Bed Nets for Prevention in Western Kenya: Explorations of Care and Repair, Repurposing, and Efficacy of Education Interventions

Item Type text; Electronic Dissertation

Authors Santos, Ellen

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 03/10/2021 19:42:13

Link to Item http://hdl.handle.net/10150/634253 BED NETS FOR MALARIA PREVENTION IN WESTERN KENYA: EXPLORATIONS OF CARE AND REPAIR, REPURPOSING, AND EFFICACY OF EDUCATION INTERVENTIONS

by

Ellen Santos

______Copyright © Ellen Santos 2019

A Dissertation Submitted to the Faculty of the

MEL AND ENID ZUCKERMAN COLLEGE OF

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

EPIDEMIOLOGY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2019

2

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules for the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgement of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, permission must be obtained from the author.

SIGNED: Ellen Maureen Santos

3

ACKNOWLEDGEMENTS

I wish to acknowledge several people who supported the work in this dissertation. First and foremost, my committee chair Dr. Kacey Ernst offered her mentorship and expertise. She provided guidance through my growth as an academic, supported my research and professional interests, and modeled what it means to be a researcher who never ceases to ask questions and continuously learn. I appreciate the support, time, and thoughtfulness provided by my committee members. I thank Dr. Elizabeth Jacobs for instilling in me a passion for the field of epidemiology. In addition to her timely and constructive feedback, she taught me about perspective: the role of an epidemiologist, understanding the impact of our work, and how to remain passionate yet realistic. I thank Dr. Yann Klimentidis for his time and effort bettering this work that is outside of his typical area of research. He especially provided a critical eye, which brought clarity and focus to each chapter. I cannot thank Lindsay Hansen enough for her knowledge of writing and learning, and for modeling valuable feedback skills. She made me a better reader and writer, and provided me with the tools and confidence to continue journey in writing and teaching.

I am especially appreciative of Dr. Jenna Coalson for her data cleaning efforts and assistance interpreting results. Her contributions made this work possible, and would have taken considerably more time without her efforts and expertise with the dataset. I thank Jean

McClelland for her time, expertise, and invaluable assistance with the systematic review. A special thank you to Chris Tisch and Michael Tearne, who continuously work to support and advocate for students. I appreciate the family-friendly culture of the Mel and Enid Zuckerman

4

College of Public Health. I served as the primary caretaker for my son, Emmett, during this work and always felt welcomed to bring him along to my many meetings.

I thank my sister, Colleen Shelly, for her efforts and enthusiasm in acting as a reviewer for the systematic review. It was a joy to include family into this work. My parents, John and Sarah

Shelly, were instrumental in the final stages of this project – caring for Emmett while I worked

(and contracting norovirus in the process) and helping me get a little extra sleep. Finally, I thank my many friends, family, and fellow students for their support. I particularly thank my husband,

Zach, for always maintaining confidence in my abilities and offering the support essential for completing this endeavor.

5

DEDICATION

To my husband, Zach, and my son, Emmett

6

TABLE OF CONTENTS

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

ABSTRACT ...... 11

CHAPTER 1 ...... 14

1. INTRODUCTION ...... 14

A. Explanation of the Problem ...... 14

Bed net care and repair ...... 14

Bed net misuse ...... 15

Health education ...... 15

B. Specific Aims and Hypotheses ...... 16

C. Role of the Author in the Research ...... 18

D. Background ...... 19

Malaria pathogenesis and diagnosis ...... 19

Global malaria epidemiology ...... 20

Malaria in Kenya ...... 21

Malaria transmission ...... 21

Prevention and control ...... 23

E. Summary of the Introduction ...... 25

CHAPTER 2 ...... 27

2. PRESENT STUDY ...... 27

A. Introduction ...... 27

B. Specific Aim 1: Bed net care adherence ...... 27

7

1. Methods ...... 27

2. Results ...... 32

C. Specific Aim 2: Bed net repurposing and misuse ...... 41

1. Methods ...... 41

2. Results ...... 43

D. Specific Aim 3: Systematic review ...... 55

1. Methods ...... 55

2. Results ...... 59

CHAPTER 3 ...... 79

3. DISSERTATION CONCLUSIONS ...... 79

A. Specific Aim 1: Bed net care adherence ...... 79

B. Specific Aim 2: Bed net repurposing and misuse ...... 83

C. Specific Aim 3: Systematic review of malaria education interventions ...... 87

D. Limitations...... 94

REFERENCES ...... 97

APPENDIX A: DETERMINATION OF HUMAN RESEARCH ...... 107

APPENDIX B: MANUSCRIPT 1 ...... 108

APPENDIX C: ADDITIONAL FILE 1 ...... 135

APPENDIX D: MANUSCRIPT 2 ...... 136

APPENDIX E: MANUSCRIPT 3 ...... 158

APPENDIX F: ADDITIONAL FILE 2 ...... 184

APPENDIX G: PROSPERO REGISTRATION ...... 185

APPENDIX H: SEARCH STRATEGIES ...... 196

8

LIST OF TABLES

Table 1. Elements of the bed net care adherence scoring system...... 30

Table 2. Household characteristics of study sample by site ...... 33

Table 3. Distributions of net care adherence and attitudes and environmental factors ...... 36

Table 4. GEE results association between net condition and overall net care adherence score...38

Table 5. Odds ratios (95% confidence intervals) of univariate GEE models measuring associations between attitudes and environmental factors with binary net care practices ...... 40

Table 6. Household characteristics and malaria perceptions comparing households with and without repurposed bed nets ...... 45

Table 7. Household-reported reasons for repurposing bed nets ...... 49

Table 8. Search terms used in database searches for peer-reviewed studies ...... 56

Table 9. Databases and malaria organizations searched for malaria education intervention records ...... 58

Table 10. Education intervention quality criteria tool ...... 59

Table 11. Care and repair conclusions in relation to theory use of education interventions ...... 63

Table 12. Care and repair conclusions in relation to education methods ...... 64

Table 13. Characteristics, key findings, and quality scores of peer-reviewed articles with a malaria education main focus ...... 69

Table 14. Characteristics and key findings of peer-reviewed articles with a minor malaria education focus ...... 73

9

LIST OF FIGURES

Figure 1. Attitudes and environmental factors related to adherence to net care and repair behaviors ...... 28

Figure 2. Individual bed net care adherence score components by site ...... 35

Figure 3. Various uses of repurposed bed nets as observed by the study team (3A) and as observed by study participants (3B) ...... 51

Figure 4. Observed number of holes among repurposed bed nets ...... 52

Figure 5. Univariate logistic regression results (odds ratios and 95% confidence intervals) assessing factors associated with the presence of bed net repurposing at the household level, adjusted for sublocation ...... 54

Figure 6. PRISMA scheme of record selection and exclusion criteria ...... 61

Figure 7. Care and repair conclusions in relation to education methods and theory use ...... 66

Figure 8. Bed net care and repair messages reported among Against Malaria Foundation-funded bed net distributions ...... 77

10

ABSTRACT

Background:

The effectiveness of Long Lasting Insecticidal Nets (LLINs) for malaria prevention is well known. To reach maximum effectiveness and ensure LLIN longevity, households must have access to enough nets, use nets regularly, and maintain nets properly. We aimed to gain a more holistic understanding of LLIN care, misuse, and the educational strategies used to improve these practices through secondary analyses of cross-sectional data in two regions of western Kenya.

Factors associated with bed net care and repair practices are not well known. Reports of LLIN misuse are varied, controversial, often rely on self-report, and accurate measurements are needed to maintain political and financial will for LLIN distribution programs. Health education interventions are strategies often used to improve net care, repair, and use practices. However, the effectiveness of these interventions are poorly understood. Additionally, these interventions often have a basis in health behavior theory, but the degree to which education components are informed by education theory is unknown.

