) with normal temperature, resulting in a case 51 52 337 fatality risk ratio of 2.0 (95% CI 0.60-6.9). 53 338 54 339 55 340 56 341 57 342 58 343 59 60 344 4. Discussion

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 26

1 2

3 345 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 346 The incidence of hypothermia in this study was high. Half of the newborns developed 5 6 347 hypothermia; 33% developed mild hypothermia; 19% developed moderate 7 348 hypothermia. Similar findings were observed in a community-based study in Nepal, 8 349 where 59% of neonates developed hypothermia on the first day [29], and in another 9 350 community-based study in India where the incidence of hypothermia was 45% [30]. 10 351 However, the incidence of hypothermia observed in our study was much higher than 11 352 that observed in two other studies in India, which observed an incidence of 11% [31] 12 and 17% [32]. The difference could be explained by the different definitions of 13 353 14 354 hypothermia used in the studies. We defined hypothermia as a temperature less than 15 355 36.5 °C in accordance with recommendations from the World Health Organization 16 356 [6], whereas Kumar et al defined hypothermia as temperature less than 35.6 °C, and 17 357 Bang et al defined hypothermia as a temperature less than 35 °C. 18 358 For peer review only 19 359 Newborns who had low birth weight were more likely to be hypothermic compared to 20 21 360 newborns with normal birth weight. This finding is not surprising. Similar findings 22 361 were observed in a community-based study conducted in Nepal [33] and in many 23 362 other hospital-based studies in Uganda, Ethiopia [24,25], and other countries [20,34]. 24 363 Low birth weight infants have less capability to conserve and generate heat. This is 25 364 mainly because of physiological factors such as the reduced amount of brown fat and 26 365 a poor shivering reflex [15,35]. These thermo-protective mechanisms are needed to 27 366 maintain normal temperature in newborns who are exposed to hypothermic situations. 28 29 367 30 368 Babies born at home were more likely to be hypothermic compared to babies born in 31 369 health facilities. This finding has also been reported in other settings [35]. A study in 32 370 Uganda found that mothers who gave birth at home were more likely to practice sub- 33 371 optimal thermal care practices [36]. Mothers who give birth at home are more likely 34 372 to bathe their babies soon after birth [37,38], which could explain the increased risk of 35 373 hypothermia observed in babies born at home. The main reason for bathing newborns 36 374 early is the belief that newborns are dirty, having come into contact with maternal 37 http://bmjopen.bmj.com/ 38 375 fluids and the vernix caseosa [17,39-41]. Bathing newborns is also perceived as a 39 376 prerequisite to good neonatal rest and sleep after birth [39]. 40 377 41 378 Despite the generally impoverished nature of the study area, belonging to a relatively 42 379 lower social economic status was also a risk factor for hypothermia in this population. 43 44 380 Mothers with low socioeconomic status often lack resources to buy materials that can

45 381 keep the baby warm[42] and may have limited access to health information [43]. This on September 29, 2021 by guest. Protected copyright. 46 382 should not be a big problem if the mother practices skin to skin care. Unfortunately, 47 383 many mothers in Uganda and other countries in sub-Saharan Africa do not practice 48 384 skin-to-skin care[41,44-46]. Reasons for not practicing adequate newborn care 49 385 include beliefs that skin-to-skin care could result in the transmission of diseases to the 50 386 baby and could hurt the umbilical cord of the baby [41,45]. 51 52 387 53 388 Mothers who delayed putting their babies to the breast were more likely to have 54 389 hypothermic babies. This finding was also observed in the community-based study in 55 390 Nepal [33]. Newborns who are breastfed early receive warmth from their mothers and 56 391 this explains the reduction in hypothermia [23,47]. Mothers who had higher education 57 392 were less likely to have hypothermic babies, although this finding was imprecise. 58 393 There was also no difference between mothers in the intervention group and the 59 60

8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 26 BMJ Open

1 2

3 394 control group, meaning the intervention did not prevent the newborns from becoming BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 395 hypothermic. 5 6 396 7 397 Methodological considerations 8 398 We did not measure some risk factors such as delivery room temperature and 9 399 maternal temperature. We could also have underestimated hypothermia by using a 10 400 digital thermometer, placed in the axilla. Digital thermometers might slightly over- or 11 401 under-estimate temperature readings as compared to mercury thermometers [48-50]. 12 We used these because they are inexpensive, locally available, and easy to use by 13 402 14 403 community workers [29]. In addition, digital thermometers are easier to use in poorly 15 404 lit rural homes [29]. We used axillary measurements because they were easier to do, 16 405 safer, and more acceptable than rectal measurements [29]. In a systematic review 17 406 studying differences between rectal and axillary temperatures, the pooled mean 18 407 difference of rectalFor minus peer axillary temperature review was estimated only to be 0.17 °C, ranging 19 408 from -0.15 °C to 0.5 °C [51]. Our study could have suffered from a selection bias 20 21 409 since only 75% of eligible participants were recruited. From our sensitivity analysis, 22 410 we believe hypothermia is still a big challenge, and that this selection bias might have 23 411 slightly underestimated the burden, since it was possibly the very sick who were not 24 412 visited within 72 hours of birth. We believe this selection bias also greatly 25 413 underestimated the mortality attributed to hypothermia since many more children died 26 414 in the unmeasured group. This is understandable since the majority of newborn deaths 27 415 in the study, as would be expected, occurred in the first few hours after birth, before 28 29 416 our team were able to reach the scene. The lack of gestational age data is another 30 417 limitation in our study. We believe that our findings are generalizable to rural areas in 31 418 tropical low-income countries with similar newborn care practices. 32 419 33 420 34 421 6 Conclusion 35 422 The incidence of neonatal hypothermia was very high, demonstrating that 36 423 communities in tropical climates should not ignore neonatal hypothermia. 37 http://bmjopen.bmj.com/ 38 424 Interventions designed to address neonatal hypothermia should consider ways of 39 425 reaching newborns born at home, as these are at greater risk of hypothermia. Low 40 426 birth weight newborns, and newborns born to mothers in the poorest socioeconomic 41 427 status, should also be prioritized. We recommend low cost interventions such as skin- 42 428 to-skin care for all newborns born in similar settings to prevent neonatal hypothermia. 43 44 429

45 430 Ethics approval and consent to participate on September 29, 2021 by guest. Protected copyright. 46 431 Ethical approval to conduct the study was obtained from the following bodies: 1) 47 432 Research and Ethics committee School of Medicine, Makerere University (SOMREC: 48 433 REF 2015-121); 2) Uganda National Council of Science and Technology (UNCST: 49 434 SS 3954); 3) Regional Committees for Medical and Health Research Ethics (REK 50 435 VEST 2017/2079) and the trial was registered at ClinicalTrial.gov as NCT02605369. 51 52 436 We also obtained permission from the Ministry of Health and Lira Local Government. 53 437 Written informed consent was obtained from the respondents in the study. Research 54 438 assistants were trained in confidentiality and the right of the respondent to withdraw 55 439 their participation at any time during the study. At the community level, we obtained 56 440 permission to include clusters during community sensitization meetings, after which 57 441 the community members democratically elected recruiters, and peer buddies when 58 442 applicable, from amongst themselves. 59 60 443

9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 26

1 2

3 444 Consent for publication BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 445 Not applicable 5 6 446 7 447 Availability of data and materials 8 448 The datasets used and/or analysed during the current study are available from the 9 449 corresponding author on reasonable request. 10 450 11 451 Funding 12 Funding was obtained from the Survival Pluss project; grant number UGA-13-0030 at 13 452 14 453 Makerere University. Survival Pluss project is funded by The Norwegian Program for 15 454 Capacity Development in Higher Education and Research for Development 16 455 (NORHED) under The Norwegian Agency for Development Cooperation (NORAD). 17 456 18 457 Author contributionsFor peer review only 19 458 David Mukunya (D.M), James K. Tumwine (J.K.T), Victoria Nankabirwa (V.N), 20 21 459 Grace Ndeezi (G.N), and Thorkild Tylleskar (T.T) conceived, designed, supervised 22 460 the study, analyzed the data, and wrote the first draft of manuscript. Milton W. 23 461 Musaba (M.W.M), Josephine Tumuhamye (J.T), Justin. B. Tongun (J.B.T), Agnes 24 462 Napyo (A.N), Vivian Zalwango (V.Z), Vicentina Achora (V.A), Beatrice Odongkara 25 463 (B.O), and Agnes Anna Arach (A.A.A) were instrumental in the design and 26 464 supervision of the study, and in drafting of the manuscript. All authors read and 27 465 approved the final version to be published. 28 29 466 30 467 Conflict of interest 31 468 All authors declare no conflict of interest. 32 469 33 470 Acknowledgments 34 471 In a special way, we acknowledge the District Health Office of Lira district, and the 35 472 various district, sub-county, parish, and village leaders for their assistance in this 36 473 study. We thank the study participants for accepting to be part of the study and 37 http://bmjopen.bmj.com/ 38 474 research assistants for working tirelessly to make this work a reality. In a special way, 39 475 we acknowledge the excellent work performed by our recruiters in making this study 40 476 possible. Finally, we extend heartfelt appreciation to Ms. Jo Weeks for the excellent 41 477 English editing. 42 478 43 44 479 Figure and Title legends

45 480 on September 29, 2021 by guest. Protected copyright. 46 481 Figure 1: Study profile of neonates assessed for hypothermia in Lira district, 47 482 Northern Uganda. 48 483 Table 1: Participant characteristics of newborns assessed for hypothermia in Northern 49 484 Uganda 50 485 Table 2a: Prevalence of hypothermia in Lira district, Northern Uganda, defined by 51 52 486 the World Health Organization classification 53 487 Table 2b: Prevalence of hypothermia in Lira district, Northern Uganda, defined by 54 488 the Mullany classification 55 489 Table 2c: Sensitivity analyses assuming all unmeasured temperatures in the first three 56 490 days were normo-thermic 57 491 Table 2d: Sensitivity analyses assuming all unmeasured temperatures in the first three 58 492 days had similar distribution of hypothermia as observed, based on place of birth 59 60

10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 26 BMJ Open

1 2

3 493 Table 3: Factors associated with moderate to severe hypothermia among newborn in BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 494 Lira district Northern Uganda 5 6 495 7 496 8 497 9 498 References 10 499 11 500 1 GBD 2016 Mortality Collaborators. Global, regional, and national under-5 12 mortality, adult mortality, age-specific mortality, and life expectancy, 1970-2016: a 13 501 14 502 systematic analysis for the Global Burden of Disease Study 2016. Lancet (London, 15 503 England) 2017;390(10100):1084-150. 16 17 504 2 United Nations. Sustainable Development Goals. Secondary Sustainable 18 505 Development GoalsFor 2015. peer http://www.un.org/sustainabledevelopment/summit/ review only . 19 20 506 3 Darmstadt GL, Bhutta ZA, Cousens S, et al. Evidence-based, cost-effective 21 507 interventions: how many newborn babies can we save? Lancet (London, England) 22 23 508 2005;365(9463):977-88. 24 25 509 4 Knippenberg R, Lawn JE, Darmstadt GL, et al. Systematic scaling up of 26 510 neonatal care in countries. Lancet (London, England) 2005;365(9464):1087-98. 27 28 511 5 Lunze K, Bloom DE, Jamison DT, et al. The global burden of neonatal 29 512 hypothermia: systematic review of a major challenge for newborn survival. BMC 30 513 Medicine 2013;11:24. 31 32 514 6 World Health Organization. Thermal Protection of the Newborn: a practical 33 34 515 guide. Geneva: World Health Organization, 1997. 35 36 516 7 Lunze K, Yeboah-Antwi K, Marsh DR, et al. Prevention and management of

37 517 neonatal hypothermia in rural Zambia. PloS One 2014;9(4):e92006. http://bmjopen.bmj.com/ 38 39 518 8 Christensson K, Bhat GJ, Eriksson B, et al. The effect of routine hospital care 40 519 on the health of hypothermic newborn infants in Zambia. Journal of Tropical 41 520 Pediatrics 1995;41(4):210-4. 42 43 521 9 Kambarami R, Chidede O. Neonatal hypothermia levels and risk factors for 44 522 mortality in a tropical country. The Central African Journal of Medicine 2003;49(9- 45 on September 29, 2021 by guest. Protected copyright. 46 523 10):103-6. 47 48 524 10 Sodemann M, Nielsen J, Veirum J, et al. Hypothermia of newborns is 49 525 associated with excess mortality in the first 2 months of life in Guinea-Bissau, West 50 526 Africa. Tropical medicine & international health : TM & IH 2008;13(8):980-6. 51 52 527 11 Kumar V, Shearer JC, Kumar A, et al. Neonatal hypothermia in low resource 53 54 528 settings: a review. Journal of perinatology : official journal of the California 55 529 Perinatal Association 2009;29(6):401-12. 56 57 530 12 Bhutta ZA, Das JK, Bahl R, et al. Can available interventions end preventable 58 531 deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet (London, 59 532 England) 2014;384(9940):347-70. 60

11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 26

1 2

3 533 13 Glass L, Silverman WA, Sinclair JC. Relationship of thermal environment and BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 534 caloric intake to growth and resting metabolism in the late neonatal period. Biologia 5 6 535 Neonatorum Neo-natal studies 1969;14(5):324-40. 7 8 536 14 Lunze K, Hamer DH. Thermal protection of the newborn in resource-limited 9 537 environments. Journal of perinatology : official journal of the California Perinatal 10 538 Association 2012;32(5):317-24. 11 12 539 15 Adamson SK, Jr., Towell ME. Thermal homeostasis in the fetus and newborn. 13 540 Anesthesiology 1965;26:531-48. 14 15 16 Bergstrom A, Byaruhanga R, Okong P. The impact of newborn bathing on the 16 541 17 542 prevalence of neonatal hypothermia in Uganda: a randomized, controlled trial. Acta 18 543 Paediatrica (Oslo,For Norway peer : 1992) 2005; review94(10):1462-7. only 19 20 544 17 Waiswa P, Kemigisa M, Kiguli J, et al. Acceptability of evidence-based 21 545 neonatal care practices in rural Uganda - implications for programming. BMC 22 546 Pregnancy and Childbirth 2008;8:21. 23 24 547 18 Hill Z, Tawiah-Agyemang C, Manu A, et al. Keeping newborns warm: beliefs, 25 26 548 practices and potential for behaviour change in rural Ghana. Tropical Medicine & 27 549 International Health : TM & IH 2010;15(10):1118-24. 28 29 550 19 Coalter WS, Patterson SL. Sociocultural factors affecting uptake of home- 30 551 based neonatal thermal care practices in Africa: A qualitative review. Journal of Child 31 552 Health Care : for professionals working with children in the hospital and community 32 553 2017;21(2):132-41. 33 34 554 20 Manji KP, Kisenge R. Neonatal hypothermia on admission to a special care 35 36 555 unit in Dar-es-Salaam, Tanzania: a cause for concern. The Central African Journal of

37 556 Medicine 2003;49(3-4):23-7. http://bmjopen.bmj.com/ 38 39 557 21 Byaruhanga R, Bergstrom A, Okong P. Neonatal hypothermia in Uganda: 40 558 prevalence and risk factors. Journal of Tropical Pediatrics 2005;51(4):212-5. 41 42 559 22 Smales OR, Kime R. Thermoregulation in babies immediately after birth. 43 560 Archives of Disease in Childhood 1978;53(1):58-61. 44 45 561 23 Mullany LC. Neonatal hypothermia in low-resource settings. Seminars in on September 29, 2021 by guest. Protected copyright. 46 47 562 Perinatology 2010;34(6):426-33. 48 49 563 24 Tasew H, Gebrekristos K, Kidanu K, et al. Determinants of hypothermia on 50 564 neonates admitted to the intensive care unit of public hospitals of Central Zone, 51 565 Tigray, Ethiopia 2017: unmatched case-control study. BMC Research Notes 52 566 2018;11(1):576. 53 54 567 25 Demissie BW, Abera BB, Chichiabellu TY, et al. Neonatal hypothermia and 55 56 568 associated factors among neonates admitted to neonatal intensive care unit of public 57 569 hospitals in Addis Ababa, Ethiopia. BMC Pediatrics 2018;18(1):263. 58 59 570 26 Uganda Bureau of Statistics. The National Population and Housing Census 60 571 2014 – Area

12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 26 BMJ Open

1 2

3 572 Specific Profile Series, Kampala, Uganda. Secondary The National Population and BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 573 Housing Census 2014 – Area 5 6 574 Specific Profile Series, Kampala, Uganda 2017. 7 575 http://www.ubos.org/onlinefiles/uploads/ubos/2014CensusProfiles/MUKONO.pdf. 8 9 576 27 Uganda Bureau of Statistics (UBOS) and ICF. Uganda Demographic and 10 577 Health Survey 2016. Kampala, Uganda and Rockville, Maryland, USA: UBOS and 11 578 ICF, 2018. 12 13 579 28 Peel MC, Finlayson BL, McMahon TA. Updated world map of the Koppen- 14 Geiger climate classification. Hydrol Earth Syst Sci 2007;11:1633–44. 15 580 16 17 581 29 Mullany LC, Katz J, Khatry SK, et al. Incidence and seasonality of 18 582 hypothermia amongFor newborns peer in southern review Nepal. Archives only of Pediatrics & Adolescent 19 583 Medicine 2010;164(1):71-7. 20 21 584 30 Darmstadt GL, Kumar V, Yadav R, et al. Introduction of community-based 22 585 skin-to-skin care in rural Uttar Pradesh, India. Journal of perinatology : official 23 586 journal of the California Perinatal Association 2006;26(10):597-604. 24 25 26 587 31 Kumar R, Aggarwal AK. Body temperatures of home delivered newborns in 27 588 north India. Tropical Doctor 1998;28(3):134-6. 28 29 589 32 Bang AT, Reddy HM, Baitule SB, et al. The incidence of morbidities in a 30 590 cohort of neonates in rural Gadchiroli, India: seasonal and temporal variation and a 31 591 hypothesis about prevention. Journal of Perinatology : official Journal of the 32 592 California Perinatal Association 2005;25 Suppl 1:S18-28. 33 34 593 33 Mullany LC, Katz J, Khatry SK, et al. Neonatal hypothermia and associated 35 36 594 risk factors among newborns of southern Nepal. BMC Medicine 2010;8:43.

37 http://bmjopen.bmj.com/ 38 595 34 Zayeri F, Kazemnejad A, Ganjali M, et al. Hypothermia in Iranian newborns. 39 596 Incidence, risk factors and related complications. Saudi Medical Journal 40 597 2005;26(9):1367-71. 41 42 598 35 Onalo R. Neonatal hypothermia in sub-Saharan Africa: a review. Niger J Clin 43 599 Pract 2013;16(2):129-38. 44 45 600 36 Kabwijamu L, Waiswa P, Kawooya V, et al. Newborn care practices among on September 29, 2021 by guest. Protected copyright. 46 47 601 adolescent mothers in Hoima district, Western Uganda. PloS One 48 602 2016;11(11):e0166405. 49 50 603 37 Mrisho M, Schellenberg JA, Mushi AK, et al. Understanding home-based 51 604 neonatal care practice in rural southern Tanzania. Transactions of the Royal Society of 52 605 Tropical Medicine and Hygiene 2008;102(7):669-78. 53 54 606 38 Salasibew MM, Filteau S, Marchant T. A qualitative study exploring newborn 55 56 607 care behaviours after home births in rural Ethiopia: implications for adoption of 57 608 essential interventions for saving newborn lives. BMC Pregnancy and Childbirth 58 609 2014;14:412. 59 60

13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 26

1 2

3 610 39 Adejuyigbe EA, Bee MH, Amare Y, et al. "Why not bathe the baby today?": BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 611 A qualitative study of thermal care beliefs and practices in four African sites. BMC 5 6 612 Pediatrics 2015;15:156. 7 8 613 40 Shamba D, Schellenberg J, Hildon ZJ, et al. Thermal care for newborn babies 9 614 in rural southern Tanzania: a mixed-method study of barriers, facilitators and 10 615 potential for behaviour change. BMC Pregnancy and Childbirth 2014;14:267. 11 12 616 41 Byaruhanga RN, Nsungwa-Sabiiti J, Kiguli J, et al. Hurdles and opportunities 13 617 for newborn care in rural Uganda. Midwifery 2011;27(6):775-80. 14 15 42 Lunze K, Dawkins R, Tapia A, et al. Market mechanisms for newborn health 16 618 17 619 in Nepal. BMC Pregnancy and Childbirth 2017;17(1):428. 18 For peer review only 19 620 43 Owor MO, Matovu JKB, Murokora D, et al. Factors associated with adoption 20 621 of beneficial newborn care practices in rural Eastern Uganda: a cross-sectional study. 21 622 BMC Pregnancy and Childbirth 2016;16:83. 22 23 623 44 Bee M, Shiroor A, Hill Z. Neonatal care practices in sub-Saharan Africa: a 24 624 systematic review of quantitative and qualitative data. Journal of Health, Population, 25 26 625 and Nutrition 2018;37(1):9. 27 28 626 45 Byaruhanga RN, Bergstrom A, Tibemanya J, et al. Perceptions among post- 29 627 delivery mothers of skin-to-skin contact and newborn baby care in a periurban 30 628 hospital in Uganda. Midwifery 2008;24(2):183-9. 31 32 629 46 Waiswa P, Peterson S, Tomson G, et al. Poor newborn care practices - a 33 630 population based survey in eastern Uganda. BMC Pregnancy and Childbirth 34 631 2010;10:9. 35 36 632 47 Huffman SL, Zehner ER, Victora C. Can improvements in breast-feeding 37 http://bmjopen.bmj.com/ 38 633 practices reduce neonatal mortality in developing countries? Midwifery 39 634 2001;17(2):80-92. 40 41 635 48 Smith J. Are electronic thermometry techniques suitable alternatives to 42 636 traditional mercury in glass thermometry techniques in the paediatric setting? Journal 43 637 of Advanced Nursing 1998;28(5):1030-9. 44 45 638 49 Jones HL, Kleber CB, Eckert GJ, et al. Comparison of rectal temperature on September 29, 2021 by guest. Protected copyright. 46 47 639 measured by digital vs. mercury glass thermometer in infants under two months old. 48 640 Clinical Pediatrics 2003;42(4):357-9. 49 50 641 50 Latman NS, Hans P, Nicholson L, et al. Evaluation of clinical thermometers 51 642 for accuracy and reliability. Biomedical Instrumentation & Technology 52 643 2001;35(4):259-65. 53 54 644 51 Craig JV, Lancaster GA, Williamson PR, et al. Temperature measured at the 55 56 645 axilla compared with rectum in children and young people: systematic review. BMJ 57 646 (Clinical research ed) 2000;320(7243):1174-8. 58 647 59 648 60 649

14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Page 17 of 26 BMJ Open

1 2 3 4 5 6 7 Table 1: Participant characteristics of newborns assessed for hypothermia in Northern Uganda 8 9 † 10 All participants Late participants* Missed participants 11 No No Hypothermia Hypothermia Unknown 12 HypothermiaFor peerHypothermia review only 13 N=652 N=678 N=88 N=109 N=241 14 15 n (%) n (%) n (%) n (%) n (%) 16

None 74 (11.4) 105 (15.5) 6 (06.8) 12 (11.0) 34 (14.1) on September 29, 2021 by guest. Protected copyright. 25 Primary 513 (78.7) 519 (76.6) 73 (83.0) 85 (78.0) 190 (78.8) 26 27 Secondary 51 (7.8) 47 (6.9) 7 (8.0) 9 (8.3) 17 (7.1) 28 Tertiary 14 (02.2) 07 (01.0) 2 (02.3) 3 (02.8) - 29 Father’s 30 31 education 32 None 14 (2.2) 11 (1.6) 1 (1.1) 1 (0.92) 6 (2.5) 33 Primary 377 (57.8) 416 (61.4) 52 (59.1) 55 (50.5) 151 (62.7) 34 35 Secondary 177 (27.2) 147 (21.7) 23 (26.1) 28 (25.7) 51 (21.2) 36 Tertiary 41 (6.3) 38 (5.6) 4 (4.6) 8 (7.3) 12 (5.0) 37 Missing 43 (6.6) 66 (9.7) 8 (9.1) 17 (15.6) 21 (8.7) 38 39 40 41 42 15 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 18 of 26

1 2 3 4 5 Parity 6 <=1 286 (43.9) 325 (47.9) 44 (50.0) 55 (50.5) 101 (41.9) 7 8 2-4 219 (33.6) 218 (32.2) 36 (40.9) 28 (25.7) 86 (35.7) 9 >4 147 (22.6) 135 (19.9) 8 (9.1) 26 (23.9) 54 (22.4) 10 Place of birth 11 12 Home 157 (24.1) For254 (37.5) peer26 (29.6) review40 (36.7) only100 (41.5) 13 Health facility 495 (75.9) 424 (62.5) 62 (70.5) 69 (63.3) 141 (58.5) 14 Caesarean

15 section 16 17 No 641 (98.3) 670 (98.8) 79 (89.8) 94 (86.2) 232 (96.3) http://bmjopen.bmj.com/ 18 Yes 11(1.7) 8 (1.2) 9 (10.2) 15 (13.8) 9 (3.7) 19 20 Marital status 21 Married 609 (93.4) 612 (90.3) 80 (90.9) 92 (84.4) 220 (91.3) 22 Single 43 (6.6) 66 (9.7) 8 (9.1) 17 (15.6) 21 (8.7) 23 24 Electricity 25 Yes 71 (10.9) 86 (12.7) 4 (4.6) 6 (5.5) 24 (10.0) on September 29, 2021 by guest. Protected copyright. 26 No 581 (89.1) 592 (87.3) 84 (95.5) 103 (94.5) 217 (90.0) 27 28 Presence of 29 mobile phone in 30 the household 31 Yes 346 (53.1) 363 (53.5) 42 (47.7) 53 (48.6) 159 (66.0) 32 33 No 306 (46.9) 315 (46.5) 46 (52.3) 56 (51.4) 82 (34.0) 34 Source of

35 drinking water 36 37 Borehole 319 (48.9) 340 (50.2) 54 (61.4) 58 (53.2) 138 (57.3) 38 Tap/piped water 88 (13.5) 84 (12.4) 9 (10.2) 10 (09.2) 20 (8.3) 39 40 41 42 16 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 19 of 26 BMJ Open

15 weight 16 17 No 613 (94.0) 622 (91.7) 83 (94.3) 101 (92.7) 15 (6.2) http://bmjopen.bmj.com/ 18 Yes 35 (5.4) 45 (6.6) 4 (4.6) 6 (5.5) 1 (0.41) 19 Missing 4 (0.6) 11 (01.6) 1 (1.1) 2 (1.8) 225 (93.4) 20 21 Wealth quintiles 22 1 (Poorest) 146 (22.4) 140 (20.7) 19 (21.6) 23 (21.1) 35 (14.5) 23 2 143 (21.9) 185 (27.3) 20 (22.7) 22 (20.2) 63 (26.1) 24 25 3 123 (18.9) 121 (17.9) 19 (21.6) 18 (16.5) 45 (18.7) on September 29, 2021 by guest. Protected copyright. 26 4 105 (16.1) 114 (16.8) 10 (11.4) 20 (18.4) 49 (20.3) 27 5 (Richest) 135 (20.7) 118 (17.4) 20 (22.7) 26 (23.9) 49 (20.3) 28 29 Season 30 Wet 589 (90.3) 579 (85.4) 74 (84.1) 87 (79.8) 47 (19.5) 31 Dry 63 (9.7) 99 (14.6) 14 (15.9) 22 (20.2) 194 (80.5) 32 33 Place baby 34 immediately 35 after birth 36 Mother chest or 37 547 (83.9) 548 (80.8) 68 (77.3) 76 (69.7) 163 (67.6) 38 abdomen 39 40 41 42 17 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 20 of 26

14 before visit 15 16 No 326 (50.0) 274 (40.4) 1 (1.1) 2 (1.8) 81 (34.3) 17 Yes 326 (50.0) 404 (59.6) 87 (98.9) 107 (98.2) 155 (65.7) http://bmjopen.bmj.com/ 18 Died in first 19 20 month 21 No 643 (98.6) 675 (99.6) 88 (100.0) 108 (99.1) 227 (94.2) 22 Yes 9 (1.4) 3 (0.44) 0 (0.0) 1 (0.92) 14 (5.8) 23 24 Early 25 breastfeeding on September 29, 2021 by guest. Protected copyright. 26 initiation 27 28 No 208 (31.9) 257 (37.9) 35 (39.8) 58 (53.2) 110 (48.0) 29 30 Yes 444 (68.1) 421 (62.1) 53 (60.2) 51 (46.8) 119 (52.0) 31 32 *Participants whose temperature was measured after 3 days 33 34 Missed participants†: Eligible participants whose temperature was not measured 35 36 37 38 39 40 41 42 18 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 21 of 26 BMJ Open

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 26

1 2

3 Table 3: Factors associated with moderate to severe hypothermia among newborn in BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 Lira district Northern Uganda 5 6 7 8 Bi-variable Multivariable 9 N=1330 RR N=1315 10 11 cPR (95% C.I) aPR (95% C.I) 12 13 Intervention group 14 Control 1 1 15 16 Intervention 0.85 (0.62-1.2) 1.0 (0.79-1.4) 17 Age of mother 18 For peer review only 19 <=19 1 1 20 20-30 0.71 (0.58-0.88) 0.81 (0.59-1.1) 21 22 >30 0.70 (0.50-0.96) 0.75 (0.43-1.3) 23 Mother’s education 24 25 None 1 1 26 Primary 0.93 (0.69-1.2) 0.94 (0.70-1.3) 27 28 >=Secondary 0.53 (0.31-0.88) 0.63 (0.39-1.0) 29 Father’s education 30 31 None 1 32 Primary 1.2 (0.58-2.6) 33 34 Secondary 0.81 (0.35-1.9) - 35 Tertiary 0.73 (0.27-2.0) 36

37 Parity http://bmjopen.bmj.com/ 38 <=1 1 1 39 40 2-4 0.75 (0.57-0.99) 0.85 (0.57-1.3) 41 >4 0.77 (0.55-1.1) 0.84 (0.50-1.4) 42 43 Place of birth 44 Health Facility 1 1 45 on September 29, 2021 by guest. Protected copyright. 46 Home 2.0 (1.5-2.6) 1.9 (1.4-2.6) 47 48 Caesarean section 49 No 1 1 50 0.94 (0.44-2.0) 0.82 (0.31-2.1) 51 Yes 52 Marital status 53 Single 1 - 54 55 Married 0.77 (0.55-1.1) 56 Low birth weight* 57 58 (less than 2.5) 59 No 1 1 60

20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 26 BMJ Open

1 2

3 Yes 1.9 (1.4-2.6) 1.7 (1.3-2.3) BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 Wealth quintiles 6 1 (Poorest) 1 1 7 8 2 1.1 (0.82-1.6) 1.3 (0.91-1.7) 9 3 0.81 (0.57-1.1) 0.93 (0.67-1.3) 10 11 4 0.71 (0.46-1.1) 0.87 (0.59-1.3) 12 5 (Richest) 0.59 (0.40-0.87) 0.79 (0.53-1.2) 13 14 Season 15 Wet 1 1 16 17 Dry 1.3 (0.92-1.8) 1.4 (1.0-1.9) 18 Baby For peer review only 19 20 placed on mother’s 21 chest or abdomen 22 immediately after 23 birth 24 25 No 1 1 26 27 Yes 0.78 (0.61-0.99) 0.98 (0.76-1.3) 28 29 Clean and dry 30 baby immediately 31 32 No 1 1 33 Yes 0.87 (0.59-1.3) 0.96 (0.65-1.4) 34 35 Bathed baby before 36 visit

37 http://bmjopen.bmj.com/ 38 No 1 1 39 Yes 1.2 (0.98-1.5) 1.0 (0.81-1.2) 40 41 Breastfeeding 42 initiation 43 44 Early 1 1

45 Late 1.4 (1.1-1.8) 1.2 (1.0-1.5) on September 29, 2021 by guest. Protected copyright. 46 47 Child’s sex 48 Male 1 49 50 Female 1.1 (0.95-1.3) - 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 24 of 26

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 Women enrolled: 1877 6 7 8 9 Lost to follow up: 43 10 Death: 1 11 12 13 14 Followed up to birth: 1833 15 16 17 Still births: 37 18 For peer review onlyChild death during birth: 26 19 Mum died: 2 20 21 22 23 Eligible for hypothermia measurement: 1768 24 25 26 27 Lost to follow up: 241 28 Accessed after 72 hours: 197 29 30 31 Accessed for hypothermia within 72 hours after 32 birth: 1330 33 34 35 36 37 Figure 1: Study profile of neonates assessed for hypothermia in Lira district, Northern http://bmjopen.bmj.com/ 38 Uganda. 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Page 25 of 26 BMJ Open

1 2 3 4 5 Table 1: Sensitivity analyses assuming all unmeasured temperatures in the first three days were normothermic 6 7 Hypothermia Best Case Scenario* Worst case scenario** 8 n/N % 9 10 Mild (36.0-36.5) 429/1768 24.3 (22.3-26.3) 867/1768 49.0 (46.7-51.4) 11 Moderate (32.0-35.9) 249/1768 14.1 (12.5-15.8) 687/1768 38.9 (36.6-41.2) 12 Severe (<32.0) 0/1768 For 0 peer review 0/1768 0 only 13 Any hypothermia 678/1768 38.3 (36.1- 40.7) 1554/1768 87.9 (86.3-89.4) 14 *: Assuming all unmeasured temperatures were normothermic 15 **: Assuming all unmeasured temperatures were hypothermic 16 17 Confidence intervals calculated by the exact method http://bmjopen.bmj.com/ 18 19 Table 2: Sensitivity analyses assuming all unmeasured temperatures in the first three days had similar distribution of hypothermia as observed, 20 based on place of birth 21 22 Hypothermia n/N % 23 24 Any hypothermia in 678/1330 51.0 (46.9-55.1) 25 measured infants on September 29, 2021 by guest. Protected copyright. 26 Any hypothermia in un 228/438 34.2 (30.6-38.0) 27 measured infants 28 29 Please note: 272 of the unmeasured infants were delivered at a health facility (prevalence of hypothermia for health facility births is 46.1% in 30 31 first 3 days) and 32 166 of the unmeasured infants were delivered at home (prevalence of hypothermia for home births is 61.8% in first 3 days). Confidence intervals 33 of unobserved calculated by the exact method. 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 26 of 26