Methods:

We conducted secondary analyses of a cross-sectional study conducted in two areas of differential malaria transmission in western Kenya. We developed a novel bed net care adherence score including items regarding net washing, drying, and physical maintenance to understand the factors associated with net care adherence to recommended practices.

Furthermore, we quantified and described observations of alternatively used nets to differentiate between true misuse, defined as the use of an intact net for any purpose other than covering a sleeping space, and potential repurposing of nets. Finally, we performed a systematic literature

11

review of malaria education interventions aimed at improving adherence to bed net care and repair recommendations, with an emphasis on health behavior and education theory.

Results:

While overall care practices are highly adherent, particularly in the highlands, practices related to daily storage, washing frequency, and drying location need improvement in the lowlands.

Seventy-seven percent of nets in the lowlands were washed <3 months prior to the survey compared to 23% of nets in the highlands. More nets were dried in the sun in the lowlands (32% of nets) compared to the highlands (4% of nets). Different elements of care are influenced by various malaria-related attitudes and environmental factors, highlighting the complexity of factors associated with net care. For example, households that learned about net care from community events, that share a sleeping structure with animals, and that have nets used by adult males tend to adhere to washing frequency recommendations. Regarding net repurposing, of 643 households in the highlands we observed 52 (8.1%) with repurposed nets, while 184 (33.0%) of the 574 lowlands households were observed to have repurposed nets. Among the observed repurposed nets, 77.6% of nets in the highlands and 69.8% in the lowlands were repurposed for use outdoors, commonly serving as chicken coops, fences, and tree covers. Repurposed nets found indoors served functions including curtains, covering bathrooms, and cut up and used as washing sponges. In both sites, the vast majority of nets were repurposed because households determined them to be either torn, old and worn out, or both torn and old. We found no impact of net repurposing on malaria prevention efforts. In both sites, there was no association between observed net repurposing and household (1) net ownership, (2) net use, (3) malaria in the last year, or (4) fever in the last two weeks. Net misuse was rare. Of 379 repurposed nets, 4 (1.06%) were in good condition with no holes. Of 1,758 functional nets inside homes, 13 (0.74%) were

12

reported to have ever been used for an alternative purpose. Finally, the systematic review resulted in data abstraction of 29 peer reviewed studies and 289 grey literature records. There were 16 studies with malaria education as a main activity, 10 of which were grounded in theory.

There was inconclusive evidence that education on bed net care and repair practices led to improvements in net durability or longevity, and there was no evidence that interventions grounded in theory achieved more positive results.

Conclusions:

Taken together, the three focus areas of LLIN care and repair, net misuse, and health education interventions advance our understanding of how nets are being maintained and used in field conditions, and the most effective strategies to improve care and repair knowledge and practices.

The results of this work will contribute to developing more robust and comprehensive health education interventions in key areas of bed net care and use.

13

CHAPTER 1

INTRODUCTION

A. Explanation of the Problem

The 2018 WHO Malaria Report indicates that, at best, progress has stagnated in reducing malaria in highly endemic areas and, at worst, that rates of malaria are slightly increasing [1].

Funding is cited as one of the primary barriers to progress on malaria control and elimination programs. The primary target of funding for malaria prevention is the distribution of long-lasting insecticidal nets (LLINs). Insecticide treated nets have been shown to reduce malaria incidence by half in children under 5 years of age [2]. To achieve this level of effectiveness LLINs must be available as a resource (supply), accessible to households (ownership), used regularly (use) and kept in good condition (maintenance). The impact of LLIN distribution programs can be impacted by both external and internal factors at each of these steps. Previous research has successfully informed policy to address gaps, including the provision of LLINs for free to overcome access barriers [3]. Less studied gaps include alternative purposing of LLINs and household bed net care and repair behaviors that influence longevity. In areas without continuous net distribution programs, decreased net longevity may contribute to malaria resurgence.

Bed net care and repair:

Most malaria control programs distribute LLINs every 3-5 years, the theoretical lifespan of an LLIN. Early failure of LLINs could result in resurgence of malaria in areas not scheduled for more frequent distributions. The President’s Malaria Initiative (PMI) recognized the importance of monitoring durability of LLINs and has developed a network of monitoring sites.

Studies from these monitoring sites have suggested that the serviceable lifespan of LLINs is less than anticipated, ranging from 1 to 2.5 years [4]. This reduced lifespan has been explained in part

14

by inadequate bed net care and repair behaviors. Proper care and repair of LLINs is needed to ensure effective and stable vector control between mass community distributions [5, 6, 7].

Recommended practices include hanging nets up while not in use, washing gently and infrequently and with no soap or mild soap, drying indoors or in the shade, and repairing holes

[8]. Reported adherence to these practices has been mixed.

Bed net misuse:

Distributed LLINs being used for purposes other than protecting individuals from malaria has been reported [9] but not rigorously investigated across multiple contexts. Often termed misuse, there is controversy about how often intact LLINs are misused as compared to how often older worn-out LLINs are being repurposed [9]. Estimates have ranged from 3% in work conducted in Zambia [9, 10] to 87% in Lake Tanganyika in Tanzania, where LLINs were used for fishing purposes [11]. More rigor in the definition of misuse is needed to ensure that results are reproducible across studies. A better understanding of LLIN misuse and repurposing is required to maintain political and financial support of malaria prevention efforts [9], and to identify solutions to maximize effective use of a limited resource. Most assessments of misuse have relied on self-report, which may limit data validity due to sampling, recall, and social desirability biases [12, 13, 14].

Health Education:

Optimally, households will continuously use their LLINs and follow guidelines for net care and repair so that nets retain their effectiveness to prevent malaria transmission. Health education interventions are one strategy to improve LLIN care practices and use. Intervention strategies vary between studies, ranging from instruction cards attached to net packages [15], providing malaria prevention education to adult microfinance clients [16], to targeting school

15

children populations [17, 18]. While many education interventions report success in increased knowledge and changed behaviors of bed net care and use [3], it is difficult to evaluate their impact or their methods for replication. To understand how impactful education interventions may be and inform future interventions to achieve the WHO Global Technical Strategy goals, it must become clear how they are conducted and how health behavior theories form their rationale. Health behavior theories should guide the development of education interventions [19] and furthermore these interventions must be described in appropriate detail for replication and comparability so that theory can translate into public health practice [20]. Beyond theory, interventions must describe the specific instructional strategies employed [21], as delivery of the educational events is highly important to the intervention’s effectiveness.

B. Specific Aims and Hypotheses

Improving bed net maintenance practices and understanding the nuances of alternative bed net use are essential to developing education interventions aimed at maximizing bed net effectiveness. Education interventions work to improve maintenance and increase use, but their impact on these improvements and the role of specific teaching strategies on their effectiveness is unclear. The following specific aims for this proposed work address gaps in understanding factors related to bed net maintenance behavior, explore bed net repurposing, and clarify the impact of education interventions aimed at improving bed net practices.

Specific Aim 1:

To describe adherence to bed net care and repair recommendations and examine the factors associated with bed net care practices among all bed nets.

16

Specific Aim 2:

To examine observations of nets alternatively used for purposes other than covering a sleeping area and to determine if this alternative use can be properly classified as repurposing rather than misuse.

Hypotheses:

Compared to households without repurposed nets, (1) m ore households with repurposed nets will meet the WHO recommended net ownership threshold of one net for every two household members. (2) The percent of household members who slept under a net will be greater among households with repurposed nets. (3) Malaria diagnosis within the last year of an individual within a household will be lower among households with repurposed nets. 4) Fever in the last two weeks of an individual within a household will be lower among households with repurposed nets.