1 2 3 4 5 Table 3: Prevalence of hypothermia in Lira district, Northern Uganda, stratified by the age of neonate on the day of examination 6 7 Day of examination since birth 8 All participants Day 1 (24-h) Day 2 (48-h) Day 3 (72-h) 9 10 Hypothermia n/N [% (95% CI)] n/N [% (95% CI)] n/N [% (95% CI)] n/N [% (95% CI)] 11 Mild (36.0-36.5) 429/1330 [32.3 (29.5-35.2)] 322/1049 [30.7 (27.8-33.8)] 76/207 [36.7 (29.1-45.1)] 31/74 [41.9 (30.0-54.9)] 12 Moderate (32.0-35.9) 249/1330 [18.7For (15.8-22.0)] peer 208/1049 [19.8review (16.8-23.3)] 24/207 only [11.6 (7.8-16.9)] 17/74 [23.0 (15.7-32.3)] 13 Severe (<32.0) 0 0 0 0 14 Any 678/1330 [51.0 (46.9-55.1)] 530/1049 [50.5 (46.2-54.9)] 100/207 [48.3 (39.7-57.1)] 48/74 [64.9 (50.6-76.9)] 15 16 17 http://bmjopen.bmj.com/ 18 Table 4: Prevalence of hypothermia in Lira district, Northern Uganda, stratified by the place of birth 19 20 Place of birth 21 All participants Health facility Home 22 23 Hypothermia n/N [% (95% CI)] n/N [% (95% CI)] n/N [% (95% CI)] 24 Mild (36.0-36.5) 429/1330 [32.3 (29.5-35.2)] 289/919 [31.5 (27.9-35.2)] 140/411 [34.1 (29.1-39.4)] 25 Moderate (32.0-35.9) 249/1330 [18.7 (15.8-22.0)] 135/919 [14.7 (11.3-18.9)] 114/411 [27.7 (23.0-33.1)] on September 29, 2021 by guest. Protected copyright. 26 Severe (<32.0) 0 0 0 27 Any 678/1330 [51.0 (46.9-55.1)] 424/919 [46.1 (41.0-51.4)] 254/411 [61.8 (56.0-67.3)] 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 27 of 26 BMJ Open

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 26

1 (b) Report category boundaries when continuous variables were categorized 2 (c) If relevant, consider translating estimates of relative risk into absolute risk for NA

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Neonatal hypothermia in Northern Uganda: a community- based cross-sectional study

Journal: BMJ Open ManuscriptFor ID peerbmjopen-2020-041723.R2 review only Article Type: Original research

Date Submitted by the 12-Jan-2021 Author:

Complete List of Authors: Mukunya, David ; University of Bergen Department of Medicine, Center for Intervention Science in Maternal and Child Health, Center for International Health; Busitema University Tumwine, James; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Nankabirwa, Victoria; Makerere University College of Health Sciences, Department of Epidemiology and Biostatistics; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Odongkara, Beatrice; Gulu University, Department of Paediatrics and Child Health Tongun, Justin; University of Juba, Department of Paediatrics and Child Health Arach, Agnes; Lira University, Department of Nursing and Midwifery http://bmjopen.bmj.com/ Tumuhamye, Josephine; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Napyo, Agnes; Busitema University, Department of Public Health Zalwango, Vivian; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Achora, Vicentina; Gulu University, Department of Obstetrics and Gynaecology Musaba, Milton; Busitema University, Department of Obstetrics and Gynaecology Ndeezi, Grace; Makerere University College of Health Sciences, on September 29, 2021 by guest. Protected copyright. Department of Paediatrics and Child Health Tylleskär, Thorkild; Universitetet i Bergen, Centre for International health

Primary Subject Public health Heading:

Secondary Subject Heading: Paediatrics

Epidemiology < TROPICAL MEDICINE, PAEDIATRICS, Public health < Keywords: INFECTIOUS DISEASES

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 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

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 3 of 29 BMJ Open

1 2

3 1 Original research BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 2 5 6 3 Neonatal hypothermia in Northern Uganda: a community-based cross-sectional 7 4 study 8 5 9 6 David Mukunya1,3,12*, James K. Tumwine4, Victoria Nankabirwa2,3,5, Beatrice 10 7 Odongkara7, Justin B. Tongun8, Agnes A. Arach9, Josephine Tumuhamye3, Agnes 11 8 Napyo12, Vivian Zalwango4, Vicentina Achora10, Milton W. Musaba11, Grace 12 Ndeezi4, Thorkild Tylleskar6 13 9 14 10 15 11 1Sanyu Africa Research Institute, Mbale, Uganda 16 12 17 13 2Department of Epidemiology and Biostatistics, School of Public Health, Makerere 18 14 University CollegeFor of Health peer Sciences, review Kampala, Uganda only 19 15 20 3 21 16 Centre for Intervention Science in Maternal and Child Health (CISMAC), Centre for 22 17 International health, University of Bergen. Bergen, Norway 23 18 24 19 4Department of Paediatrics and Child Health, School of Medicine, Makerere 25 20 University College of Health Sciences, Kampala, Uganda 26 21 27 22 5Department of Epidemiology and Biostatistics, School of Public Health, Makerere 28 29 23 University College of Health Sciences, Kampala, Uganda 30 24 31 25 7Department of Paediatrics and Child Health, Gulu University, Gulu, Uganda 32 26 33 27 8Department of Paediatrics and Child Health, Juba University, Juba, Uganda 34 28 35 29 9Department of Nursing and Midwifery, Lira University, Lira, Uganda 36 30 37 http://bmjopen.bmj.com/ 38 31 10Department of Obstetrics and Gynaecology, Gulu University, Gulu, Uganda 39 32 40 33 11Department of Obstetrics and Gynaecology, Busitema University Faculty of Health 41 34 Sciences, Mbale, Uganda 42 35 43 12 44 36 Department of Public Health, Busitema University Faculty of Health Sciences,

45 37 Mbale, Uganda on September 29, 2021 by guest. Protected copyright. 46 38 47 39 48 40 *Corresponding author 49 41 David Mukunya, P.O Box 2190, Mbale, Uganda 50 42 Email; [email protected], Mob; +256775152316 51 52 43 53 44 Key words: newborn care, neonatal care, kangaroo mother care 54 45 55 46 Word count: 3339 56 47 57 48 58 49 59 60 50

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 29

1 2

3 51 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 52 Abstract 5 6 53 7 54 8 55 Objective: To determine the prevalence, predictors, and case fatality risk of 9 56 hypothermia among neonates in Lira district, Northern Uganda. 10 57 11 58 Setting: Three sub counties of Lira district in Northern Uganda. 12 13 59 14 60 Design: This was a community-based cross-sectional study nested in a cluster 15 61 randomized controlled trial. 16 62 17 63 Participants: Mother – baby pairs enrolled in a cluster randomised controlled trial. 18 64 An axillary temperatureFor waspeer taken during review a home visit onlyusing a lithium battery- 19 65 operated digital thermometer. 20 21 66 22 67 Primary and secondary outcomes: The primary outcome measure was the 23 68 prevalence of hypothermia. Hypothermia was defined as mild if the axillary 24 69 temperature was 36.0 °C - <36.5 °C, moderate if the temperature was 32.0 °C - <36.0 25 70 °C, and severe hypothermia if the temperature was < 32.0 °C. The secondary outcome 26 71 measure was the case-fatality risk of neonatal hypothermia. Predictors of moderate to 27 72 severe hypothermia were determined using a generalized estimating equation model 28 29 73 for the Poisson family. 30 74 31 75 Results: We recruited 1330 neonates. The prevalence of hypothermia (<36.5°C) was 32 76 678/1330 [51.0%: 95% CI (46.9-55.1)]. Overall, 32% (429/1330), 95%CI (29.5-35.2)] 33 77 had mild hypothermia, whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate 34 78 hypothermia. None had severe hypothermia. At multivariable analysis, predictors of 35 79 neonatal hypothermia included: home birth [adjusted prevalence ratio, aPR, 1.9, 95% 36 80 CI (1.4-2.6)]; low birth weight [aPR 1.7, 95%CI (1.3-2.3)]; and delayed breastfeeding 37 http://bmjopen.bmj.com/ 38 81 initiation [aPR 1.2, 95%CI (1.0-1.5)]. The case fatality risk ratio of hypothermic 39 82 compared to normothermic neonates was 2.0 (95% CI 0.60-6.9). 40 83 41 84 Conclusion: The prevalence of neonatal hypothermia was very high, demonstrating 42 85 that communities in tropical climates should not ignore neonatal hypothermia. 43 44 86 Interventions designed to address neonatal hypothermia should consider ways of

45 87 reaching neonates born at home and those with low birth weight. The promotion of on September 29, 2021 by guest. Protected copyright. 46 88 early breastfeeding initiation and skin-to-skin care could reduce the effects of 47 89 neonatal hypothermia. 48 90 49 91 50 92 51 52 93 53 94 Article summary: 54 95 55 96 Strengths and limitations of this study; 56 97 57 98  This is the first purely community based assessment of neonatal hypothermia 58 99 in sub-Saharan Africa 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 29 BMJ Open

1 2

3 100  Estimates obtained are generalizable to settings with a significant proportion BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 101 of home births unlike previous estimates from health facility based studies 5 6 102  We included a large number of neonates (1330) which increased the precision 7 103 of our estimates 8 104  The choice of a digital thermometer, placed in the axilla could have 9 105 underestimated hypothermia, but this was the most socially acceptable option 10 106  We did not measure some predictors such as delivery room temperature and 11 107 maternal body temperature 12 13 108 14 109 15 110 1 Introduction 16 111 17 112 Neonatal mortality (death of neonates less than 28 days) in Uganda is unacceptably 18 113 high, at 22.3 deathsFor per 1,000peer live births review compared to 1.6only deaths per 1,000 live births 19 114 in high-income countries [1]. In order to attain the global target of reducing neonatal 20 21 115 mortality to under 12 deaths per 1,000 live births by 2030 [2], there is a need to 22 116 identify and quantify the predictors of neonatal mortality; especially those that are 23 117 preventable by available low-cost interventions [3,4]. One of the predictors of 24 118 neonatal mortality that can easily be solved by available low-cost interventions is 25 119 neonatal hypothermia [5]. 26 120 27 121 Neonatal hypothermia, defined as an axillary temperature less than 36.5 °C [6,7], is 28 29 122 associated with increased neonatal morbidity and mortality [8-10]. Countries with 30 123 high neonatal mortality have high rates of neonatal hypothermia [11]. Hypothermia 31 124 mainly contributes to mortality by worsening outcomes of severe neonatal infections, 32 125 preterm birth, and birth asphyxia [5,6,11]. It is estimated that 20% of deaths due to 33 126 prematurity and 10% of deaths in term babies could be prevented by improved 34 127 thermal care [12]. In addition, neonatal hypothermia results in reduced growth and 35 128 development [13]. 36 129 37 http://bmjopen.bmj.com/ 38 130 Neonates are unable to maintain their body temperature without thermal protection 39 131 [14]. They are susceptible to hypothermia due to physical and environmental factors. 40 132 Physical factors that predispose neonates to hypothermia include a large surface area 41 133 to volume ratio, thin skin, and low amounts of insulating fat [5,11,14,15]. 42 134 Environmental factors that predispose neonates to hypothermia include poor thermal 43 44 135 practices around the time of birth, such as keeping the neonate away from the mother

45 136 and bathing the newborn within 24 hours of birth [16], which are common practices in on September 29, 2021 by guest. Protected copyright. 46 137 sub-Saharan Africa [17,18]. The World Health Organization recommends a ten-step 47 138 warm chain to prevent neonatal hypothermia: a warm delivery room, immediate 48 139 drying, delayed bathing, skin to skin contact, early and exclusive breastfeeding, 49 140 appropriate clothing/bedding, keeping the baby with the mother, warm transportation 50 141 and resuscitation, and training/raising awareness on the dangers of hypothermia [6]. 51 52 142 However, these actions are often suboptimal in most communities in sub-Saharan 53 143 Africa [19], and disregarded with the misguided assumption that a warm climate 54 144 guarantees thermal protection to the neonates [20,21]. Neonates are at greatest risk of 55 145 hypothermia on the first day of life and this is mainly a result of evaporation of 56 146 amniotic fluid and the neonate’s limited ability to generate heat [15,22]. 57 147 58 148 Despite a significant proportion of births and deaths taking place at home in sub- 59 60 149 Saharan Africa, there is little to no data on hypothermia obtained from community

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 29

1 2

3 150 studies [5,23]. Previous estimates of hypothermia in sub-Saharan Africa have mostly BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 151 been obtained from health facility studies [9,10,20,21,24,25] and may therefore not be 5 6 152 representative of populations with poor health-seeking behaviors. Researchers 7 153 conducting community-based studies have been encouraged to incorporate axillary 8 154 temperatures with standard inexpensive digital thermometers in their study protocols 9 155 in order to enrich the literature on community estimates of neonatal hypothermia [5]. 10 156 This information is necessary when advocating for the scale-up of existing 11 157 interventions known to reduce hypothermia [23]. Therefore, in this study, we 12 determined the prevalence, predictors, and case fatality risk of hypothermia among 13 158 14 159 neonates in Lira district, Northern Uganda. 15 160 16 161 2 Materials and Methods 17 162 18 163 For peer review only 19 164 Study setting 20 21 165 This study was conducted in Lira district, located in Lango region a post-conflict area 22 166 in Northern Uganda, in the sub-counties of Aromo, Agweng, and Ogur between 23 167 January 2018 and March 2019. About 400,000 people live in Lira; the majority live in 24 168 rural areas and practice subsistence farming [26]. In Lango region, 97% of pregnant 25 169 women attend at least one antenatal care visit from a skilled provider, only 66% of 26 170 births take place in a health facility, and approximately 29 out of every 1,000 27 171 neonates died in the first 28 days of life [27]. During the period of this study, the 28 29 172 average monthly temperatures ranged from 27.8 °C to 35.0 °C (Ngeta weather station, 30 173 Lira district). Women who give birth vaginally are discharged from health facilities 31 174 within 24 hours and those who give birth by caesarean section are discharged within 32 175 72 hours, unless complications occur. 33 176 34 177 Study design 35 178 This was a cross sectional study conducted between January 2018 and March 2019. 36 179 The study was nested in a cluster randomized controlled trial designed to promote 37 http://bmjopen.bmj.com/ 38 180 health facility birth, newborn care practices (early and exclusive breastfeeding, skin to 39 181 skin care), and timely postnatal health facility visits (Survival Pluss study registered 40 182 on ClinicalTrial.gov as NCT02605369). 41 183 42 184 Study participants 43 44 185 All neonates born to mothers participating in the cluster randomized controlled trial

45 186 were eligible for this study. We excluded neonates whose mothers were too sick to on September 29, 2021 by guest. Protected copyright. 46 187 participate in the interview, and neonates that died before we visited. 47 188 48 189 Power and sample size 49 190 A total of 1330 neonates participated in our study. The participants were initially 50 191 enrolled in a cluster randomized controlled study which had a neonatal hypothermia 51 52 192 intra cluster correlation coefficient of 0.044, and average cluster sample size of 65, 53 193 giving us a design effect of 3.8, and effective sample size of 350, resulting in absolute 54 194 precision of 1.5% to 5.2%, i.e. the difference between the point estimate and the 95% 55 195 confidence interval (CI) for prevalence values ranging from 2% to 50%. Since we 56 196 were studying a very common outcome, we deemed this precision adequate. 57 197 58 198 Main variables 59 60 199

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 29 BMJ Open

1 2

3 200 Outcome variable BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 201 The outcome variable in this study was hypothermia, which was defined as mild 5 6 202 hypothermia if the axillary temperature was between 36.0 °C and less than 36.5 °C, 7 203 moderate if the temperature was between 32.0 °C and less than 36.0 °C, and severe 8 204 hypothermia if the temperature was less than 32.0°C. We also graded hypothermia 9 205 according to a classification proposed by Mullany et al. [23]. Briefly, Mullany et al. 10 206 classified hypothermia as follows: Grade 1 (36.0 °C -36.5 °C), Grade 2 (35.0 °C - 11 207 <36.0 °C), Grade 3 (34.0 °C -<35.0 °C), and Grade 4 (<34.0 °C). 12 13 208 14 15 209 Exposure variables 16 17 210 Data were collected on several predictors during pregnancy and immediately after 18 211 birth. These included:For maternal peer age, parity,review maternal education, only paternal education, 19 212 wealth, singleton or multiple birth, sex of the newborn, place of birth, birth weight, 20 21 213 early breastfeeding initiation, bathing of the newborn, and whether the baby was 22 214 placed on the mother’s chest or abdomen immediately after birth. We classified the 23 215 season as wet if the average monthly precipitation was 60 mm or more (Koppen- 24 216 Geiger climate classification) [28]. The average monthly precipitation and 25 217 temperature for the study period were obtained from the Ngeta weather station in Lira 26 218 district. Wealth quintiles were calculated from an asset-based index using principal 27 219 component analysis. The following assets and house characteristics were considered: 28 29 220 cupboard, bicycle, radio, mobile phone, motorcycle, cement floor, iron sheets, burnt 30 221 bricks, and land ownership. We defined early breastfeeding initiation as the initiation 31 222 of breastfeeding within one hour of birth. Education level was categorised into 32 223 primary, secondary and tertiary. The primary level corresponds to 1-7 years of 33 224 education, the secondary level to 8-13 years of education and the tertiary level to 34 225 more than 13 years of education. 35 36

37 226 http://bmjopen.bmj.com/ 38 39 227 Data collection 40 41 228 As part of the trial in which this study was nested, a team of 42 research assistants 42 229 collected data and conducted measurements on the first day of birth, or as soon as 43 44 230 possible after birth at the mother’s home. A temperature was taken high in the axilla