Specific Aim 3:

To perform a systematic review of the efficacy of educational interventions targeting bed net longevity practices.

Hypothesis:

The efficacy of education campaigns is modified by the specific longevity practices targeted and by the degree to which the intervention applied educational and health behavior theories.

17

C. Role of Author in the Research

To complete Specific Aims 1 and 2, the author planned and conducted secondary data analyses using data generated by Dr. Kacey Ernst and her field team from a cross-sectional study involving household surveys in western Kenya. Oversight for the primary cross-sectional study was deferred to the Kenya Medical Research Institute (KEMRI). For the secondary analyses, the author submitted a Determination of Human Research form to the University of Arizona Human

Subjects Protections Program, which determined that this work did not meet the definition of human subjects research on 12 December, 2017 (Appendix A). These analyses assessed factors associated with adherence to bed net care and repair recommendations and described and analyzed the occurrence of bed net repurposing in the context of malaria prevention. The author wrote and submitted the manuscript titled, Bed net care practices and associated factors in western Kenya to Malaria Journal (under review). The author wrote the manuscript titled, “After those nets are torn most people use them for other purposes”: An examination of bed net repurposing and misuse in western Kenya.

To complete Specific Aim 3, the author planned, developed, and refined search terms for the systematic review in collaboration with Jean McClelland. The author performed all database searches and compiled literature. Jean McClelland deduplicated literature results within EndNote

X8 (Clarivate Analytics, Philadelphia PA). The author acted as Reviewer 1 for both title and abstract screening, Colleen Shelly acted as Reviewer 2 for both screenings, and Kacey Ernst acted as Reviewer 3 to assess discrepancies for both screenings. All data were abstracted and summarized by the author. With contributions by Jean McClelland to the Methods, the author wrote the manuscript titled, “Malaria education interventions addressing bed net care and repair practices: a systematic review”.

18

D. Background

Malaria is a mosquito-borne disease transmitted by mosquitoes, with pathogenesis caused by infection with parasites in the genus , which are contained in the order Haemospororida and family Plasmodiidae [22]. Four major species cause human malaria including , , , and

Plasmodium vivax .

Malaria Pathogenesis and Diagnosis

The hallmark symptom of malaria is the cyclical fever that persists intensely for a few days before subsiding for a time and then returning, caused by the erythrocytic cycle in the parasite’s life cycle. Human malaria may present as uncomplicated or severe malaria.

Uncomplicated malaria presents with generalized symptoms including fever, headache, vomiting, and joint pain. Most severe malaria is caused by P. falciparum, and may involve hemolytic anemia, jaundice, multi-organ damage, convulsions, and death [23]. Young children and those living in areas of sporadic, low, or seasonal transmission are at greater risk for developing cerebral malaria than adults and those living in areas of intense transmission [23, 24].

Diagnosing malaria only through evaluation of clinical symptoms is challenging due to malaria’s often non-specific symptoms [25]. The World Health Organization (WHO) recommends through microscopy or Rapid Diagnostic Testing (RDT) [1].

Microscopy is used to identify Plasmodium blood stage parasites using stained thick and thin peripheral blood smears (PBS) [25]. Microscopy requires laboratory equipment and specialized training to detect the parasites and often has low sensitivity for the diagnosis of malaria [22, 25,

26], because parasite levels in the blood are often not high enough to be detected by laboratory

19

technicians. RDTs overcome some of the challenges of microscopy, as they do not require complex equipment or specially trained healthcare workers. RDTs detect malaria antigen in peripheral blood samples [25]. Other diagnostic techniques including molecular methods are available [25, 26], however microscopy and RDTs are the currently WHO recommended diagnostic methods due to their availability and feasibility in low-resource settings [1].

Global Malaria Epidemiology

Malaria transmission currently occurs where the parasite, vectors, and susceptible human hosts coexist and interact, predominantly in tropical and subtropical areas [23] where the climate is suitable for parasite and vector survival. Though many countries have experienced significant declines in malaria transmission in recent years due to elimination efforts through reductions in estimated cases from 239 million in 2010 to 217 million in 2016, the WHO estimated an increase in global malaria cases from 217 million cases in 2016 to 219 million cases in 2017 [1]. While the epidemiology of malaria is changing over time, sub-Saharan Africa consistently experiences the greatest malaria burden with 92% of all malaria cases in 2017 [World Malaria Report 2018].

Similarly, of the 435,000 worldwide deaths from malaria in 2017, 93% occurred in sub-Saharan

Africa [1].

Within smaller geographic zones, the epidemiology of malaria differs based on four categories of endemicity and seasonality of transmission (seasonal and year-round). A location is considered hypoendemic for malaria if the parasite prevalence among children between 2-10 years of age is 0-10% [27, 28]. A mesoendemic area has a parasite prevalence between 10-50% of children, a hyperendemic area has 50-75% parasite prevalence among children, and a holoendemic area has >75% parasite prevalence among children [27, 28]. Endemicity is

20

important for determining the risk for and presentation of severe malaria, likely due to parasite prevalence and partial acquired immunity. In highly endemic areas, the risk for severe malaria is lower than in less endemic areas [23]. Therefore, maintaining malaria control efforts after major successes in reducing malaria transmission is highly important because reducing transmission to low levels without completely eliminating it may increase the risk of severe malaria [24].

With consistent exposure to Plasmodium parasites, partial immunity can develop, conferring some protection against clinical malaria. Thus, populations with underdeveloped or weakened immune systems are at highest risk for clinical and severe malaria.

Malaria in Kenya

Approximately 70% of the population lives in high-malaria transmission areas (areas with > 1 case per 100,000 population) in Kenya [29]. The distribution of malaria in Kenya reflects its range of climate conditions [30]. Transmission is unstable and epidemic prone in the dry northeastern areas of the country [30], seasonal in the western highlands, and holoendemic in the western lowlands [31]. Malaria in Kenya is due to Plasmodium falciparum, and the primary vectors include Anopheles gambiae, Anopheles arabiensis, Anopheles funestus, and Anopheles menus [29].

Malaria Transmission

The distribution of competent Anopheles mosquito vectors dictates the geographic and climatic limits of malaria transmission. Approximately 20 Anopheles species are important vectors capable of transmitting Plasmodium , while nearly 50 species are capable of transmission

21

[32]. The epidemiology of malaria depends on the parasite’s life cycle, which illustrates how

Anopheles vectors interact with susceptible human hosts for disease transmission to occur.

Within the mosquito vectors Plasmodium must undergo key development phases in both the vector and the human host to complete a highly developed life cycle [33]. While an infected female mosquito takes a blood meal from a susceptible human, sporozoites (the parasite stage infective to humans) travel from the insect’s salivary glands to the human’s blood stream.

The sporozoites travel to the liver where they invade hepatocytes, mature into schizonts, and replicate. The schizonts burst and merozoites are released from the hepatocytes [34].

The merozoites then infect erythrocytes (red blood cells). This stage is known as the erythrocytic cycle, and is responsible for malaria symptoms. During this cycle, more merozoites are produced asexually as merozoites within erythrocytes develop into trophozoites and schizonts, which then rupture releasing more merozoites. These merozoites continue to infect new red blood cells, propagating the erythrocytic cycle. The rupturing of infected red blood cells during this stage is highly synchronized and multitudes of infected erythrocytes burst at one time, causing the intense fever cycles of the disease.

During this time, some of the parasites differentiate into male and female gametocytes, which are the parasite stages that are infective to mosquitos. A susceptible mosquito is then able to ingest the gametocytes while blood-feeding on the infected human. The gametocytes travel to the mosquito’s midgut where they sexually reproduce, first forming ookinetes, then oocysts. The oocysts release sporozoites upon rupturing, which travel to the mosquito’s salivary glands where they can be injected into another human host.

22

Prevention and Control

Historical Efforts

Historically, there have been multiple efforts to eliminate or eradicate malaria through treatment and vector control that have experienced some localized success followed by malaria resurgence [23]. The WHO’s Global Malaria Eradication Program 1955 ran from 1955-1972 and set out to spray dichloro-diphenyl-trichloroethane (DDT) and provide mass treatment with chloroquine on a global scale until drug and insecticide resistance became prevalent and political will receded [23, 35]. Though it was a global eradication plan, there were no efforts to eradicate malaria in sub-Saharan Africa, which has by far the greatest burden of malaria globally [36].

With the development of new treatments and vector control tools, the WHO revised the global malaria strategy in the early 1990s, setting national goals to expand diagnosis and prompt treatment, to control outbreaks, and to build local capacity to prevent and control local malaria conditions [35]. Shortly following the turn of the century, evidence of the effectiveness of -based combination therapies (ACTs), insecticide-treated bed nets (ITNs), and indoor residual spraying (IRS) with pyrethroid insecticides prompted renewed energy and investments toward global malaria elimination [23, 36].

Current Efforts

Currently there is substantial political will, coordination, and collaboration to eliminate malaria as outlined in the WHO Global Technical Strategy for Malaria 2016-2030, which maintains the ultimate goal of reducing the global malaria burden by 90% by the year 2030 [37].

The key emphasis of the strategy is to accelerate malaria elimination efforts in ways specific to local contexts to provide equitable control and prevention efforts to the most vulnerable populations [37].

23

The current recommendations for prevention and control include ensuring universal access to and use of long lasting insecticide treated bed nets (LLINs), with the addition of IRS where appropriate and using insecticides with a different mode of action from bed nets to mitigate the potential for insecticide resistance among vector populations [37]. Other key recommendations include expansion of chemoprevention to include intermittent preventive treatment of pregnant women and infants, and seasonal chemoprevention among children under 5 years of age [37, 38].

The third largest area of recommendations includes ensuring universal diagnostic testing using rapid diagnostic tests or microscopy, and universal treatment [37].

Proper Bed Net Use

Under the purview of the recommendation for universal LLIN use, the three specific aims of this dissertation hone in on proper LLIN use in western Kenya, specifically LLIN maintenance, repurposing and misuse, and education interventions targeting elements of proper LLIN use.

Adherence to recommended net care and repair practices is mixed throughout the literature. A study in Laos examined maintenance behaviors 2-3 years after LLIN distribution and found that

38.2% of households followed the recommended washing frequency, 83% followed drying recommendations, and 32.7% of households with torn LLINs had repaired them [15]. However, the results showed weak associations between maintenance behaviors and malaria knowledge and prevention attitudes [15]. Other studies qualitatively describe motivations for bed net care and repair as a way to care for the family and save money, though few LLIN were actually repaired [39, 40]. Perceptions of social acceptability and color of the LLIN have been associated with washing frequency [41, 42]. In an effort to standardize terminology and further understand bed net maintenance attitudes and practices, the NetWorks project established the term “care and repair” and developed a conceptual framework of the attitudes, intentions, and behaviors related

24

to care and repair practices [7]. While these studies suggest that mixed social, LLIN, and individual factors influence LLIN care and repair, more evidence is needed to identify consistent obstacles that can be targeted for education-based intervention.

As bed net distributions continue, older nets from previous distributions are disposed or repurposed. Differentiating between bed net misuse and repurposing is important for maintaining financial and political will to continue net distributions for malaria prevention.

Recently, the Roll Back Malaria Partnership (RBM) explicitly defined bed net repurposing and misuse, and recommended adoption of the terms and definitions for future net monitoring [43].

Specifically, net repurposing is classified as beneficial or neutral repurposing, depending on whether the new purpose for the net still provides some protection against mosquitos [43]. All repurposing is defined as the alternative use of a net that is no longer used for protecting a sleeping space [43]. Misuse, on the other hand, is the alternative use of a bed net that is still viable for sleeping protection [43].

E. Summary of the Introduction

LLINs are a cornerstone of current malaria prevention efforts, and ensuring their proper use is essential to maintaining their longevity and effectiveness for malaria prevention between distributions. A large focus of the LLIN literature regards improving household net ownership and nightly use, while less is known about adherence to care and repair practices. Specific Aim 1 will evaluate adherence to recommended care and repair practices. A topic often discussed with bed net use is bed net misuse, a phenomenon frequently reported but likely misclassified.

Specific Aim 2 distinguishes between bed net repurposing and misuse and describes both practices among two sites of differential malaria transmission in western Kenya. Furthermore,

Specific Aim 2 examines whether the practice of repurposing is associated with malaria

25

prevention efforts and malaria outcomes. Finally, multiple studies regarding bed net care and repair conclude that health education interventions are necessary to improve adherence to care and repair practices. Specific Aim 3 carefully examines the education intervention literature through a systematic review to evaluate the effectiveness of the interventions on bed net care and repair and malaria outcomes. The overall objective of this work is to obtain a more nuanced understanding of the proper use of bed nets for malaria prevention and the educational strategies commonly used to improve practices.

26

CHAPTER 2

PRESENT STUDY

A. Introduction

Bed nets, specifically LLINs, are a key pillar in malaria elimination efforts, but to have substantial impacts on reducing malaria incidence they must be deployed and used on a mass scale [37]. Mass LLIN distributions every 3-5 years help to achieve high levels of net ownership,

[44, 45, 46] though maintaining universal access within households (1 net per 2 people) remains difficult if net longevity is compromised due to damage or attrition [47].

Bed net care and repair behaviors include the recommended maintenance practices that help nets retain their intended physical and chemical quality [7] and ultimately, their longevity for providing continuous malaria protection. In this dissertation, we describe levels of adherence to net care recommendations, we explore what happens to nets that become too damaged for use, and we review the education intervention methods commonly used to try and improve adherence to bed net care and repair recommendations.

B. Specific Aim 1: Bed net care adherence

To describe adherence to bed net care recommendations and examine the factors

associated with bed net care practices among all bed nets.

1. Methods

Overview: We describe bed net care practices from a cross-sectional study administered in two regions of western Kenya with differing Plasmodium transmission patterns. Though the vast majority (95.3%) of nets were LLINs, the term ‘bed net’ or ‘net’ was chosen because we could not verify the brand or type of all nets in the sample. We developed a bed net care

27

adherence score and examined the relationship between overall adherence to care recommendations and net condition. We then examined the relationship between adherence and attitudes and environmental elements of the NetWorks conceptual model (Figure 1) [7].

Figure 1. Attitudes and environmental factors related to adherence to net care and repair behaviors, adapted from the NetWorks conceptual framework [7].

Study design and data collection: Household surveys were administered during a cross- sectional study in western Kenya. Surveys were administered between June and August of 2015 in two sites (highlands and lowlands) of differing Plasmodium transmission patterns. Data collected included household, individual, and bed net items encompassing demographic information, malaria history, knowledge and perceptions, and bed net attributes, including ownership, universal access, care, and use. Bed net and household level data were used for the present analyses.

Study sites : Kapkangani location encompasses the rural highland sublocations of

Chepsonoi, Kiborgok, and Tindinyo at an altitude range of 1600 – 2100 m above sea level and

28

seasonal Plasmodium transmission following the rainy season from April-May. In this region, the population is primarily comprised of Kalenjin and Luhya ethnic groups. The primary occupation is rural subsistence agriculture, while some work as casual laborers in the commercial tea industry. Miwani is located on the Kano Plain and includes the rural lowland sublocations of

Kabar Central and Kabar West at an altitude of 1200 m above sea level. Transmission is holo- endemic and year-round with seasonal increases. People are primarily of the Luo ethnic group and are subsistence farmers with some employed by area commercial sugarcane and rice growers.