45 231 during the study visit. We used a lithium battery-operated digital thermometer: Model on September 29, 2021 by guest. Protected copyright. 46 232 TM01 (manufactured by Cotronic Manufacturing, Shenzhen). The research assistants 47 233 were trained on how to measure temperature and supervised by a team consisting of 48 234 three paediatricians, one obstetrician, two general practitioners, two nurses, and one 49 235 data analyst. Temperature measurements were mostly conducted before taking the 50 51 236 baby's anthropometric measurements, with emphasis placed on minimizing the time 52 237 the babies were exposed to the cold. Measurements involved putting the tip of the 53 238 thermometer high up in the apex of the axilla, halfway between the anterior and 54 239 posterior margins, and holding the arm in place until an automatic audible beep was 55 240 heard. Two measurement readings in degrees Celsius were taken and the average of 56 241 these used. Thermometers were cleaned with cotton wool soaked in 70% alcohol after 57 242 the examination. 58 59 60 243 Recruitment and follow-up

5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 29

1 2

3 244 All villages had a recruiter who was elected during the community sensitization BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 245 meetings of the trial. The recruiter was a female resident in the cluster. Recruiters 5 6 246 identified pregnant women and accompanied research assistants to the home of the 7 247 women during recruitment. They were trained during a one-day workshop, which 8 248 emphasized ethics, confidentiality, and good record keeping. Recruiters were also 9 249 given a cell phone to contact the team (site supervisor/research assistants) whenever 10 250 they identified a pregnant woman or whenever a pregnant woman had given birth. 11 251 They were paid Uganda Shillings 5,000 (USD 1.4) whenever they identified an 12 eligible participant and whenever they informed the team within 24 hours of a mother 13 252 14 253 giving birth. Approximately 250 recruiters were trained. After a recruiter informed the 15 254 team of an eligible participant, a research assistant visited the mother to ascertain 16 255 eligibility, to obtain informed consent and to conduct the interview. To ensure that 17 256 recruiters were reporting all pregnant women, we employed community health 18 257 workers (villageFor health teampeer members) review to conduct a census only of all pregnant women in 19 258 the area. Pregnant mothers and their relatives were encouraged to contact the study 20 21 259 team immediately after giving birth. Research assistants also obtained phone numbers 22 260 of pregnant women and their relatives and periodically conducted follow up phone 23 261 calls and visits to ensure that mothers were visited immediately after birth. The 24 262 process of notification was similar between health facility and home births. Data 25 263 collectors conducted follow up visits to assess whether the neonates were alive at 7 26 264 days and at 28 days. 27 265 28 29 266 Patient and Public Involvement 30 267 The public was not involved in the design and conceptualisation of the study but they 31 268 were involved in the recruitment of participants. We held community meetings in 32 269 each village during which a recruiter was elected from among the village members. 33 270 The recruiter was responsible for recruitment in their village. The results of this study 34 271 will be disseminated to the wider community through community dialogue meetings 35 272 at parish level in each participating village. 36 273 37 http://bmjopen.bmj.com/ 38 274 Statistical analysis 39 275 Data were analyzed using Stata version 14.0 (StataCorp; College Station, TX, USA). 40 276 Study characteristics were compared across the exposure status and summarized as 41 277 proportions for categorical data and means for continuous data. Hypothermia was 42 278 categorized using both the WHO classification [6], and a classification suggested by 43 44 279 Mullany et al [23], and presented as proportions with corresponding 95% confidence

45 280 intervals adjusted for clustering. Factors associated with moderate to severe on September 29, 2021 by guest. Protected copyright. 46 281 hypothermia were determined using a generalized estimating equation model for the 47 282 Poisson family, with a log link, allowing for the clustering and assuming an 48 283 exchangeable correlation. We used robust variance estimation in our model. 49 284 Predictors of hypothermia included in our multivariable model were determined a 50 285 priori during a review of the literature on the subject. Factors included as predictors in 51 52 286 our model included: mother’s age, mother’s education, mode of birth, place of birth, 53 287 low birth weight, wealth, parity, season, baby placed on mother’s chest or abdomen 54 288 immediately after birth, cleaning/drying the baby immediately after birth, bathing the 55 289 baby, delayed initiation of breastfeeding [5,6,11,16,20-23]. All variables included in 56 290 the model were assessed for collinearity and considered collinear if they had a 57 291 variance inflation factor greater than 10. In the case of collinearity, we retained the 58 292 variable with greater biological plausibility and/or measure of association. The 59 60 293 multivariable analyses were based on a complete case analysis. However, we

6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 29 BMJ Open

1 2

3 294 conducted sensitivity analyses of best case, worst case and most realistic scenarios to BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 295 assess the potential effect of the missing data. We also conducted sub-group analysis 5 6 296 of the prevalence of hypothermia by date of neonate on examination and by place of 7 297 birth. Since this study was nested in a cluster randomized controlled trial, the trial arm 8 298 was added as a fixed effect in all the models. 9 299 10 300 11 301 12 3 Results 13 302 14 303 15 304 Participant characteristics 16 305 On our visits to the mothers, we were able to take the temperature measurements of 17 306 1527 neonates; of these we used the data of 1330 for whom temperatures were taken 18 307 within the first 72For hours afterpeer birth (figure review 1). The mean only age of mothers was 24.6 19 308 years (standard deviation (sd) 6.8) and their median education was 5 years (inter 20 21 309 quartile range (iqr) 3-6). The mean weight of neonates was 3.2 kg (sd 0.5) (table 1). 22 310 23 311 Hypothermia 24 312 The mean temperature was 36.4 °C (sd 0.7), and the median temperature was 36.4 °C 25 313 (iqr 36.1 °C -36.8 °C). The minimum temperature recorded was 32.0 °C and the 26 314 maximum temperature recorded was 39.4 °C. The prevalence of hypothermia 27 315 (temperature less than 36.5°C) was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Overall, 28 29 316 32% (429/1330), 95%CI (29.5-35.2)] had mild hypothermia (temperature 36.0 °C - 30 317 <36.5 °C), whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate hypothermia 31 318 (temperature 32.0 °C - <36.0 °C). No neonate had severe hypothermia (temperature 32 319 less than 32.0°C) (table 2a). We also graded hypothermia according to a classification 33 320 proposed by Mullany et al. [23] and present the results in table 2b. Sensitivity 34 321 analyses conducted suggested that we might have underestimated the burden (online 35 322 supplementary appendix table 1,2). Hypothermia was more common among home 36 323 births and on the first day of birth. Results of the third day of life were very imprecise 37 http://bmjopen.bmj.com/ 38 324 (online supplementary appendix table 3,4). 39 325 40 326 Factors associated with hypothermia 41 327 Using multivariable analysis, the factors associated with neonatal hypothermia 42 328 included: home birth [adjusted prevalence ratio, aPR, 1.9, 95% CI (1.4-2.6)], low 43 44 329 birth weight [aPR 1.7, 95%CI (1.3-2.3)], and delayed breastfeeding initiation [aPR

45 330 1.2, 95%CI (1.0-1.5] (table 3). on September 29, 2021 by guest. Protected copyright. 46 331 47 332 Case fatality risk 48 333 The risk of death among neonates with moderate hypothermia was 3/249 (1.2%, 49 334 95%CI 0.38-3.7), compared to 6/1023 (0.59%, 95% CI 0.28-1.2) among neonates 50 335 with normal temperature, resulting in a case fatality risk ratio of 2.0 (95% CI 0.60- 51 52 336 6.9). 53 337 54 338 55 339 56 340 57 341 58 342 59 60 343 4. Discussion

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 29

1 2

3 344 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 345 The prevalence of hypothermia in this study was high. Half of the neonates 5 6 346 developed hypothermia; 33% developed mild hypothermia; 19% developed moderate 7 347 hypothermia. Similar findings were observed in a community-based study in Nepal, 8 348 where 59% of neonates developed hypothermia on the first day [29], and in another 9 349 community-based study in India where the prevalence of hypothermia was 45% [30]. 10 350 However, the prevalence of hypothermia observed in our study was much higher than 11 351 that observed in two other studies in India, which observed a prevalence of 11% [31] 12 and 17% [32]. The difference could be explained by the different definitions of 13 352 14 353 hypothermia used in the studies. We defined hypothermia as a temperature less than 15 354 36.5 °C in accordance with recommendations from the World Health Organization 16 355 [6], whereas Kumar et al defined hypothermia as a temperature less than 35.6 °C, and 17 356 Bang et al defined hypothermia as a temperature less than 35 °C. 18 357 For peer review only 19 358 Neonates who had low birth weight were more likely to be hypothermic compared to 20 21 359 neonates with normal birth weight. This finding is not surprising. Similar findings 22 360 were observed in a community-based study conducted in Nepal [33] and in many 23 361 other hospital-based studies in Uganda, Ethiopia [24,25], and other countries [20,34]. 24 362 Low birth weight neonates have less capability to conserve and generate heat. This is 25 363 mainly because of physiological factors such as the reduced amount of brown fat and 26 364 a poor shivering reflex [15,35]. These thermo-protective mechanisms are needed to 27 365 maintain a normal temperature in neonates who are exposed to hypothermic 28 29 366 situations. In addition to the previously reported practice of bathing the neonates soon 30 367 after birth [17], we observed neonates wrapped in thin pieces of clothing and left in 31 368 the houses as mothers carried out their daily duties. Promotion of skin-to-skin care 32 369 could reduce both hypothermia and neonatal mortality in this community [12]. 33 370 34 371 Neonates born at home were more likely to be hypothermic compared to neonates 35 372 born in health facilities. This finding has also been reported in other settings [35]. A 36 373 study in Uganda found that mothers who gave birth at home were more likely to 37 http://bmjopen.bmj.com/ 38 374 practice sub-optimal thermal care practices [36]. Mothers who give birth at home are 39 375 more likely to bathe their neonates soon after birth [37,38], which could explain the 40 376 increased risk of hypothermia observed in neonates born at home. The main reason 41 377 for bathing neonates early is the belief that neonates are dirty, having come into 42 378 contact with maternal fluids and the vernix caseosa [17,39-41]. Bathing neonates is 43 44 379 also perceived as a prerequisite to good rest and sleep [39]. However, early bathing

45 380 has been shown to result in a substantial drop in the neonate’s temperatures [16]. We on September 29, 2021 by guest. Protected copyright. 46 381 recommend that neonates are not bathed within the first 24 hours after birth [14] and 47 382 that bathing be done with warm water, after which the neonate should be placed on 48 383 the skin of the caregiver or placed in adequate warm clothing if available. Mothers of 49 384 low socioeconomic status also have increased odds of home deliveries, poor access to 50 385 health care, low birth preparedness, and complication readiness beside the inability to 51 52 386 buy materials and limited information. 53 387 54 388 Despite the generally impoverished nature of the study area, belonging to a relatively 55 389 lower social-economic status was also a predictor of hypothermia in this population. 56 390 Mothers with low socioeconomic status often lack resources to buy materials that can 57 391 keep the neonate warm[42] and may have limited access to health information [43]. 58 392 This should not be a big problem if the mother practices skin-to-skin care. 59 60 393 Unfortunately, many mothers in Uganda, Ethiopia, Ghana, Tanzania and Mali do not

8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 29 BMJ Open

1 2

3 394 practice skin-to-skin care[41,44-46]. Reasons for not practicing adequate neonatal BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 395 care include beliefs that skin-to-skin care could result in the transmission of diseases 5 6 396 to the neonate and could hurt the umbilical cord of the neonate [41,45]. 7 397 8 398 Mothers who delayed putting their neonates to the breast were more likely to have 9 399 hypothermic babies. This finding was also observed in the community-based study in 10 400 Nepal [33]. Neonates who are breastfed early receive warmth from their mothers and 11 401 this explains the reduction in hypothermia [23,47]. Mothers who had higher education 12 were less likely to have hypothermic babies, although this finding was imprecise. 13 402 14 403 There was also no difference between mothers in the intervention group and the 15 404 control group, meaning the intervention did not prevent the neonate from becoming 16 405 hypothermic. 17 406 18 407 MethodologicalFor considerations peer review only 19 408 We did not measure some predictors such as delivery room temperature and maternal 20 21 409 temperature. We could also have underestimated hypothermia by using a digital 22 410 thermometer, placed in the axilla. Digital thermometers might slightly over or 23 411 underestimate temperature readings as compared to mercury thermometers [48-50]. 24 412 We used these because they are inexpensive, locally available, and easy to use by 25 413 community workers [29]. In addition, digital thermometers are easier to use in poorly 26 414 lit rural homes [29]. We used axillary measurements because they were easier to do, 27 415 safer, and more acceptable than rectal measurements [29]. In a systematic review 28 29 416 studying differences between rectal and axillary temperatures, the pooled mean 30 417 difference of rectal minus axillary temperature was estimated to be 0.17 °C, ranging 31 418 from -0.15 °C to 0.5 °C [51]. Our study could have suffered from a selection bias 32 419 since only 75% of eligible participants were recruited. From our sensitivity analysis, 33 420 we believe hypothermia is still a big challenge, and that this selection bias might have 34 421 slightly underestimated the burden since it was possibly the very sick who were not 35 422 visited within 72 hours of birth. We believe this selection bias also greatly 36 423 underestimated the mortality attributed to hypothermia since many more neonates 37 http://bmjopen.bmj.com/ 38 424 died in the unmeasured group. This is understandable since the majority of newborn 39 425 deaths in the study, as would be expected, occurred in the first few hours after birth 40 426 before our teams were able to reach the scene. The lack of gestational age data is 41 427 another limitation in our study. We believe that our findings are generalizable to rural 42 428 areas in tropical low-income countries with similar newborn care practices. 43 44 429

45 430 on September 29, 2021 by guest. Protected copyright. 46 431 6 Conclusion 47 432 The prevalence of neonatal hypothermia was very high, demonstrating that 48 433 communities in tropical climates should not ignore neonatal hypothermia. 49 434 Interventions designed to address neonatal hypothermia should consider ways of 50 435 reaching neonates born at home, as these are at greater risk of hypothermia. Low birth 51 52 436 weight neonates, and neonates born to mothers in the poorest socioeconomic status, 53 437 should also be prioritized. We recommend low-cost interventions such as skin-to-skin 54 438 care for all neonates born in similar settings to prevent neonatal hypothermia. 55 439 56 440 Ethics approval and consent to participate 57 441 Ethical approval to conduct the study was obtained from the following bodies: 1) 58 442 Research and Ethics committee School of Medicine, Makerere University (SOMREC: 59 60 443 REF 2015-121); 2) Uganda National Council of Science and Technology (UNCST:

9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 29

1 2

3 444 SS 3954); 3) Regional Committees for Medical and Health Research Ethics (REK BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 445 VEST 2017/2079) and the trial was registered at ClinicalTrial.gov as NCT02605369. 5 6 446 We also obtained permission from the Ministry of Health and Lira Local Government. 7 447 Written informed consent was obtained from the respondents in the study. Research 8 448 assistants were trained in confidentiality and the right of the respondent to withdraw 9 449 their participation at any time during the study. At the community level, we obtained 10 450 permission to include clusters during community sensitization meetings, after which 11 451 the community members democratically elected recruiters, and peer buddies when 12 applicable, from amongst themselves. 13 452 14 453 15 454 Consent for publication 16 455 Not applicable 17 456 18 457 Availability of Fordata and peer materials review only 19 458 The datasets used and/or analysed during the current study are available from the 20 21 459 corresponding author on reasonable request. 22 460 23 461 Funding 24 462 Funding was obtained from the Survival Pluss project; grant number UGA-13-0030 at 25 463 Makerere University. Survival Pluss project is funded by The Norwegian Program for 26 464 Capacity Development in Higher Education and Research for Development 27 465 (NORHED) under The Norwegian Agency for Development Cooperation (NORAD). 28 29 466 30 467 Author contributions 31 468 David Mukunya (D.M), James K. Tumwine (J.K.T), Victoria Nankabirwa (V.N), 32 469 Grace Ndeezi (G.N), and Thorkild Tylleskar (T.T) conceived, designed, supervised 33 470 the study, analyzed the data, and wrote the first draft of manuscript. Milton W. 34 471 Musaba (M.W.M), Josephine Tumuhamye (J.T), Justin. B. Tongun (J.B.T), Agnes 35 472 Napyo (A.N), Vivian Zalwango (V.Z), Vicentina Achora (V.A), Beatrice Odongkara 36 473 (B.O), and Agnes Anna Arach (A.A.A) were instrumental in the design and 37 http://bmjopen.bmj.com/ 38 474 supervision of the study, and in drafting of the manuscript. All authors read and 39 475 approved the final version to be published. 40 476 41 477 Conflict of interest 42 478 All authors declare no conflict of interest. 43 44 479

45 480 Acknowledgments on September 29, 2021 by guest. Protected copyright. 46 481 In a special way, we acknowledge the District Health Office of Lira district, and the 47 482 various district, sub-county, parish, and village leaders for their assistance in this 48 483 study. We thank the study participants for accepting to be part of the study and 49 484 research assistants for working tirelessly to make this work a reality. In a special way, 50 485 we acknowledge the excellent work performed by our recruiters in making this study 51 52 486 possible. Finally, we extend heartfelt appreciation to Ms. Jo Weeks for the excellent 53 487 English editing. 54 488 55 489 Figure and Title legends 56 490 57 491 Figure 1: Study profile of neonates assessed for hypothermia in Lira district, 58 492 Northern Uganda 59 60

10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 29 BMJ Open

1 2

3 493 Table 1: Participant characteristics of neonates assessed for hypothermia in Northern BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 494 Uganda 5 6 495 Table 2a: Prevalence of hypothermia (defined by the World Health Organization 7 496 classification) in Lira district, Northern Uganda 8 497 Table 2b: Prevalence of hypothermia (defined by the Mullany classification) in Lira 9 498 district, Northern Uganda 10 499 Table 3: Factors associated with moderate to severe hypothermia among neonates in 11 500 Lira district Northern Uganda 12 13 501 14 502 15 503 16 504 References 17 505 18 506 1 GBD 2016For Mortality peer Collaborators. review Global, regional, only and national under-5 19 507 mortality, adult mortality, age-specific mortality, and life expectancy, 1970-2016: a 20 21 508 systematic analysis for the Global Burden of Disease Study 2016. Lancet (London, 22 509 England) 2017;390(10100):1084-150. 23 24 510 2 United Nations. Sustainable Development Goals. Secondary Sustainable 25 511 Development Goals 2015. http://www.un.org/sustainabledevelopment/summit/. 26 27 512 3 Darmstadt GL, Bhutta ZA, Cousens S, et al. Evidence-based, cost-effective 28 513 interventions: how many newborn babies can we save? Lancet (London, England) 29 514 2005;365(9463):977-88. 30 31 32 515 4 Knippenberg R, Lawn JE, Darmstadt GL, et al. Systematic scaling up of 33 516 neonatal care in countries. Lancet (London, England) 2005;365(9464):1087-98. 34 35 517 5 Lunze K, Bloom DE, Jamison DT, et al. The global burden of neonatal 36 518 hypothermia: systematic review of a major challenge for newborn survival. BMC

37 519 Medicine 2013;11:24. http://bmjopen.bmj.com/ 38 39 520 6 World Health Organization. Thermal Protection of the Newborn: a practical 40 521 guide. Geneva: World Health Organization, 1997. 41 42 43 522 7 Lunze K, Yeboah-Antwi K, Marsh DR, et al. Prevention and management of 44 523 neonatal hypothermia in rural Zambia. PloS One 2014;9(4):e92006.

45 on September 29, 2021 by guest. Protected copyright. 46 524 8 Christensson K, Bhat GJ, Eriksson B, et al. The effect of routine hospital care 47 525 on the health of hypothermic newborn infants in Zambia. Journal of Tropical 48 526 Pediatrics 1995;41(4):210-4. 49 50 527 9 Kambarami R, Chidede O. Neonatal hypothermia levels and risk factors for 51 52 528 mortality in a tropical country. The Central African Journal of Medicine 2003;49(9- 53 529 10):103-6. 54 55 530 10 Sodemann M, Nielsen J, Veirum J, et al. Hypothermia of newborns is 56 531 associated with excess mortality in the first 2 months of life in Guinea-Bissau, West 57 532 Africa. Tropical medicine & international health : TM & IH 2008;13(8):980-6. 58 59 60

11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 29

1 2

3 533 11 Kumar V, Shearer JC, Kumar A, et al. Neonatal hypothermia in low resource BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 534 settings: a review. Journal of perinatology : official journal of the California 5 6 535 Perinatal Association 2009;29(6):401-12. 7 8 536 12 Bhutta ZA, Das JK, Bahl R, et al. Can available interventions end preventable 9 537 deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet (London, 10 538 England) 2014;384(9940):347-70. 11 12 539 13 Glass L, Silverman WA, Sinclair JC. Relationship of thermal environment and 13 540 caloric intake to growth and resting metabolism in the late neonatal period. Biologia 14 Neonatorum Neo-natal studies 1969;14(5):324-40. 15 541 16 17 542 14 Lunze K, Hamer DH. Thermal protection of the newborn in resource-limited 18 543 environments. JournalFor of peer perinatology review : official journal only of the California Perinatal 19 544 Association 2012;32(5):317-24. 20 21 545 15 Adamson SK, Jr., Towell ME. Thermal homeostasis in the fetus and newborn. 22 546 Anesthesiology 1965;26:531-48. 23 24 547 16 Bergstrom A, Byaruhanga R, Okong P. The impact of newborn bathing on the 25 26 548 prevalence of neonatal hypothermia in Uganda: a randomized, controlled trial. Acta 27 549 Paediatrica (Oslo, Norway : 1992) 2005;94(10):1462-7. 28 29 550 17 Waiswa P, Kemigisa M, Kiguli J, et al. Acceptability of evidence-based 30 551 neonatal care practices in rural Uganda - implications for programming. BMC 31 552 Pregnancy and Childbirth 2008;8:21. 32 33 553 18 Hill Z, Tawiah-Agyemang C, Manu A, et al. Keeping newborns warm: beliefs, 34 554 practices and potential for behaviour change in rural Ghana. Tropical Medicine & 35 36 555 International Health : TM & IH 2010;15(10):1118-24.

37 http://bmjopen.bmj.com/ 38 556 19 Coalter WS, Patterson SL. Sociocultural factors affecting uptake of home- 39 557 based neonatal thermal care practices in Africa: A qualitative review. Journal of Child 40 558 Health Care : for professionals working with children in the hospital and community 41 559 2017;21(2):132-41. 42 43 560 20 Manji KP, Kisenge R. Neonatal hypothermia on admission to a special care 44 561 unit in Dar-es-Salaam, Tanzania: a cause for concern. The Central African Journal of 45 on September 29, 2021 by guest. Protected copyright. 46 562 Medicine 2003;49(3-4):23-7. 47 48 563 21 Byaruhanga R, Bergstrom A, Okong P. Neonatal hypothermia in Uganda: 49 564 prevalence and risk factors. Journal of Tropical Pediatrics 2005;51(4):212-5. 50 51 565 22 Smales OR, Kime R. Thermoregulation in babies immediately after birth. 52 566 Archives of Disease in Childhood 1978;53(1):58-61. 53 54 567 23 Mullany LC. Neonatal hypothermia in low-resource settings. Seminars in 55 56 568 Perinatology 2010;34(6):426-33. 57 58 569 24 Tasew H, Gebrekristos K, Kidanu K, et al. Determinants of hypothermia on 59 570 neonates admitted to the intensive care unit of public hospitals of Central Zone, 60

12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 29 BMJ Open

1 2

3 571 Tigray, Ethiopia 2017: unmatched case-control study. BMC Research Notes BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 572 2018;11(1):576. 5 6 7 573 25 Demissie BW, Abera BB, Chichiabellu TY, et al. Neonatal hypothermia and 8 574 associated factors among neonates admitted to neonatal intensive care unit of public 9 575 hospitals in Addis Ababa, Ethiopia. BMC Pediatrics 2018;18(1):263. 10 11 576 26 Uganda Bureau of Statistics. The National Population and Housing Census 12 577 2014 – Area 13 578 Specific Profile Series, Kampala, Uganda. Secondary The National Population and 14 Housing Census 2014 – Area 15 579 16 580 Specific Profile Series, Kampala, Uganda 2017. 17 581 http://www.ubos.org/onlinefiles/uploads/ubos/2014CensusProfiles/MUKONO.pdf. 18 For peer review only 19 582 27 Uganda Bureau of Statistics (UBOS) and ICF. Uganda Demographic and 20 583 Health Survey 2016. Kampala, Uganda and Rockville, Maryland, USA: UBOS and 21 584 ICF, 2018. 22 23 585 28 Peel MC, Finlayson BL, McMahon TA. Updated world map of the Koppen- 24 25 586 Geiger climate classification. Hydrol Earth Syst Sci 2007;11:1633–44. 26 27 587 29 Mullany LC, Katz J, Khatry SK, et al. Incidence and seasonality of 28 588 hypothermia among newborns in southern Nepal. Archives of Pediatrics & Adolescent 29 589 Medicine 2010;164(1):71-7. 30 31 590 30 Darmstadt GL, Kumar V, Yadav R, et al. Introduction of community-based 32 591 skin-to-skin care in rural Uttar Pradesh, India. Journal of perinatology : official 33 592 journal of the California Perinatal Association 2006;26(10):597-604. 34 35 36 593 31 Kumar R, Aggarwal AK. Body temperatures of home delivered newborns in

37 594 north India. Tropical Doctor 1998;28(3):134-6. http://bmjopen.bmj.com/ 38 39 595 32 Bang AT, Reddy HM, Baitule SB, et al. The incidence of morbidities in a 40 596 cohort of neonates in rural Gadchiroli, India: seasonal and temporal variation and a 41 597 hypothesis about prevention. Journal of Perinatology : official Journal of the 42 598 California Perinatal Association 2005;25 Suppl 1:S18-28. 43 44 599 33 Mullany LC, Katz J, Khatry SK, et al. Neonatal hypothermia and associated 45 on September 29, 2021 by guest. Protected copyright. 46 600 risk factors among newborns of southern Nepal. BMC Medicine 2010;8:43. 47 48 601 34 Zayeri F, Kazemnejad A, Ganjali M, et al. Hypothermia in Iranian newborns. 49 602 Incidence, risk factors and related complications. Saudi Medical Journal 50 603 2005;26(9):1367-71. 51 52 604 35 Onalo R. Neonatal hypothermia in sub-Saharan Africa: a review. Niger J Clin 53 605 Pract 2013;16(2):129-38. 54 55 56 606 36 Kabwijamu L, Waiswa P, Kawooya V, et al. Newborn care practices among 57 607 adolescent mothers in Hoima district, Western Uganda. PloS One 58 608 2016;11(11):e0166405. 59 60

13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 29

1 2

3 609 37 Mrisho M, Schellenberg JA, Mushi AK, et al. Understanding home-based BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 610 neonatal care practice in rural southern Tanzania. Transactions of the Royal Society of 5 6 611 Tropical Medicine and Hygiene 2008;102(7):669-78. 7 8 612 38 Salasibew MM, Filteau S, Marchant T. A qualitative study exploring newborn 9 613 care behaviours after home births in rural Ethiopia: implications for adoption of 10 614 essential interventions for saving newborn lives. BMC Pregnancy and Childbirth 11 615 2014;14:412. 12 13 616 39 Adejuyigbe EA, Bee MH, Amare Y, et al. "Why not bathe the baby today?": 14 A qualitative study of thermal care beliefs and practices in four African sites. BMC 15 617 16 618 Pediatrics 2015;15:156. 17 18 619 40 Shamba ForD, Schellenberg peer J, Hildon review ZJ, et al. Thermal only care for newborn babies 19 620 in rural southern Tanzania: a mixed-method study of barriers, facilitators and 20 621 potential for behaviour change. BMC Pregnancy and Childbirth 2014;14:267. 21 22 622 41 Byaruhanga RN, Nsungwa-Sabiiti J, Kiguli J, et al. Hurdles and opportunities 23 623 for newborn care in rural Uganda. Midwifery 2011;27(6):775-80. 24 25 26 624 42 Lunze K, Dawkins R, Tapia A, et al. Market mechanisms for newborn health 27 625 in Nepal. BMC Pregnancy and Childbirth 2017;17(1):428. 28 29 626 43 Owor MO, Matovu JKB, Murokora D, et al. Factors associated with adoption 30 627 of beneficial newborn care practices in rural Eastern Uganda: a cross-sectional study. 31 628 BMC Pregnancy and Childbirth 2016;16:83. 32 33 629 44 Bee M, Shiroor A, Hill Z. Neonatal care practices in sub-Saharan Africa: a 34 630 systematic review of quantitative and qualitative data. Journal of Health, Population, 35 36 631 and Nutrition 2018;37(1):9.

37 http://bmjopen.bmj.com/ 38 632 45 Byaruhanga RN, Bergstrom A, Tibemanya J, et al. Perceptions among post- 39 633 delivery mothers of skin-to-skin contact and newborn baby care in a periurban 40 634 hospital in Uganda. Midwifery 2008;24(2):183-9. 41 42 635 46 Waiswa P, Peterson S, Tomson G, et al. Poor newborn care practices - a 43 636 population based survey in eastern Uganda. BMC Pregnancy and Childbirth 44 637 2010;10:9. 45 on September 29, 2021 by guest. Protected copyright. 46 47 638 47 Huffman SL, Zehner ER, Victora C. Can improvements in breast-feeding 48 639 practices reduce neonatal mortality in developing countries? Midwifery 49 640 2001;17(2):80-92. 50 51 641 48 Smith J. Are electronic thermometry techniques suitable alternatives to 52 642 traditional mercury in glass thermometry techniques in the paediatric setting? Journal 53 643 of Advanced Nursing 1998;28(5):1030-9. 54 55 56 644 49 Jones HL, Kleber CB, Eckert GJ, et al. Comparison of rectal temperature 57 645 measured by digital vs. mercury glass thermometer in infants under two months old. 58 646 Clinical Pediatrics 2003;42(4):357-9. 59 60

14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 29 BMJ Open

1 2

3 647 50 Latman NS, Hans P, Nicholson L, et al. Evaluation of clinical thermometers BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 648 for accuracy and reliability. Biomedical Instrumentation & Technology 5 6 649 2001;35(4):259-65. 7 8 650 51 Craig JV, Lancaster GA, Williamson PR, et al. Temperature measured at the 9 651 axilla compared with rectum in children and young people: systematic review. BMJ 10 652 (Clinical research ed) 2000;320(7243):1174-8. 11 653 12 654 13 14 655 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 18 of 29

1 2 3 4 5 6 7 Table 1: Participant characteristics of neonates assessed for hypothermia in Northern Uganda 8 9 † 10 All participants Late participants* Missed participants 11 No No Hypothermia Hypothermia Unknown 12 HypothermiaFor peerHypothermia review only 13 N=652 N=678 N=88 N=109 N=241 14 15 n (%) n (%) n (%) n (%) n (%) 16

None 74 (11.4) 105 (15.5) 6 (06.8) 12 (11.0) 34 (14.1) on September 29, 2021 by guest. Protected copyright. 25 Primary 513 (78.7) 519 (76.6) 73 (83.0) 85 (78.0) 190 (78.8) 26 27 Secondary 51 (7.8) 47 (6.9) 7 (8.0) 9 (8.3) 17 (7.1) 28 Tertiary 14 (02.2) 07 (01.0) 2 (02.3) 3 (02.8) - 29 Father’s 30 31 education 32 None 14 (2.2) 11 (1.6) 1 (1.1) 1 (0.92) 6 (2.5) 33 Primary 377 (57.8) 416 (61.4) 52 (59.1) 55 (50.5) 151 (62.7) 34 35 Secondary 177 (27.2) 147 (21.7) 23 (26.1) 28 (25.7) 51 (21.2) 36 Tertiary 41 (6.3) 38 (5.6) 4 (4.6) 8 (7.3) 12 (5.0) 37 Missing 43 (6.6) 66 (9.7) 8 (9.1) 17 (15.6) 21 (8.7) 38 39 40 41 42 16 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 19 of 29 BMJ Open

35 drinking water 36 37 Borehole 319 (48.9) 340 (50.2) 54 (61.4) 58 (53.2) 138 (57.3) 38 Tap/piped water 88 (13.5) 84 (12.4) 9 (10.2) 10 (09.2) 20 (8.3) 39 40 41 42 17 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 20 of 29

15 weight 16 17 No 613 (94.0) 622 (91.7) 83 (94.3) 101 (92.7) 15 (6.2) http://bmjopen.bmj.com/ 18 Yes 35 (5.4) 45 (6.6) 4 (4.6) 6 (5.5) 1 (0.41) 19 Missing 4 (0.6) 11 (01.6) 1 (1.1) 2 (1.8) 225 (93.4) 20 21 Wealth quintiles 22 1 (Poorest) 146 (22.4) 140 (20.7) 19 (21.6) 23 (21.1) 35 (14.5) 23 2 143 (21.9) 185 (27.3) 20 (22.7) 22 (20.2) 63 (26.1) 24 25 3 123 (18.9) 121 (17.9) 19 (21.6) 18 (16.5) 45 (18.7) on September 29, 2021 by guest. Protected copyright. 26 4 105 (16.1) 114 (16.8) 10 (11.4) 20 (18.4) 49 (20.3) 27 5 (Richest) 135 (20.7) 118 (17.4) 20 (22.7) 26 (23.9) 49 (20.3) 28 29 Season 30 Wet 589 (90.3) 579 (85.4) 74 (84.1) 87 (79.8) 47 (19.5) 31 Dry 63 (9.7) 99 (14.6) 14 (15.9) 22 (20.2) 194 (80.5) 32 33 34 35 36 37 38 39 40 41 42 18 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 21 of 29 BMJ Open

1 2 3 4 5 6 Baby 7 placed on

8 mother’s chest 9 10 or abdomen 11 immediately 12 after birth For peer review only 13 14 15 Yes 547 (83.9) 548 (80.8) 68 (77.3) 76 (69.7) 163 (67.6) 16 17 No 105 (16.1) 130 (19.2) 20 (22.7) 33 (30.3) 78 (32.4) http://bmjopen.bmj.com/ 18 19 Clean and dry 20 baby 21 immediately 22 23 No 68 (10.4) 104 (15.3) 10 (11.4) 21 (19.3) 41 (17.0) 24 Yes 584 (89.6) 574 (84.7) 78 (88.6) 88 (80.7) 200 (83.0) on September 29, 2021 by guest. Protected copyright. 25 Bathed baby

26 before visit 27 28 No 326 (50.0) 274 (40.4) 1 (1.1) 2 (1.8) 81 (34.3) 29 Yes 326 (50.0) 404 (59.6) 87 (98.9) 107 (98.2) 155 (65.7) 30 31 Died in first 32 month 33 No 643 (98.6) 675 (99.6) 88 (100.0) 108 (99.1) 227 (94.2) 34 Yes 9 (1.4) 3 (0.44) 0 (0.0) 1 (0.92) 14 (5.8) 35 36 37 38 39 40 41 42 19 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 22 of 29