Household selection: The study site was mapped and enumerated prior to sample selection. Households were randomly selected from an enumerated list. Households were oversampled by 20% and then randomly ordered. Visits were made to each household in subsequent order until the required sample size for each village had been met. Male or female heads of household were approached in person by field team staff for recruitment. Households were included if they had resided in the study area for a period of at least one month. Households of all sizes and configurations were included. A household was defined as individuals who regularly eat meals together. Data from 1,217 households (n = 640 in Kapkangani and n = 572 in

Miwani) were used in the present study.

Ethical considerations: The current analysis involved secondary analyses of the de- identified data from the primary study. The primary study was approved by the Kenyan Medical

Research Institute (KEMRI) (SSC #2810) and the University of Arizona ethical review boards with deferral of primary oversight from the University of Arizona to KEMRI. Informed consent was obtained from all household heads. Parental consent was obtained for children less than 7 years of age, and assent was obtained for those 8 to 17 years of age.

29

Bed Net Care Adherence Score: Data on each net owned by the household were collected in addition to overall net care practices reported by the household head. A 7-item care scoring system was developed from factors including net storage, washing, and drying. Each net received a score between 0-2 for each item and the scores were totaled for an overall net care adherence score, ranging from 0-14 in which lower scores indicate better adherence (Table 1).

This scoring system was chosen to be more flexible and nuanced than a strict binary system of good-inadequate care. Similar scoring systems are used in cancer prevention literature [48, 49].

We measured the association between net care adherence score with overall net condition. Field staff classified each net as excellent, good, fair, or poor condition in comparison to standard images provided on the survey forms (Appendix C).

Individual net care elements with ≥ 20% poor adherence (scores of 2) were then selected for further analysis, as these indicate the elements that may need improvement in the study population. We measured the associations of attitudes and environmental factors with adherence to these selected individual net care elements to inform focused interventions.

Table 1. Elements of the bed net care adherence scoring system. Scores of 0 indicate best practice, scores of 1 indicate moderately adherent practice, and scores of 2 indicate non-adherent practice. Wash Where Wash Drying Score Storage Soap Type Ever Retreat Freq Washed Manner Location Never if 0 Washtub or None; Never Not scrubbed/ Indoors; Never LLIN; ever Best Tied up Never bucket; Never Wash or Bar beaten; Never wash if ITN ≥ 6 Practice wash soap wash months

1 Remov ≥ 3 Outdoors in Moderate Lake Detergent _____ Ever if LLIN ed months shade Practice

2 Never if Non- Leave Scrubbed or ITN if net < 3 months Stream or river Bleach In the sun adherent as is beaten age ≥ 6 Practice months

30

Attitudes and Environmental Factors: Knowledge of bed net care is defined as the interviewee’s (household’s) knowledge of particular net care and repair practices taken from an open-ended survey question. Participants were asked what things they should do to take care of their bed net. Whether participants mentioned a particular net care practice indicated knowledge of that practice. Risk perception operates both intra and inter-personally [50]. Perception of malaria risk was measured from participant’s perception of the seriousness of malaria for their family (intrapersonal) and for their community (interpersonal). Both of these items are scored on a 5-point likert scale where 1 indicates ‘not at all serious’, and 5 ‘extremely serious’. We added these items to obtain an overall risk perception score (range 2-10). Confidence in the ability to prevent malaria indicates how well participants felt they are able to protect their family from malaria from a 5-point likert scale where 1 indicates ‘not at all’ and 5 ‘extremely well’.

Household structural factors include both family and physical structures. Family structure includes the bed net user , which is a series of binary variables indicating which household members slept under the net. Net user variables measured in years include: child < 5, child 6-18, adult male 19-50, adult female 19-50, adult male 50+, and adult female 50+. Source of Net Care

Knowledge includes a series of binary variables indicating where the interviewee learned net care. These variables include: clinic staff, community education session, or during a net distribution. Self-reported ability to pay for a bed net is a binary variable where ‘yes’ indicates the household could afford a bed net if one was not provided for free. The ability to pay for a net may influence how well nets are cared for depending if it is easy or difficult to obtain a new net.

Physical structural factors include universal access , defined as households that meet the

WHO recommendation of one net for every two household members, amount of shade in the house compound , defined as no shade, little, half-shaded, or very shaded, and animals sleep in

31

structure at night , a binary variable where ‘yes’ indicates an animal shares the sleeping structure space at night.

Statistical Analyses: Stata 12.1 (College Station, TX) was used for all analyses with

α=0.05 at the bed net level. Due to the clustered, dependent nature of the data (multiple nets per household), Generalized Estimating Equations (GEEs) were used to measure the association between net condition and overall net care adherence score, and to measure the associations between the identified key independent variables and binary bed net care adherence elements.

Binary care adherence elements were used to assess associations with non-adherent practices, thus best and moderate practices were grouped together to serve as the reference category. GEEs are flexible and ideal for analyzing dependent data that do not meet the assumptions of classical regression techniques [51, 52].

These models can also be used for ordinal discrete data [53]. The GEE parameters used for adherence score analyses were adjusted for bed net age and included Gaussian family, identity link function, exchangeable correlation structure, and robust standard errors. The parameters used for individual net care element analyses included binomial family, identity link function, exchangeable correlation structure, and robust standard errors. GEE models assessing washing frequency were adjusted for net age because this element was measured as “when the net was last washed”. Because the highlands and lowlands differed on nearly all demographic characteristics, level of malaria endemicity, and history of LLIN distributions and community education programs, GEEs were run separately for each site.

2. Results

Net ownership was greater in the lowlands where 98% of households owned at least one net compared to 68.6% of households in the highlands (Table 2). Most households had one

32

sleeping structure, though households in the highlands tended to have both more individuals in the household as well as more sleeping structures than in the lowlands, where 29.6% vs. 10.2% had more than one structure, respectively. Females are primarily responsible for net care [54], thus it is important to understand attributes of female household heads. In both sites, most female household heads had some primary education or have completed primary education (Table 2).

Table 2: Household characteristics of study sample by site Highlands Lowlands (643 households) (574 households) N (%) N (%) People in household 1-3 241 (37.5) 382 (66.6) 4-6 296 (46.0) 169 (29.4) ≥ 7 106 (16.5) 23 (4.0)

Sleeping structures in household 1 453 (70.5) 516 (89.9) 2 157 (24.4) 50 (8.7) 3 27 (4.2) 7 (1.3) 4 5 (0.8) 1 (0.2) 5 1 (0.2) 0

Nets per household 0 203 (31.6) 11 (1.9) 1 167 (26.0) 318 (55.4) 2 154 (24.0) 185 (32.2) 3 86 (13.4) 45 (7.8) 4 28 (4.4) 14 (2.4) ≥ 5 5 (0.8) 1 (0.2) Education of the female household head None 55 (8.6) 22 (3.8) Some Primary 261 (40.6) 200 (34.8) Completed Primary 152 (23.6) 190 (33.1) Some Secondary 66 (10.3) 57 (9.9) Completed Secondary 30 (4.7) 39 (6.8) Any Tertiary 20 (3.1) 8 (1.4) No female household head 42 (6.5) 50 (8.7)

LLIN Care Adherence : Bed nets in the highlands had better overall care adherence scores compared to the lowlands. Adherence in the highlands was quite good, where nets had a median adherence score of 2, compared to scores in the lowlands, while still good, had a less

33

adherent score of 4 (possible range 0-14) (Table 3). When each care element was measured individually, 2 elements in the highlands and 3 in the lowlands had inadequate adherence (<80% of nets cared for according to recommendations) (Figure 2). Adherence to washing location

(bucket vs. river or stream), washing manner (gently washed or scrubbed/beaten), soap type, and retreating practices were adequate in both sites and excluded from further analyses. However

<80% of nets met adherence recommendations for daily storage, washing frequency, and drying location in at least one of the sites (highlands or lowlands). Frequent net washing is not recommended, and nets in the lowlands were washed more frequently (77% <3 months) than the highlands (23% <3 months). In the lowlands, 32% of nets were inappropriately dried in the sun compared to only 4% in the highlands (Figure 2).