1 2 3 4 5 Early 6 breastfeeding 7 initiation 8 9 No 208 (31.9) 257 (37.9) 35 (39.8) 58 (53.2) 110 (48.0) 10 11 Yes 444 (68.1) 421 (62.1) 53 (60.2) 51 (46.8) 119 (52.0) 12 For peer review only 13 *Participants whose temperature was measured after 3 days 14 15 Missed participants†: Eligible participants whose temperature was not measured 16 17 http://bmjopen.bmj.com/ 18 19 20 21 22 23 24 25 on September 29, 2021 by guest. Protected copyright. 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 20 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 23 of 29 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 Table 2a: Prevalence of hypothermia (defined by the World Health Organization 7 classification) in Lira district, Northern Uganda 8 9 Hypothermia n/N (all) % (95% CI) 10 Mild (36.0-36.5) 429/1330 32.3 (29.5-35.2) 11 Moderate (32.0-35.9) 249/1330 18.7 (15.8-22.0) 12 13 Severe (<32.0) 0/1330 0 14 Any 678/1330 51.0 (46.9-55.1) 15 16 17 Table 2b: Prevalence of hypothermia (defined by the Mullany classification) in Lira 18 district, NorthernFor Uganda peer review only 19 20 21 Hypothermia n/N (all) % (95% CI) 22 Grade 1 (36.0-36.5) 429/1330 32.3 (29.5-35.2) 23 Grade 2 (35.0-35.99) 218/1330 16.4 (14.0-19.1) 24 Grade 3 (34.0-34.99) 26/1330 2.0 (1.2-3.1) 25 26 Grade 4 (less than 34.0) 5/1330 0.38 (0.16-0.90) 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 24 of 29

1 2

3 Table 3: Factors associated with moderate to severe hypothermia among neonates in BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 Lira district Northern Uganda 5 6 7 8 Bi-variable Multivariable 9 N=1330 RR N=1315 10 11 cPR (95% C.I) aPR (95% C.I) 12 13 Trial arm 14 Control 1 1 15 16 Intervention 0.85 (0.62-1.2) 1.0 (0.79-1.4) 17 Age of mother 18 For peer review only 19 <=19 1 1 20 20-30 0.71 (0.58-0.88) 0.81 (0.59-1.1) 21 22 >30 0.70 (0.50-0.96) 0.75 (0.43-1.3) 23 Mother’s education 24 25 None 1 1 26 Primary 0.93 (0.69-1.2) 0.94 (0.70-1.3) 27 28 >=Secondary 0.53 (0.31-0.88) 0.63 (0.39-1.0) 29 Father’s education 30 31 None 1 32 Primary 1.2 (0.58-2.6) 33 34 Secondary 0.81 (0.35-1.9) - 35 Tertiary 0.73 (0.27-2.0) 36

22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 29 BMJ Open

1 2

37 http://bmjopen.bmj.com/ 38 No 1 1 39 Yes 1.2 (0.98-1.5) 1.0 (0.81-1.2) 40 41 Breastfeeding 42 initiation 43 44 Early 1 1

45 Late 1.4 (1.1-1.8) 1.2 (1.0-1.5) on September 29, 2021 by guest. Protected copyright. 46 47 Child’s sex 48 Male 1 49 50 Female 1.1 (0.95-1.3) - 51 52 cPR: crude prevalence ratio 53 aPR: adjusted prevalence ratio 54 55 56 57 58 59 60

23 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 29

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 27 of 29 BMJ Open

1 2

45 on September 29, 2021 by guest. Protected copyright. 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: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 28 of 29

Page 29 of 29 BMJ Open

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 29 BMJ Open

1 (b) Report category boundaries when continuous variables were categorized 2 (c) If relevant, consider translating estimates of relative risk into absolute risk for NA

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

Neonatal hypothermia in Northern Uganda: a community- based cross-sectional study

Journal: BMJ Open ManuscriptFor ID peerbmjopen-2020-041723.R3 review only Article Type: Original research

Date Submitted by the 23-Jan-2021 Author:

Complete List of Authors: Mukunya, David ; University of Bergen Department of Medicine, Center for Intervention Science in Maternal and Child Health, Center for International Health; Busitema University Tumwine, James; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Nankabirwa, Victoria; Makerere University College of Health Sciences, Department of Epidemiology and Biostatistics; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Odongkara, Beatrice; Gulu University, Department of Paediatrics and Child Health Tongun, Justin; University of Juba, Department of Paediatrics and Child Health Arach, Agnes; Lira University, Department of Nursing and Midwifery http://bmjopen.bmj.com/ Tumuhamye, Josephine; Universitet i Bergen, Center for Intervention Science in Maternal and Child Health, Center for International Health Napyo, Agnes; Busitema University, Department of Public Health Zalwango, Vivian; Makerere University College of Health Sciences, Department of Paediatrics and Child Health Achora, Vicentina; Gulu University, Department of Obstetrics and Gynaecology Musaba, Milton; Busitema University, Department of Obstetrics and Gynaecology Ndeezi, Grace; Makerere University College of Health Sciences, on September 29, 2021 by guest. Protected copyright. Department of Paediatrics and Child Health Tylleskär, Thorkild; Universitetet i Bergen, Centre for International health

Primary Subject Public health Heading:

Secondary Subject Heading: Paediatrics

Epidemiology < TROPICAL MEDICINE, PAEDIATRICS, Public health < Keywords: INFECTIOUS DISEASES

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 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

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 3 of 29 BMJ Open

1 2

3 1 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 2 Original research 5 6 3 7 4 Neonatal hypothermia in Northern Uganda: a community-based cross-sectional 8 5 study 9 6 10 7 David Mukunya1,3,12*, James K. Tumwine4, Victoria Nankabirwa2,3,5, Beatrice 11 8 Odongkara7, Justin B. Tongun8, Agnes A. Arach9, Josephine Tumuhamye3, Agnes 12 Napyo12, Vivian Zalwango4, Vicentina Achora10, Milton W. Musaba11, Grace 13 9 4 6 14 10 Ndeezi , Thorkild Tylleskar 15 11 16 12 1Sanyu Africa Research Institute, Mbale, Uganda 17 13 18 14 2Department of ForEpidemiology peer and Biostatistics, review School only of Public Health, Makerere 19 15 University College of Health Sciences, Kampala, Uganda 20 21 16 3 22 17 Centre for Intervention Science in Maternal and Child Health (CISMAC), Centre for 23 18 International health, University of Bergen. Bergen, Norway 24 19 25 20 4Department of Paediatrics and Child Health, School of Medicine, Makerere 26 21 University College of Health Sciences, Kampala, Uganda 27 22 28 5 29 23 Department of Epidemiology and Biostatistics, School of Public Health, Makerere 30 24 University College of Health Sciences, Kampala, Uganda 31 25 32 26 7Department of Paediatrics and Child Health, Gulu University, Gulu, Uganda 33 27 34 28 8Department of Paediatrics and Child Health, Juba University, Juba, Uganda 35 29 36 30 9Department of Nursing and Midwifery, Lira University, Lira, Uganda 37 http://bmjopen.bmj.com/ 38 31 39 32 10Department of Obstetrics and Gynaecology, Gulu University, Gulu, Uganda 40 33 41 34 11Department of Obstetrics and Gynaecology, Busitema University Faculty of Health 42 35 Sciences, Mbale, Uganda 43 44 36 12

45 37 Department of Public Health, Busitema University Faculty of Health Sciences, on September 29, 2021 by guest. Protected copyright. 46 38 Mbale, Uganda 47 39 48 40 49 41 *Corresponding author 50 42 David Mukunya, P.O Box 2190, Mbale, Uganda 51 52 43 Email; [email protected], Mob; +256775152316 53 44 54 45 Key words: newborn care, neonatal care, kangaroo mother care 55 46 56 47 Word count: 3264 57 48 58 49 59 60 50

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 29

1 2

3 51 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 52 5 6 53 Abstract 7 54 8 55 9 56 Objective: To determine the prevalence, predictors, and case fatality risk of 10 57 hypothermia among neonates in Lira district, Northern Uganda. 11 58 12 Setting: Three sub counties of Lira district in Northern Uganda. 13 59 14 60 15 61 Design: This was a community-based cross-sectional study nested in a cluster 16 62 randomized controlled trial. 17 63 18 64 Participants: MotherFor – babypeer pairs enrolled review in a cluster only randomised controlled trial. 19 65 An axillary temperature was taken during a home visit using a lithium battery- 20 21 66 operated digital thermometer. 22 67 23 68 Primary and secondary outcomes: The primary outcome measure was the 24 69 prevalence of hypothermia. Hypothermia was defined as mild if the axillary 25 70 temperature was 36.0 °C - <36.5 °C, moderate if the temperature was 32.0 °C - <36.0 26 71 °C, and severe hypothermia if the temperature was < 32.0 °C. The secondary outcome 27 72 measure was the case-fatality risk of neonatal hypothermia. Predictors of moderate to 28 29 73 severe hypothermia were determined using a generalized estimating equation model 30 74 for the Poisson family. 31 75 32 76 Results: We recruited 1330 neonates. The prevalence of hypothermia (<36.5°C) was 33 77 678/1330 [51.0%: 95% CI (46.9-55.1)]. Overall, 32% (429/1330), 95%CI (29.5-35.2)] 34 78 had mild hypothermia, whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate 35 79 hypothermia. None had severe hypothermia. At multivariable analysis, predictors of 36 80 neonatal hypothermia included: home birth [adjusted prevalence ratio, aPR, 1.9, 95% 37 http://bmjopen.bmj.com/ 38 81 CI (1.4-2.6)]; low birth weight [aPR 1.7, 95%CI (1.3-2.3)]; and delayed breastfeeding 39 82 initiation [aPR 1.2, 95%CI (1.0-1.5)]. The case fatality risk ratio of hypothermic 40 83 compared to normothermic neonates was 2.0 (95% CI 0.60-6.9). 41 84 42 85 Conclusion: The prevalence of neonatal hypothermia was very high, demonstrating 43 44 86 that communities in tropical climates should not ignore neonatal hypothermia.

45 87 Interventions designed to address neonatal hypothermia should consider ways of on September 29, 2021 by guest. Protected copyright. 46 88 reaching neonates born at home and those with low birth weight. The promotion of 47 89 early breastfeeding initiation and skin-to-skin care could reduce the risk of neonatal 48 90 hypothermia. 49 91 50 92 51 52 93 53 94 54 95 Article summary: 55 96 56 97 Strengths and limitations of this study; 57 98 58 99  This is the first purely community based assessment of neonatal hypothermia 59 60 100 in sub-Saharan Africa

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 29 BMJ Open

1 2

3 101  Estimates obtained are generalizable to settings with a significant proportion BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 102 of home births unlike previous estimates from health facility based studies 5 6 103  We included a large number of neonates (1330) which increased the precision 7 104 of our estimates 8 105  The choice of a digital thermometer, placed in the axilla could have 9 106 underestimated hypothermia, but this was the most socially acceptable option 10 107  We did not measure some predictors such as delivery room temperature and 11 108 maternal body temperature 12 13 109 14 110 15 111 1 Introduction 16 112 17 113 Neonatal mortality (death of neonates less than 28 days) in Uganda is unacceptably 18 114 high, at 22.3 deathsFor per 1,000peer live births review compared to 1.6only deaths per 1,000 live births 19 115 in high-income countries [1]. In order to attain the global target of reducing neonatal 20 21 116 mortality to under 12 deaths per 1,000 live births by 2030 [2], there is a need to 22 117 identify and quantify the predictors of neonatal mortality; especially those that are 23 118 preventable by available low-cost interventions [3,4]. One of the predictors of 24 119 neonatal mortality that can easily be solved by available low-cost interventions is 25 120 neonatal hypothermia [5]. 26 121 27 122 Neonatal hypothermia, defined as an axillary temperature less than 36.5 °C [6,7], is 28 29 123 associated with increased neonatal morbidity and mortality [8-10]. Countries with 30 124 high neonatal mortality have high rates of neonatal hypothermia [11]. Hypothermia 31 125 mainly contributes to mortality by worsening outcomes of severe neonatal infections, 32 126 preterm birth, and birth asphyxia [5,6,11]. It is estimated that 20% of deaths due to 33 127 prematurity and 10% of deaths in term babies could be prevented by improved 34 128 thermal care [12]. In addition, neonatal hypothermia results in reduced growth and 35 129 development [13]. 36 130 37 http://bmjopen.bmj.com/ 38 131 Neonates are unable to maintain their body temperature without thermal protection 39 132 [14]. They are susceptible to hypothermia due to physical and environmental factors. 40 133 Physical factors that predispose neonates to hypothermia include a large surface area 41 134 to volume ratio, thin skin, and low amounts of insulating fat [5,11,14,15]. 42 135 Environmental factors that predispose neonates to hypothermia include poor thermal 43 44 136 practices around the time of birth, such as keeping the neonate away from the mother

45 137 and bathing the newborn within 24 hours of birth [16], which are common practices in on September 29, 2021 by guest. Protected copyright. 46 138 sub-Saharan Africa [17,18]. The World Health Organization recommends a ten-step 47 139 warm chain to prevent neonatal hypothermia: a warm delivery room, immediate 48 140 drying, delayed bathing, skin to skin contact, early and exclusive breastfeeding, 49 141 appropriate clothing/bedding, keeping the baby with the mother, warm transportation 50 142 and resuscitation, and training/raising awareness on the dangers of hypothermia [6]. 51 52 143 However, these actions are often suboptimal in most communities in sub-Saharan 53 144 Africa [19], and disregarded with the misguided assumption that a warm climate 54 145 guarantees thermal protection to the neonates [20,21]. Neonates are at greatest risk of 55 146 hypothermia on the first day of life and this is mainly a result of evaporation of 56 147 amniotic fluid and the neonate’s limited ability to generate heat [15,22]. 57 148 58 149 Despite a significant proportion of births and deaths taking place at home in sub- 59 60 150 Saharan Africa, there is little to no data on hypothermia obtained from community

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 29

1 2

3 151 studies [5,23]. Previous estimates of hypothermia in sub-Saharan Africa have mostly BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 152 been obtained from health facility studies [9,10,20,21,24,25] and may therefore not be 5 6 153 representative of populations with poor health-seeking behaviors. Researchers 7 154 conducting community-based studies have been encouraged to incorporate axillary 8 155 temperatures with standard inexpensive digital thermometers in their study protocols 9 156 in order to enrich the literature on community estimates of neonatal hypothermia [5]. 10 157 This information is necessary when advocating for the scale-up of existing 11 158 interventions known to reduce hypothermia [23]. Therefore, in this study, we 12 determined the prevalence, predictors, and case fatality risk of hypothermia among 13 159 14 160 neonates in Lira district, Northern Uganda. 15 161 16 162 2 Materials and Methods 17 163 18 164 For peer review only 19 165 Study setting 20 21 166 This study was conducted in Lira district, located in Lango region a post-conflict area 22 167 in Northern Uganda, in the sub-counties of Aromo, Agweng, and Ogur between 23 168 January 2018 and March 2019. About 400,000 people live in Lira; the majority live in 24 169 rural areas and practice subsistence farming [26]. In Lango region, 97% of pregnant 25 170 women attend at least one antenatal care visit from a skilled provider, only 66% of 26 171 births take place in a health facility, and approximately 29 out of every 1,000 27 172 neonates died in the first 28 days of life [27]. During the period of this study, the 28 29 173 average monthly temperatures ranged from 27.8 °C to 35.0 °C (Ngeta weather station, 30 174 Lira district). Women who give birth vaginally are discharged from health facilities 31 175 within 24 hours and those who give birth by caesarean section are discharged within 32 176 72 hours, unless complications occur. 33 177 34 178 Study design 35 179 This was a cross sectional study conducted between January 2018 and March 2019. 36 180 The study was nested in a cluster randomized controlled trial designed to promote 37 http://bmjopen.bmj.com/ 38 181 health facility birth, newborn care practices (early and exclusive breastfeeding, skin to 39 182 skin care), and timely postnatal health facility visits (Survival Pluss study registered 40 183 on ClinicalTrial.gov as NCT02605369). 41 184 42 185 Study participants 43 44 186 All neonates born to mothers participating in the cluster randomized controlled trial

45 187 were eligible for this study. We excluded neonates whose mothers were too sick to on September 29, 2021 by guest. Protected copyright. 46 188 participate in the interview, and neonates that died before we visited. 47 189 48 190 Power and sample size 49 191 A total of 1330 neonates participated in our study. The participants were initially 50 192 enrolled in a cluster randomized controlled study which had a neonatal hypothermia 51 52 193 intra cluster correlation coefficient of 0.044, and average cluster sample size of 65, 53 194 giving us a design effect of 3.8, and effective sample size of 350, resulting in absolute 54 195 precision of 1.5% to 5.2%, i.e. the difference between the point estimate and the 95% 55 196 confidence interval (CI) for prevalence values ranging from 2% to 50%. Since we 56 197 were studying a very common outcome, we deemed this precision adequate. 57 198 58 199 Main variables 59 60 200

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 29 BMJ Open

1 2

3 201 Outcome variable BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 202 The outcome variable in this study was hypothermia, which was defined as mild 5 6 203 hypothermia if the axillary temperature was between 36.0 °C and less than 36.5 °C, 7 204 moderate if the temperature was between 32.0 °C and less than 36.0 °C, and severe 8 205 hypothermia if the temperature was less than 32.0°C. We also graded hypothermia 9 206 according to a classification proposed by Mullany et al. [23]. Briefly, Mullany et al. 10 207 classified hypothermia as follows: Grade 1 (36.0 °C -36.5 °C), Grade 2 (35.0 °C - 11 208 <36.0 °C), Grade 3 (34.0 °C -<35.0 °C), and Grade 4 (<34.0 °C). 12 13 209 14 15 210 Exposure variables 16 17 211 Data were collected on several predictors during pregnancy and immediately after 18 212 birth. These included:For maternal peer age, parity,review maternal education, only paternal education, 19 213 wealth, singleton or multiple birth, sex of the newborn, place of birth, birth weight, 20 21 214 early breastfeeding initiation, bathing of the newborn, and whether the baby was 22 215 placed on the mother’s chest or abdomen immediately after birth. We classified the 23 216 season as wet if the average monthly precipitation was 60 mm or more (Koppen- 24 217 Geiger climate classification) [28]. The average monthly precipitation and 25 218 temperature for the study period were obtained from the Ngeta weather station in Lira 26 219 district. Wealth quintiles were calculated from an asset-based index using principal 27 220 component analysis. The following assets and house characteristics were considered: 28 29 221 cupboard, bicycle, radio, mobile phone, motorcycle, cement floor, iron sheets, burnt 30 222 bricks, and land ownership. We defined early breastfeeding initiation as the initiation 31 223 of breastfeeding within one hour of birth. Education level was categorised into 32 224 primary, secondary and tertiary. The primary level corresponds to 1-7 years of 33 225 education, the secondary level to 8-13 years of education and the tertiary level to 34 226 more than 13 years of education. 35 36

37 227 http://bmjopen.bmj.com/ 38 39 228 Data collection 40 41 229 As part of the trial in which this study was nested, a team of 42 research assistants 42 230 collected data and conducted measurements on the first day of birth, or as soon as 43 44 231 possible after birth at the mother’s home. A temperature was taken high in the axilla

45 232 during the study visit. We used a lithium battery-operated digital thermometer: Model on September 29, 2021 by guest. Protected copyright. 46 233 TM01 (manufactured by Cotronic Manufacturing, Shenzhen). The research assistants 47 234 were trained on how to measure temperature and supervised by a team consisting of 48 235 three paediatricians, one obstetrician, two general practitioners, two nurses, and one 49 236 data analyst. Temperature measurements were mostly conducted before taking the 50 51 237 baby's anthropometric measurements, with emphasis placed on minimizing the time 52 238 the babies were exposed to the cold. Measurements involved putting the tip of the 53 239 thermometer high up in the apex of the axilla, halfway between the anterior and 54 240 posterior margins, and holding the arm in place until an automatic audible beep was 55 241 heard. Two measurement readings in degrees Celsius were taken and the average of 56 242 these used. Thermometers were cleaned with cotton wool soaked in 70% alcohol after 57 243 the examination. 58 59 60 244 Recruitment and follow-up

5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 29

1 2

3 245 All villages had a recruiter who was elected during the community sensitization BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 246 meetings of the trial. The recruiter was a female resident in the cluster. Recruiters 5 6 247 identified pregnant women and accompanied research assistants to the home of the 7 248 women during recruitment. They were trained during a one-day workshop, which 8 249 emphasized ethics, confidentiality, and good record keeping. Recruiters were also 9 250 given a cell phone to contact the team (site supervisor/research assistants) whenever 10 251 they identified a pregnant woman or whenever a pregnant woman had given birth. 11 252 They were paid Uganda Shillings 5,000 (USD 1.4) whenever they identified an 12 eligible participant and whenever they informed the team within 24 hours of a mother 13 253 14 254 giving birth. Approximately 250 recruiters were trained. After a recruiter informed the 15 255 team of an eligible participant, a research assistant visited the mother to ascertain 16 256 eligibility, to obtain informed consent and to conduct the interview. To ensure that 17 257 recruiters were reporting all pregnant women, we employed community health 18 258 workers (villageFor health teampeer members) review to conduct a census only of all pregnant women in 19 259 the area. Pregnant mothers and their relatives were encouraged to contact the study 20 21 260 team immediately after giving birth. Research assistants also obtained phone numbers 22 261 of pregnant women and their relatives and periodically conducted follow up phone 23 262 calls and visits to ensure that mothers were visited immediately after birth. The 24 263 process of notification was similar between health facility and home births. Data 25 264 collectors conducted follow up visits to assess whether the neonates were alive at 7 26 265 days and at 28 days. 27 266 28 29 267 Patient and Public Involvement 30 268 The public was not involved in the design and conceptualisation of the study but they 31 269 were involved in the recruitment of participants. We held community meetings in 32 270 each village during which a recruiter was elected from among the village members. 33 271 The recruiter was responsible for recruitment in their village. The results of this study 34 272 will be disseminated to the wider community through community dialogue meetings 35 273 at parish level in each participating village. 36 274 37 http://bmjopen.bmj.com/ 38 275 Statistical analysis 39 276 Data were analyzed using Stata version 14.0 (StataCorp; College Station, TX, USA). 40 277 Study characteristics were compared across the exposure status and summarized as 41 278 proportions for categorical data and means for continuous data. Hypothermia was 42 279 categorized using both the WHO classification [6], and a classification suggested by 43 44 280 Mullany et al [23], and presented as proportions with corresponding 95% confidence

45 281 intervals adjusted for clustering. Factors associated with moderate to severe on September 29, 2021 by guest. Protected copyright. 46 282 hypothermia were determined using a generalized estimating equation model for the 47 283 Poisson family, with a log link, allowing for the clustering and assuming an 48 284 exchangeable correlation. We used robust variance estimation in our model. 49 285 Predictors of hypothermia included in our multivariable model were determined a 50 286 priori during a review of the literature on the subject. Factors included as predictors in 51 52 287 our model were: mother’s age, mother’s education, mode of birth, place of birth, low 53 288 birth weight, wealth, parity, season, baby placed on mother’s chest or abdomen 54 289 immediately after birth, cleaning/drying the baby immediately after birth, bathing the 55 290 baby, delayed initiation of breastfeeding [5,6,11,16,20-23]. All variables included in 56 291 the model were assessed for collinearity and considered collinear if they had a 57 292 variance inflation factor greater than 10. In the case of collinearity, we retained the 58 293 variable with greater biological plausibility and/or measure of association. The 59 60 294 multivariable analyses were based on a complete case analysis. However, we

6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 29 BMJ Open

1 2

3 295 conducted sensitivity analyses of best case, worst case and most realistic scenarios to BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 296 assess the potential effect of the missing data. We also conducted sub-group analysis 5 6 297 of the prevalence of hypothermia by date of neonate on examination and by place of 7 298 birth. Since this study was nested in a cluster randomized controlled trial, the trial arm 8 299 was added as a fixed effect in all the models. 9 300 10 301 11 302 12 3 Results 13 303 14 304 15 305 Participant characteristics 16 306 On our visits to the mothers, we were able to take the temperature measurements of 17 307 1527 neonates; of these we used the data of 1330 for whom temperatures were taken 18 308 within the first 72For hours afterpeer birth (figure review 1). The mean only age of mothers was 24.6 19 309 years (standard deviation (sd) 6.8) and their median education was 5 years (inter 20 21 310 quartile range (iqr) 3-6). The mean weight of neonates was 3.2 kg (sd 0.5) (table 1). 22 311 23 312 Hypothermia 24 313 The mean temperature was 36.4 °C (sd 0.7), and the median temperature was 36.4 °C 25 314 (iqr 36.1 °C -36.8 °C). The minimum temperature recorded was 32.0 °C and the 26 315 maximum temperature recorded was 39.4 °C. The prevalence of hypothermia 27 316 (temperature less than 36.5°C) was 678/1330 [51.0%: 95% CI (46.9-55.1)]. Overall, 28 29 317 32% (429/1330), 95%CI (29.5-35.2)] had mild hypothermia (temperature 36.0 °C - 30 318 <36.5 °C), whereas 18.7% (249/1330), 95% CI (15.8-22.0) had moderate hypothermia 31 319 (temperature 32.0 °C - <36.0 °C). No neonate had severe hypothermia (temperature 32 320 less than 32.0°C) (table 2a). We also graded hypothermia according to a classification 33 321 proposed by Mullany et al. [23] and present the results in table 2b. Sensitivity 34 322 analyses conducted suggested that we might have underestimated the burden (online 35 323 supplementary appendix table 1,2). Hypothermia was more common among home 36 324 births and on the first day of birth. Results of the third day of life were very imprecise 37 http://bmjopen.bmj.com/ 38 325 (online supplementary appendix table 3,4). 39 326 40 327 Factors associated with hypothermia 41 328 Using multivariable analysis, the factors associated with neonatal hypothermia 42 329 included: home birth [adjusted prevalence ratio, aPR, 1.9, 95% CI (1.4-2.6)], low 43 44 330 birth weight [aPR 1.7, 95%CI (1.3-2.3)], and delayed breastfeeding initiation [aPR

45 331 1.2, 95%CI (1.0-1.5] (table 3). on September 29, 2021 by guest. Protected copyright. 46 332 47 333 Case fatality risk 48 334 The risk of death among neonates with moderate hypothermia was 3/249 (1.2%, 49 335 95%CI 0.38-3.7), compared to 6/1023 (0.59%, 95% CI 0.28-1.2) among neonates 50 336 with normal temperature, resulting in a case fatality risk ratio of 2.0 (95% CI 0.60- 51 52 337 6.9). 53 338 54 339 55 340 56 341 57 342 58 343 59 60 344 4. Discussion

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 29

1 2

3 345 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 346 The prevalence of hypothermia in this study was high. Half of the neonates 5 6 347 developed hypothermia; 33% developed mild hypothermia; 19% developed moderate 7 348 hypothermia. Similar findings were observed in a community-based study in Nepal, 8 349 where 59% of neonates developed hypothermia on the first day [29], and in another 9 350 community-based study in India where the prevalence of hypothermia was 45% [30]. 10 351 However, the prevalence of hypothermia observed in our study was much higher than 11 352 that observed in two other studies in India, which observed a prevalence of 11% [31] 12 and 17% [32]. The difference could be explained by the different definitions of 13 353 14 354 hypothermia used in the studies. We defined hypothermia as a temperature less than 15 355 36.5 °C in accordance with recommendations from the World Health Organization 16 356 [6], whereas Kumar et al defined hypothermia as a temperature less than 35.6 °C, and 17 357 Bang et al defined hypothermia as a temperature less than 35 °C. 18 358 For peer review only 19 359 Neonates who had low birth weight were more likely to be hypothermic compared to 20 21 360 neonates with normal birth weight. This finding is not surprising. Similar findings 22 361 were observed in a community-based study conducted in Nepal [33] and in many 23 362 other hospital-based studies in Uganda, Ethiopia [24,25], and other countries [20,34]. 24 363 Low birth weight neonates have less capability to conserve and generate heat. This is 25 364 mainly because of physiological factors such as the reduced amount of brown fat and 26 365 a poor shivering reflex [15,35]. These thermo-protective mechanisms are needed to 27 366 maintain a normal temperature in neonates who are exposed to hypothermic 28 29 367 situations. 30 368 31 369 Neonates born at home were more likely to be hypothermic compared to neonates 32 370 born in health facilities. This finding has also been reported in other settings [35]. A 33 371 study in Uganda found that mothers who gave birth at home were more likely to 34 372 practice sub-optimal thermal care practices [36]. Mothers who give birth at home are 35 373 more likely to bathe their neonates soon after birth [37,38], which could explain the 36 374 increased risk of hypothermia observed in neonates born at home. The main reason 37 http://bmjopen.bmj.com/ 38 375 for bathing neonates early is the belief that neonates are dirty, having come into 39 376 contact with maternal fluids and the vernix caseosa [17,39-41]. Bathing neonates is 40 377 also perceived as a prerequisite to good rest and sleep [39]. However, early bathing 41 378 has been shown to result in a substantial drop in the neonate’s temperatures [16]. We 42 379 recommend that neonates are not bathed within the first 24 hours after birth [14] and 43 44 380 that bathing be done with warm water, after which the neonate should be placed on

45 381 the skin of the caregiver or placed in adequate warm clothing if available. on September 29, 2021 by guest. Protected copyright. 46 382 47 383 Despite the generally impoverished nature of the study area, belonging to a relatively 48 384 lower social-economic status was also a predictor of hypothermia in this population. 49 385 Mothers with low socioeconomic status often lack resources to buy materials that can 50 386 keep the neonate warm [42] and may have limited access to health information [43]. 51 52 387 This should not be a big problem if the mother practices skin-to-skin care. 53 388 Unfortunately, many mothers in Uganda [41], Ethiopia [44], Ghana [44], Tanzania 54 389 [44] and Mali [44] do not practice skin-to-skin care. Reasons for not practicing 55 390 adequate neonatal care include beliefs that skin-to-skin care could result in the 56 391 transmission of diseases to the neonate and could hurt the umbilical cord of the 57 392 neonate [41,45,46]. 58 393 59 60

8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 29 BMJ Open

1 2

3 394 Mothers who delayed putting their neonates to the breast were more likely to have BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 395 hypothermic babies. This finding was also observed in the community-based study in 5 6 396 Nepal [33]. Neonates who are breastfed early receive warmth from their mothers and 7 397 this explains the reduction in hypothermia [23,47]. Mothers who had higher education 8 398 were less likely to have hypothermic babies, although this finding was imprecise. 9 399 There was also no difference between mothers in the intervention group and the 10 400 control group, meaning the intervention did not prevent the neonate from becoming 11 401 hypothermic. 12 13 402 14 403 Methodological considerations 15 404 We did not measure some predictors such as delivery room temperature and maternal 16 405 temperature. We could also have underestimated hypothermia by using a digital 17 406 thermometer, placed in the axilla. Digital thermometers might slightly over or 18 407 underestimate temperatureFor peer readings asreview compared to mercury only thermometers [48-50]. 19 408 We used these because they are inexpensive, locally available, and easy to use by 20 21 409 community workers [29]. In addition, digital thermometers are easier to use in poorly 22 410 lit rural homes [29]. We used axillary measurements because they were easier to do, 23 411 safer, and more acceptable than rectal measurements [29]. In a systematic review 24 412 studying differences between rectal and axillary temperatures, the pooled mean 25 413 difference of rectal minus axillary temperature was estimated to be 0.17 °C, ranging 26 414 from -0.15 °C to 0.5 °C [51]. Our study could have suffered from a selection bias 27 415 since only 75% of eligible participants were recruited. From our sensitivity analysis, 28 29 416 we believe hypothermia is still a big challenge, and that this selection bias might have 30 417 slightly underestimated the burden since it was possibly the very sick who were not 31 418 visited within 72 hours of birth. We believe this selection bias also greatly 32 419 underestimated the mortality attributed to hypothermia since many more neonates 33 420 died in the unmeasured group. This is understandable since the majority of newborn 34 421 deaths in the study, as would be expected, occurred in the first few hours after birth 35 422 before our teams were able to reach the scene. The lack of gestational age data is 36 423 another limitation in our study. We believe that our findings are generalizable to rural 37 http://bmjopen.bmj.com/ 38 424 areas in tropical low-income countries with similar newborn care practices. 39 425 40 426 41 427 6 Conclusion 42 428 The prevalence of neonatal hypothermia was very high, demonstrating that 43 44 429 communities in tropical climates should not ignore neonatal hypothermia.

45 430 Interventions designed to address neonatal hypothermia should consider ways of on September 29, 2021 by guest. Protected copyright. 46 431 reaching neonates born at home, as these are at greater risk of hypothermia. Low birth 47 432 weight neonates, and neonates born to mothers in the poorest socioeconomic status, 48 433 should also be prioritized. We recommend promotion of low-cost interventions such 49 434 as skin-to-skin care for all neonates born in similar settings to prevent neonatal 50 435 hypothermia. 51 52 436 53 437 Ethics approval and consent to participate 54 438 Ethical approval to conduct the study was obtained from the following bodies: 1) 55 439 Research and Ethics committee School of Medicine, Makerere University (SOMREC: 56 440 REF 2015-121); 2) Uganda National Council of Science and Technology (UNCST: 57 441 SS 3954); 3) Regional Committees for Medical and Health Research Ethics (REK 58 442 VEST 2017/2079) and the trial was registered at ClinicalTrial.gov as NCT02605369. 59 60 443 We also obtained permission from the Ministry of Health and Lira Local Government.