34

Figure 2. Individual bed net care adherence score components by site. Scores for components in the highlands are the left-sided, bold colored bars, and the scores for components in the lowlands are the right-sided, pale colored bars.

100

90

80

70

60

50

40

Percentof BedNets 30

20

10

0 Storage Washing freq Where wash Soap type Wash manner Dry location Retreat

Highlands most adherent Lowlands most adherent Highlands moderate adherence Lowlands moderate adherence Highlands non-adherent Lowlands non-adherent

35

Table 3. Distributions of net care adherence and attitudes and environmental factors Highlands CDE Lowlands CDE Median (Range) Median (Range) or # (%) or # (%) p-value A Overall net care adherence score F 2 (0-8) 4 (0-9) <0.001*

Overall net condition <0.001* G Excellent 522 (59.9) 594 (67.2) Good 227 (26.1) 162 (18.3) Fair 98 (11.3) 85 (9.6) Poor 18 (2.1) 23 (2.6) Knowledge and Attitudes What kinds of things should you do to take care of your bed net? C Tie up in the morning 177 (27.6) 94 (16.4) <0.001* Don’t wash or wash rarely 25 (3.9) 44 (7.7) 0.005* Don’t use detergent 135 (21.1) 37 (6.5) <0.001* Don’t beat on the rocks 149 (23.2) 48 (8.4) <0.001* Don’t hang in the sun 382 (59.6) 281 (49.0) <0.001* Retreat with insecticide 213 (33.2) 271 (47.2) <0.001* Tie up all holes 166 (25.9) 418 (72.8) <0.001*

Perception of malaria risk C (2=low, 10=high) 6 (2, 10) 7.5 (2, 10) <0.001* B

Confidence in ability to prevent malaria C ( Likert scale) 2 (1, 5) 3 (1, 5) <0.001* Environmental Factors Household Family Structure: Net User (years) D Child < 5 219 (25.0) 232 (26.2) 0.386 Child 6-18 332 (38.0) 312 (35.3) 0.387 Adult Male 19-50 169 (19.3) 218 (24.7) 0.003* Adult Female 19-50 332 (38.0) 337 (38.1) 0.386 Adult Male > 50 61 (7.0) 58 (6.6) 0.811 Adult Female > 50 113 (12.9) 101 (11.4) 0.416

Ability to pay for a net C 441 (68.8) 494 (87.1) <0.001*

Source of net care knowledge C Clinic 328 (37.5) 372 (42.1) 0.127 Community event 176 (20.1) 340 (38.5) <0.001* Net distribution 179 (20.5) 3 (0.34) <0.001*

Universal Access (1 net:2 members) 239 (37.2) 414 (72.1) <0.001*

36

Amount of shade in house compound C,G <0.001* No shade 60 (9.3) 106 (18.5) Little shade 321 (49.9) 255 (44.4) Half shaded 229 (35.6) 159 (27.7) Very shaded 27 (4.2) 38 (6.6)

Sleeping structure has Earth floor E <0.001* Earth floors 916 (73.5) 893 (87.0) No earth floors 331 (26.5) 134 (13.0)

Animals present in sleeping structure at night E 0.004* Animals present 164 (13.2) 179 (17.4) No animals present 1083 (86.8) 848 (82.6) A Two sample test of proportions B Nonparametric equality of medians test C Household level data (Highlands n=641, Lowlands n=574) D Bed net level data (Highlands n=874, Lowlands n=884) E Sleeping structure level data (Highlands n=1247, Lowlands n=1027) *Statistically significant at alpha 0.05; F Lower scores indicate better net care and repair adherence practices; G Pearson chi-square test

Attitudes and Environmental Factors: Households in both sites were most knowledgeable about proper drying practices (drying nets out of the sun). In the highlands,

59.6% of households specifically mentioned nets should not be dried in the sun compared to

49.0% of households in the lowlands (Table 3). This knowledge translated to practice in the highlands (4% of nets dried in the sun), but less so in the lowlands (32% of nets dried in the sun)

(Figure 2). Notably in both sites only 25 (3.9%) of households in the highlands and 44 (7.7%) in the lowlands mentioned they should not wash their nets or wash them infrequently, suggesting that messaging regarding proper washing frequency has not been well communicated in these sites (Table 3). Perception of malaria risk and reported confidence in ability to prevent malaria were both greater in the lowlands compared to the highlands (Table 3).

Nearly a quarter of nets were used by at least one child <5 years in both sites. The distribution of net users is similar in both sites, though more nets were used by males 19-50 years in the lowlands compared to the highlands. Significantly more households in the lowlands

37

(87.1%) reported the ability to afford a bed net compared to households in the highlands (68.8%)

(Table 3).

Most households learned about net care and repair practices from clinics in both sites.

Approximately 20% of households learned from community events and nearly 20% from net distributions in the highlands, in contrast with the lowlands where 38.8% learned from community events and only 0.34% learned from a net distribution.

Factors Associated with Net Care Adherence: There was no association between net condition and overall net care adherence score in the highlands (Table 4). However in the lowlands, poor net condition was associated with worse overall net care adherence score after adjustment for net age. (Table 4).

Table 4. GEE results of association between net condition and overall net care adherence score, adjusted for net age Highlands Lowlands Overall Net Condition Coefficient 95% CI Coefficient 95% CI Excellent Ref Ref Good 0.48 -0.34, 1.29 0.46 -0.45, 1.36 Fair 0.61 -0.16, 1.38 0.64 0.42, 0.86 Poor -0.27 -1.75, 1.22 0.96 0.50, 1.41 Associations measured using GEE models with net care adherence as an ordinal variable.

Different attitudes and environmental factors were associated with elements of net care in different ways (Table 5). Knowledge of net care, and nets used by children <5 years, children 6-

18 years, adult males 19-50 years, and adult males >50 years were associated with better adherence (lower scores). Nets used by adult females >50 years were associated with worse adherence (higher scores). Many independent variables had positive associations with some elements of the net care adherence score, and negative associations with others, illustrating the complexity of the factors that influence net care behaviors.

38

As compared to those properly stored during the day, nets that remained hanging were more likely to belong to a household with high confidence to prevent malaria and less likely to have been used by a child <5 years in both sites (Table 5). This suggests that households with a sense of self-efficacy to prevent malaria were less likely to properly store bed nets, and nets used by young children were stored properly during the day. Other factors were positively and negatively associated with storage practices, but different across sites (Table 5).

Households that learned about net care from community events, that shared a sleeping structure with animals, and that had nets used by adult males tended to adhere to washing frequency recommendations (Table 5). We hypothesized this difference could be due to housing structural factors where dustier environments may be associated with more frequent washing.

However, no association was found between having dirt floors and washing behaviors in either site (Table 5).