9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 29

1 2

3 444 Written informed consent was obtained from the respondents in the study. Research BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 445 assistants were trained in confidentiality and the right of the respondent to withdraw 5 6 446 their participation at any time during the study. At the community level, we obtained 7 447 permission to include clusters during community sensitization meetings, after which 8 448 the community members democratically elected recruiters, and peer buddies when 9 449 applicable, from amongst themselves. 10 450 11 451 Consent for publication 12 Not applicable 13 452 14 453 15 454 Availability of data and materials 16 455 The datasets used and/or analysed during the current study are available from the 17 456 corresponding author on reasonable request. 18 457 For peer review only 19 458 Funding 20 21 459 Funding was obtained from the Survival Pluss project; grant number UGA-13-0030 at 22 460 Makerere University. Survival Pluss project is funded by The Norwegian Program for 23 461 Capacity Development in Higher Education and Research for Development 24 462 (NORHED) under The Norwegian Agency for Development Cooperation (NORAD). 25 463 26 464 Author contributions 27 465 David Mukunya (D.M), James K. Tumwine (J.K.T), Victoria Nankabirwa (V.N), 28 29 466 Grace Ndeezi (G.N), and Thorkild Tylleskar (T.T) conceived, designed, supervised 30 467 the study, analyzed the data, and wrote the first draft of manuscript. Milton W. 31 468 Musaba (M.W.M), Josephine Tumuhamye (J.T), Justin. B. Tongun (J.B.T), Agnes 32 469 Napyo (A.N), Vivian Zalwango (V.Z), Vicentina Achora (V.A), Beatrice Odongkara 33 470 (B.O), and Agnes Anna Arach (A.A.A) were instrumental in the design and 34 471 supervision of the study, and in drafting of the manuscript. All authors read and 35 472 approved the final version to be published. 36 473 37 http://bmjopen.bmj.com/ 38 474 Conflict of interest 39 475 All authors declare no conflict of interest. 40 476 41 477 Acknowledgments 42 478 In a special way, we acknowledge the District Health Office of Lira district, and the 43 44 479 various district, sub-county, parish, and village leaders for their assistance in this

45 480 study. We thank the study participants for accepting to be part of the study and on September 29, 2021 by guest. Protected copyright. 46 481 research assistants for working tirelessly to make this work a reality. In a special way, 47 482 we acknowledge the excellent work performed by our recruiters in making this study 48 483 possible. Finally, we extend heartfelt appreciation to Ms. Jo Weeks for the excellent 49 484 English editing. 50 485 51 52 486 Figure and Title legends 53 487 54 488 Figure 1: Study profile of neonates assessed for hypothermia in Lira district, 55 489 Northern Uganda 56 490 Table 1: Participant characteristics of neonates assessed for hypothermia in Northern 57 491 Uganda 58 492 Table 2a: Prevalence of hypothermia (defined by the World Health Organization 59 60 493 classification) in Lira district, Northern Uganda

10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 29 BMJ Open

1 2

3 494 Table 2b: Prevalence of hypothermia (defined by the Mullany classification) in Lira BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 495 district, Northern Uganda 5 6 496 Table 3: Factors associated with moderate to severe hypothermia among neonates in 7 497 Lira district Northern Uganda 8 498 9 499 10 500 11 501 References 12 13 502 14 503 1 GBD 2016 Mortality Collaborators. Global, regional, and national under-5 15 504 mortality, adult mortality, age-specific mortality, and life expectancy, 1970-2016: a 16 505 systematic analysis for the Global Burden of Disease Study 2016. Lancet (London, 17 506 England) 2017;390(10100):1084-150. 18 For peer review only 19 507 2 United Nations. Sustainable Development Goals. Secondary Sustainable 20 508 Development Goals 2015. http://www.un.org/sustainabledevelopment/summit/. 21 22 23 509 3 Darmstadt GL, Bhutta ZA, Cousens S, et al. Evidence-based, cost-effective 24 510 interventions: how many newborn babies can we save? Lancet (London, England) 25 511 2005;365(9463):977-88. 26 27 512 4 Knippenberg R, Lawn JE, Darmstadt GL, et al. Systematic scaling up of 28 513 neonatal care in countries. Lancet (London, England) 2005;365(9464):1087-98. 29 30 514 5 Lunze K, Bloom DE, Jamison DT, et al. The global burden of neonatal 31 515 hypothermia: systematic review of a major challenge for newborn survival. BMC 32 33 516 Medicine 2013;11:24. 34 35 517 6 World Health Organization. Thermal Protection of the Newborn: a practical 36 518 guide. Geneva: World Health Organization, 1997.

37 http://bmjopen.bmj.com/ 38 519 7 Lunze K, Yeboah-Antwi K, Marsh DR, et al. Prevention and management of 39 520 neonatal hypothermia in rural Zambia. PloS One 2014;9(4):e92006. 40 41 521 8 Christensson K, Bhat GJ, Eriksson B, et al. The effect of routine hospital care 42 522 on the health of hypothermic newborn infants in Zambia. Journal of Tropical 43 44 523 Pediatrics 1995;41(4):210-4.

45 on September 29, 2021 by guest. Protected copyright. 46 524 9 Kambarami R, Chidede O. Neonatal hypothermia levels and risk factors for 47 525 mortality in a tropical country. The Central African Journal of Medicine 2003;49(9- 48 526 10):103-6. 49 50 527 10 Sodemann M, Nielsen J, Veirum J, et al. Hypothermia of newborns is 51 528 associated with excess mortality in the first 2 months of life in Guinea-Bissau, West 52 53 529 Africa. Tropical medicine & international health : TM & IH 2008;13(8):980-6. 54 55 530 11 Kumar V, Shearer JC, Kumar A, et al. Neonatal hypothermia in low resource 56 531 settings: a review. Journal of perinatology : official journal of the California 57 532 Perinatal Association 2009;29(6):401-12. 58 59 60

11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 29

1 2

3 533 12 Bhutta ZA, Das JK, Bahl R, et al. Can available interventions end preventable BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 534 deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet (London, 5 6 535 England) 2014;384(9940):347-70. 7 8 536 13 Glass L, Silverman WA, Sinclair JC. Relationship of thermal environment and 9 537 caloric intake to growth and resting metabolism in the late neonatal period. Biologia 10 538 Neonatorum Neo-natal studies 1969;14(5):324-40. 11 12 539 14 Lunze K, Hamer DH. Thermal protection of the newborn in resource-limited 13 540 environments. Journal of perinatology : official journal of the California Perinatal 14 Association 2012;32(5):317-24. 15 541 16 17 542 15 Adamson SK, Jr., Towell ME. Thermal homeostasis in the fetus and newborn. 18 543 Anesthesiology For1965;26 :531-48.peer review only 19 20 544 16 Bergstrom A, Byaruhanga R, Okong P. The impact of newborn bathing on the 21 545 prevalence of neonatal hypothermia in Uganda: a randomized, controlled trial. Acta 22 546 Paediatrica (Oslo, Norway : 1992) 2005;94(10):1462-7. 23 24 547 17 Waiswa P, Kemigisa M, Kiguli J, et al. Acceptability of evidence-based 25 26 548 neonatal care practices in rural Uganda - implications for programming. BMC 27 549 Pregnancy and Childbirth 2008;8:21. 28 29 550 18 Hill Z, Tawiah-Agyemang C, Manu A, et al. Keeping newborns warm: beliefs, 30 551 practices and potential for behaviour change in rural Ghana. Tropical Medicine & 31 552 International Health : TM & IH 2010;15(10):1118-24. 32 33 553 19 Coalter WS, Patterson SL. Sociocultural factors affecting uptake of home- 34 554 based neonatal thermal care practices in Africa: A qualitative review. Journal of Child 35 36 555 Health Care : for professionals working with children in the hospital and community

37 556 2017;21(2):132-41. http://bmjopen.bmj.com/ 38 39 557 20 Manji KP, Kisenge R. Neonatal hypothermia on admission to a special care 40 558 unit in Dar-es-Salaam, Tanzania: a cause for concern. The Central African Journal of 41 559 Medicine 2003;49(3-4):23-7. 42 43 560 21 Byaruhanga R, Bergstrom A, Okong P. Neonatal hypothermia in Uganda: 44 561 prevalence and risk factors. Journal of Tropical Pediatrics 2005;51(4):212-5. 45 on September 29, 2021 by guest. Protected copyright. 46 47 562 22 Smales OR, Kime R. Thermoregulation in babies immediately after birth. 48 563 Archives of Disease in Childhood 1978;53(1):58-61. 49 50 564 23 Mullany LC. Neonatal hypothermia in low-resource settings. Seminars in 51 565 Perinatology 2010;34(6):426-33. 52 53 566 24 Tasew H, Gebrekristos K, Kidanu K, et al. Determinants of hypothermia on 54 567 neonates admitted to the intensive care unit of public hospitals of Central Zone, 55 56 568 Tigray, Ethiopia 2017: unmatched case-control study. BMC Research Notes 57 569 2018;11(1):576. 58 59 60

12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 29 BMJ Open

1 2

3 570 25 Demissie BW, Abera BB, Chichiabellu TY, et al. Neonatal hypothermia and BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 571 associated factors among neonates admitted to neonatal intensive care unit of public 5 6 572 hospitals in Addis Ababa, Ethiopia. BMC Pediatrics 2018;18(1):263. 7 8 573 26 Uganda Bureau of Statistics. The National Population and Housing Census 9 574 2014 – Area 10 575 Specific Profile Series, Kampala, Uganda. Secondary The National Population and 11 576 Housing Census 2014 – Area 12 577 Specific Profile Series, Kampala, Uganda 2017. 13 http://www.ubos.org/onlinefiles/uploads/ubos/2014CensusProfiles/MUKONO.pdf. 14 578 15 16 579 27 Uganda Bureau of Statistics (UBOS) and ICF. Uganda Demographic and 17 580 Health Survey 2016. Kampala, Uganda and Rockville, Maryland, USA: UBOS and 18 581 ICF, 2018. For peer review only 19 20 582 28 Peel MC, Finlayson BL, McMahon TA. Updated world map of the Koppen- 21 583 Geiger climate classification. Hydrol Earth Syst Sci 2007;11:1633–44. 22 23 584 29 Mullany LC, Katz J, Khatry SK, et al. Incidence and seasonality of 24 25 585 hypothermia among newborns in southern Nepal. Archives of Pediatrics & Adolescent 26 586 Medicine 2010;164(1):71-7. 27 28 587 30 Darmstadt GL, Kumar V, Yadav R, et al. Introduction of community-based 29 588 skin-to-skin care in rural Uttar Pradesh, India. Journal of perinatology : official 30 589 journal of the California Perinatal Association 2006;26(10):597-604. 31 32 590 31 Kumar R, Aggarwal AK. Body temperatures of home delivered newborns in 33 591 north India. Tropical Doctor 1998;28(3):134-6. 34 35 36 592 32 Bang AT, Reddy HM, Baitule SB, et al. The incidence of morbidities in a

37 593 cohort of neonates in rural Gadchiroli, India: seasonal and temporal variation and a http://bmjopen.bmj.com/ 38 594 hypothesis about prevention. Journal of Perinatology : official Journal of the 39 595 California Perinatal Association 2005;25 Suppl 1:S18-28. 40 41 596 33 Mullany LC, Katz J, Khatry SK, et al. Neonatal hypothermia and associated 42 597 risk factors among newborns of southern Nepal. BMC Medicine 2010;8:43. 43 44 598 34 Zayeri F, Kazemnejad A, Ganjali M, et al. Hypothermia in Iranian newborns. 45 on September 29, 2021 by guest. Protected copyright. 46 599 Incidence, risk factors and related complications. Saudi Medical Journal 47 600 2005;26(9):1367-71. 48 49 601 35 Onalo R. Neonatal hypothermia in sub-Saharan Africa: a review. Niger J Clin 50 602 Pract 2013;16(2):129-38. 51 52 603 36 Kabwijamu L, Waiswa P, Kawooya V, et al. Newborn care practices among 53 604 adolescent mothers in Hoima district, Western Uganda. PloS One 54 55 605 2016;11(11):e0166405. 56 57 606 37 Mrisho M, Schellenberg JA, Mushi AK, et al. Understanding home-based 58 607 neonatal care practice in rural southern Tanzania. Transactions of the Royal Society of 59 608 Tropical Medicine and Hygiene 2008;102(7):669-78. 60

13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 29

1 2

3 609 38 Salasibew MM, Filteau S, Marchant T. A qualitative study exploring newborn BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 610 care behaviours after home births in rural Ethiopia: implications for adoption of 5 6 611 essential interventions for saving newborn lives. BMC Pregnancy and Childbirth 7 612 2014;14:412. 8 9 613 39 Adejuyigbe EA, Bee MH, Amare Y, et al. "Why not bathe the baby today?": 10 614 A qualitative study of thermal care beliefs and practices in four African sites. BMC 11 615 Pediatrics 2015;15:156. 12 13 616 40 Shamba D, Schellenberg J, Hildon ZJ, et al. Thermal care for newborn babies 14 in rural southern Tanzania: a mixed-method study of barriers, facilitators and 15 617 16 618 potential for behaviour change. BMC Pregnancy and Childbirth 2014;14:267. 17 18 619 41 ByaruhangaFor RN, peerNsungwa-Sabiiti review J, Kiguli J, etonly al. Hurdles and opportunities 19 620 for newborn care in rural Uganda. Midwifery 2011;27(6):775-80. 20 21 621 42 Lunze K, Dawkins R, Tapia A, et al. Market mechanisms for newborn health 22 622 in Nepal. BMC Pregnancy and Childbirth 2017;17(1):428. 23 24 623 43 Owor MO, Matovu JKB, Murokora D, et al. Factors associated with adoption 25 26 624 of beneficial newborn care practices in rural Eastern Uganda: a cross-sectional study. 27 625 BMC Pregnancy and Childbirth 2016;16:83. 28 29 626 44 Bee M, Shiroor A, Hill Z. Neonatal care practices in sub-Saharan Africa: a 30 627 systematic review of quantitative and qualitative data. Journal of Health, Population, 31 628 and Nutrition 2018;37(1):9. 32 33 629 45 Byaruhanga RN, Bergstrom A, Tibemanya J, et al. Perceptions among post- 34 630 delivery mothers of skin-to-skin contact and newborn baby care in a periurban 35 36 631 hospital in Uganda. Midwifery 2008;24(2):183-9.

37 http://bmjopen.bmj.com/ 38 632 46 Waiswa P, Peterson S, Tomson G, et al. Poor newborn care practices - a 39 633 population based survey in eastern Uganda. BMC Pregnancy and Childbirth 40 634 2010;10:9. 41 42 635 47 Huffman SL, Zehner ER, Victora C. Can improvements in breast-feeding 43 636 practices reduce neonatal mortality in developing countries? Midwifery 44 637 2001;17(2):80-92. 45 on September 29, 2021 by guest. Protected copyright. 46 47 638 48 Smith J. Are electronic thermometry techniques suitable alternatives to 48 639 traditional mercury in glass thermometry techniques in the paediatric setting? Journal 49 640 of Advanced Nursing 1998;28(5):1030-9. 50 51 641 49 Jones HL, Kleber CB, Eckert GJ, et al. Comparison of rectal temperature 52 642 measured by digital vs. mercury glass thermometer in infants under two months old. 53 643 Clinical Pediatrics 2003;42(4):357-9. 54 55 56 644 50 Latman NS, Hans P, Nicholson L, et al. Evaluation of clinical thermometers 57 645 for accuracy and reliability. Biomedical Instrumentation & Technology 58 646 2001;35(4):259-65. 59 60

14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 29 BMJ Open

1 2

3 647 51 Craig JV, Lancaster GA, Williamson PR, et al. Temperature measured at the BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from 4 648 axilla compared with rectum in children and young people: systematic review. BMJ 5 6 649 (Clinical research ed) 2000;320(7243):1174-8. 7 650 8 651 9 652 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 BMJ Open: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 18 of 29

None 74 (11.4) 105 (15.5) 6 (06.8) 12 (11.0) 34 (14.1) on September 29, 2021 by guest. Protected copyright. 25 Primary 513 (78.7) 519 (76.6) 73 (83.0) 85 (78.0) 190 (78.8) 26 27 Secondary 51 (7.8) 47 (6.9) 7 (8.0) 9 (8.3) 17 (7.1) 28 Tertiary 14 (02.2) 07 (01.0) 2 (02.3) 3 (02.8) - 29 Father’s 30 31 education 32 None 14 (2.2) 11 (1.6) 1 (1.1) 1 (0.92) 6 (2.5) 33 Primary 377 (57.8) 416 (61.4) 52 (59.1) 55 (50.5) 151 (62.7) 34 35 Secondary 177 (27.2) 147 (21.7) 23 (26.1) 28 (25.7) 51 (21.2) 36 Tertiary 41 (6.3) 38 (5.6) 4 (4.6) 8 (7.3) 12 (5.0) 37 Missing 43 (6.6) 66 (9.7) 8 (9.1) 17 (15.6) 21 (8.7) 38 39 40 41 42 16 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 19 of 29 BMJ Open

BMJ Open Page 20 of 29

15 weight 16 17 No 613 (94.0) 622 (91.7) 83 (94.3) 101 (92.7) 15 (6.2) http://bmjopen.bmj.com/ 18 Yes 35 (5.4) 45 (6.6) 4 (4.6) 6 (5.5) 1 (0.41) 19 Missing 4 (0.6) 11 (01.6) 1 (1.1) 2 (1.8) 225 (93.4) 20 21 Wealth quintiles 22 1 (Poorest) 146 (22.4) 140 (20.7) 19 (21.6) 23 (21.1) 35 (14.5) 23 2 143 (21.9) 185 (27.3) 20 (22.7) 22 (20.2) 63 (26.1) 24 25 3 123 (18.9) 121 (17.9) 19 (21.6) 18 (16.5) 45 (18.7) on September 29, 2021 by guest. Protected copyright. 26 4 105 (16.1) 114 (16.8) 10 (11.4) 20 (18.4) 49 (20.3) 27 5 (Richest) 135 (20.7) 118 (17.4) 20 (22.7) 26 (23.9) 49 (20.3) 28 29 Season 30 Wet 589 (90.3) 579 (85.4) 74 (84.1) 87 (79.8) 47 (19.5) 31 Dry 63 (9.7) 99 (14.6) 14 (15.9) 22 (20.2) 194 (80.5) 32 33 34 35 36 37 38 39 40 41 42 18 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 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-2020-041723 on 11 February 2021. Downloaded from

Page 21 of 29 BMJ Open

1 2 3 4 5 6 Baby 7 placed on

BMJ Open Page 22 of 29

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 24 of 29

1 2

22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 29 BMJ Open

1 2

37 http://bmjopen.bmj.com/ 38 No 1 1 39 Yes 1.2 (0.98-1.5) 1.0 (0.81-1.2) 40 41 Breastfeeding 42 initiation 43 44 Early 1 1

23 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 29

1 2

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 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 27 of 29 BMJ Open

1 2

45 on September 29, 2021 by guest. Protected copyright. 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: first published as 10.1136/bmjopen-2020-041723 on 11 February 2021. Downloaded from

BMJ Open Page 28 of 29

Page 29 of 29 BMJ Open

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 29 BMJ Open

1 (b) Report category boundaries when continuous variables were categorized 2 (c) If relevant, consider translating estimates of relative risk into absolute risk for NA

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 29, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

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