There was a remarkable difference in drying practices between sites. In the lowlands,

32% of nets were dried in the sun compared to only 4% in the highlands (Figure 5). Drying in the shade may not be feasible for individuals depending upon the environment surrounding their compound, so we examined whether amount of shade was associated with drying practices, but found no association, though households in the lowlands had less shade compared to households in the highlands (Table 5). Interestingly, nets improperly dried in the sun were less likely to belong to a household with high knowledge of proper drying practices, indicating that increasing knowledge may positively influence this practice. Source of net care knowledge was similarly associated, suggesting that community events were effective in their messaging regarding proper drying practices (Table 5).

39

Table 5. Odds ratios (95% Confidence Intervals) of univariate GEE models measuring associations between attitudes and environmental factors with binary net care practices Highlands Lowlands

Attitudes and Environmental Factors Storage Wash Freq A Storage Wash Freq A Dry Location Knowledge of appropriate care 0.86 (0.80, 0.92) 0.86 (0.80, 0.92)

Perception of malaria risk 1.03 (1.02, 1.05)

Confidence to prevent malaria 1.06 (1.02, 1.11) 1.08 (1.04, 1.13)

Net User Child <5 0.92 (0.86, 0.99) 0.93 (0.87, 0.98) Child 6-18 0.91 (0.85, 0.97) Adult male 19-50

Adult female 19-50 Adult male >50 0.93 (0.91, 0.96) Adult female >50 1.27 (1.14, 1.40)

Ability to buy net 1.10 (1.01, 1.20) 1.15 (1.04, 1.29)

Source of Net Care Knowledge Clinic 0.87 (0.82, 0.93) 1.10 (1.02, 1.19) Community event 0.87 (0.81, 0.93) 0.93 (0.90, 0.96) 1.24 (1.17, 1.33) 1.07 (1.01, 1.13) 0.87 (0.81, 0.94) Net distribution 1.20 (1.09, 1.33)

Universal Access (1 net: 2 people) 1.11 (1.03, 1.19)

Number of trees around compound N/A N/A N/A N/A

Sleeping Structure Attributes Earth floors 1.07 (1.01, 1.14) 1.14 (1.04, 1.26) Animals occupy sleeping structure 1.13 (1.03, 1.24) 0.95 (0.92, 0.97) 0.92 (0.86, 0.99) A Adjusted for net age; Blue indicates associations with better adherence, orange indicates associations with worse adherence, White indicates no association.

40

C. Specific Aim 2: Bed net repurposing and misuse

To examine observations of nets alternatively used for purposes other than covering a sleeping area and to determine if this alternative use can be properly classified as repurposing rather than misuse.

Hypotheses: Compared to households without repurposed nets, (1) More households with repurposed nets will meet the WHO recommended net ownership threshold of one net for every two household members. (2) The percent of household members who slept under a net will be greater among households with repurposed nets. (3) Malaria diagnosis within the last year of an individual within a household will be lower among households with repurposed nets. (4) Fever in the last two weeks of an individual within a household will be lower among households with repurposed nets.

1. Methods

Definitions

Household surveys were conducted in 1,217 households among three sublocations of the highlands (Chepsonoi, Kiborgok, and Tindinyo) and among two sublocations of the lowlands

(Kabar West and Kabar Central) from March to December, 2015. Analyses were conducted separately for the highlands and lowlands because the two sites differ in terms of malaria transmission intensity, which can impact individual malaria prevention behaviors. Because the majority of surveys were conducted during the spring and summer months, households surveyed from March to May are classified as early surveys, those from June to July are mid surveys, and those from August to December are late surveys. Households where the study team observed any

41

bed nets being used for purposes other than covering sleeping areas were determined to have net repurposing.

Household heads were asked about various malaria perceptions including how serious a problem malaria is for their family (Likert scale: not at all serious to extremely serious) and the likelihood a household member is to contract malaria within the next month (Likert scale: not at all likely to extremely likely). To get a sense of socioeconomic status, households were categorized into wealth quartiles based on an inverse frequency of ownership-weighted asset- based wealth index. Additionally, households were asked whether they would be able to afford a bed net if one was not provided for free, measured as a binary Yes/No response.

Households that met the WHO-recommended net ownership threshold included those with at least one (non-repurposed) net for every two household members. The percent of household members who slept under a net was calculated using the number of household members who reported to have slept under a bed net the night prior to the household survey.

Malaria diagnosis within the last year or fever within the last two weeks is defined as at least one household member who was reported to have been diagnosed with malaria within one year or at least one household member with self-reported fever within two weeks prior to the household survey, respectively. Malaria diagnosis and fever were self-reported or reported by the household head regarding other household members. We indicated binary household ownership of animals including cattle, sheep, goats, chickens, and dogs. We also measured whether households reported observing other households in their community repurpose bed nets with a binary

Yes/No variable.

42

Statistical Analyses

We used the Pearson Chi-Square test to determine whether the distribution of various household characteristics and malaria perceptions differed between households with and without repurposed bed nets in the highlands and lowlands. To test the four stated hypotheses and to measure the magnitude of associations of structural and social factors with household bed net repurposing, we used univariate logistic regression models. Logistic regression assumptions were tested where appropriate, including independent observations and linearity in the log-odds.

Because a household’s local community likely influences both the selected independent variables and bed net repurposing, we adjusted for sublocation. We used Stata v12 (College Station, TX) and α=0.05 for all analyses.

2. Results

Household characteristics and malaria perceptions

There were no statistically significant differences in the distributions of any household characteristics between households with and without repurposed nets in both the highlands and lowlands (Table 6). Most households in the highlands have 4-6 members, 1-2 sleeping structures,

≤ 2 bed nets, a female household head that has some or completed primary education, and fall into the second wealth quartile. Most households in the lowlands have 1-3 members, 1 sleeping structure, 1-2 bed nets, a female household head that has some or completed primary education, and fall into the fourth wealth quartile. Additionally, 69.1% of households in the highlands and

87.1% in the lowlands reported the ability to afford a bed net if one was not provided for free

(Table 6).

43

Most households reported that malaria is a slightly or somewhat serious problem for their family in the highlands, and a slightly-to-serious problem in the lowlands, with no significant differences between households with and without repurposed bed nets. In the highlands and lowlands, most (91.9% and 86.6%, respectively) households report having no problems in hanging bed nets or other difficulties with net use (Table 6). A higher proportion of repurposed nets were observed earlier in the study period with a decreasing trend during the mid and late study period in both sites, though the distribution of households in each study period with and without repurposed nets was not significantly different in either site.

Observation of repurposed bed nets

Fewer households repurposed nets in the highlands compared to the lowlands. Of 643 households in the highlands, we observed 52 (8.1%) with repurposed nets, while 184 (33.0%) of the 574 lowlands households had repurposed nets observed by the study team. Some households repurposed more than one net, indicating clustering of net repurposing within households, as there were 67 repurposed nets among the 52 highlands households and 312 repurposed nets among the 184 lowlands households that had repurposed nets. Clustering of net repurposing may also occur at the community level, so we described clustering of net repurposing observations within sublocations (Table 6). In the highlands, Chepsonoi had the highest prevalence of observed repurposing and Kiborgok had the lowest prevalence (10.0% and 5.9% of households respectively). The distribution of repurposed nets between the highland sublocations was significantly different. In the lowlands, the prevalence of repurposing was similar between Kabar

Central and Kabar West (33.5% and 30.2%, respectively) (Table 6).

44

Table 6. Household characteristics and malaria perceptions comparing households with and without repurposed bed nets A Highlands Lowlands

Any No Any No Repurposing Repurposing Repurposing Repurposing Total (n= 52) (n=587) P-Value A Total (n=184) (n=374) P-Value A People per household 0.629 0.183 1-3 240 21 (8.8) 219 (91.3) 372 120 (32.3) 252 (67.7) 4-6 293 25 (8.5) 268 (91.5) 164 53 (32.3) 111 (67.7) ≥ 7 106 6 (5.7) 100 (94.3) 22 11 (50.0) 11 (50.0) # Sleeping structures 0.537 0.349 1 450 38 (8.4) 412 (91.6) 501 167 (33.3) 334 (66.7) 2 156 14 (9.0) 142 (91.0) 49 15 (30.6) 34 (69.4) 3 27 0 27 (100) 7 1 (14.3) 6 (85.7) 4 5 0 5 (100) 1 1 (100) 0 5 1 0 1 (100) 0 0 0 Nets per household 0.236 0.470 0 202 11 (5.4) 191 (94.6) 10 2 (20.0) 8 (80.0) 1 167 17 (10.2) 150 (89.8) 309 95 (30.7) 214 (69.3) 2 152 15 (9.9) 137 (90.1) 179 65 (36.3) 114 (63.7) 3 85 7 (8.2) 78 (91.8) 45 16 (35.6) 29 (64.4) 4 28 1 (3.6) 27 (96.4) 14 5 (35.7) 9 (64.3) ≥5 5 1 (20.0) 4 (80.0) 1 1 (100) 0 Education level female household head 0.497 0.076 None 55 5 (9.1) 50 (90.9) 22 10 (45.5) 12 (54.5) Some primary 261 25 (9.6) 236 (90.4) 193 54 (28.0) 139 (72.0) Completed primary 148 10 (6.8) 138 (93.2) 187 77 (41.2) 110 (58.8) Some secondary 66 6 (9.1) 60 (90.9) 54 16 (29.6) 38 (70.4) Completed secondary 30 0 30 (100) 39 10 (25.6) 29 (74.4)

45

Any tertiary 20 3 (15.0) 17 (85.0) 8 1 (12.5) 7 (87.5) No female household head 42 1 (2.4) 41 (97.6) 47 14 (29.8) 33 (70.2) How serious a problem malaria is for the family 0.383 0.938 0.273 Not at all serious 22 4 (18.2) 18 (81.8) 5 2 (40.0) 3 (60.0) Slightly serious 357 26 (7.3) 331 (92.7) 143 43 (30.1) 100 (69.9) Somewhat serious 189 16 (8.5) 173 (91.5) 100 33 (33.0) 67 (67.0) Serious 64 5 (7.8) 59 (92.2) 283 96 (33.9) 187 (66.1) Extremely serious 5 1 (20.0) 4 (80.0) 26 9 (34.6) 17 (65.4) Problems ever using a net 0.184 0.696 0.438 Yes 43 6 (14.0) 37 (86.0) 72 25 (34.7) 47 (65.3) No 491 40 (8.1) 451 (91.9) 466 151 (32.4) 315 (67.6)

Likelihood a HH member will get malaria in next month 0.687 0.526 Not at all likely 26 1 (3.8) 25 (96.2) 4 2 (50.0) 2 (50.0) Somewhat likely 66 8 (12.1) 58 (87.9) 59 21 (35.6) 38 (64.4) Equally likely as unlikely 142 11 (7.7) 131 (92.3) 277 88 (31.8) 189 (68.2) Likely 396 31 (7.8) 365 (92.2) 164 50 (30.5) 114 (69.5) Extremely likely 1 0 1 (100) 50 21 (42.0) 29 (58.0) Wealth quartile 0.809 0.234 1st Quartile 160 12 (7.5) 148 (92.5) 118 34 (28.8) 84 (71.2) 2nd Quartile 211 18 (8.5) 193 (91.5) 103 26 (25.2) 77 (74.8) 3rd Quartile 128 8 (6.3) 120 (93.8) 156 55 (35.3) 101 (64.7) 4th Quartile 129 12 (9.3) 117 (90.7) 187 69 (36.9) 118 (63.1) Ability to afford a net 0.753 0.888 Yes 442 36 (8.1) 403 (91.2) 494 157 (31.8) 322 (65.2) No 198 15 (7.6) 182 (91.9) 73 23 (31.5) 49 (67.1) Survey Timing 0.101 0.071 Early (March-May) 14 3 (21.4) 11 (78.6) 4 3 (75.0) 1 (25.0) Mid (June-July) 415 36 (8.7) 379 (91.3) 349 125 (35.8) 224 (64.2)

46

Late (Aug-Dec) 210 13 (6.2) 197 (93.8) 205 56 (27.3) 56 (27.3) Sublocation 0.038* 0.284 Chepsonoi 221 22 (10.0) 199 (90.0) Kiborgok 204 12 (5.9) 192 (94.1) Tindinyo 217 18 (8.3) 199 (91.7) Kabar Central 316 106 (33.5) 210 (66.5) Kabar West 258 78 (30.2) 180 (69.8) Owns 1 net:2 members 0.211 0.668 Yes 238 21 (8.8) 218 (91.6) 184 134 (72.8) 50 (27.2) No 400 31 (77.5) 369 (92.3) 374 272 (72.7) 102 (27.3) % HH slept under a net 0.115 0.610 0-25% 227 11 (4.8) 216 (95.2) 15 4 (26.7) 11 (73.3) 26-50% 67 8 (11.9) 59 (88.1) 12 2 (16.7) 10 (83.3) 51-75% 46 5 (10.9) 41 (89.1) 16 5 (31.2) 11 (68.8) 76-100% 293 28 (9.6) 265 (90.4) 515 173 (33.6) 342 (66.4) Malaria in the last year B 0.995 0.865 Yes 475 39 (8.2) 436 (91.8) 430 141 (32.8) 289 (67.2) No 158 13 (8.2) 145 (91.8) 128 43 (33.6) 85 (66.4) Fever in the last 2 weeks B 0.624 0.480 Yes 176 16 (9.1) 160 (90.9) 306 97 (31.7) 209 (68.3) No 456 36 (7.9) 420 (92.1) 252 87 (34.5) 165 (65.5) Own cattle 0.974 0.852 Yes 355 29 (8.2) 326 (91.8) 288 96 (33.3) 192 (66.7) No 284 23 (8.1) 261 (91.9) 270 88 (32.6) 182 (67.4) Own sheep 57 6 (8.1) 68 (91.9) 0.992 0.754 Yes 74 6 (8.1) 68 (91.9) 117 40 (34.2) 77 (65.8) No 565 46 (8.1) 519 (91.9) 441 144 (32.7) 297 (67.3) Own goats 0.191 0.028* Yes 56 2 (3.6) 54 (96.4) 175 69 (39.4) 106 (60.6) No 583 50 (8.6) 533 (91.4) 383 115 (30.0) 268 (70.0) Own chickens 0.117 0.000*

47

Yes 497 45 (9.1) 452 (90.9) 441 163 (37.0) 278 (63.0) No 141 7 (5.0) 134 (95.0) 117 21 (17.9) 96 (82.1) Own dogs 0.004* 0.212 Yes 155 4 (2.6) 151 (97.4) 325 114 (35.1) 211 (64.9) No 484 48 (9.9) 436 (90.1) 233 70 (30.0) 163 (70.0) HH reports seeing others repurpose bed nets 0.000* 0.011* Yes 450 48 (10.7) 402 (89.3) 525 180 (34.3) 345 (65.7) No 187 4 (2.1) 183 (97.9) 32 4 (12.5) 28 (87.5) APearson’s Chi-Square test; *Statistically significant p<0.05; BAt least one member in the household

48

Most nets in the highlands were repurposed for use outdoors (77.6%) (Figure 3A). The most common functions of observed repurposed nets were as chicken coops or fences (33% and

37%, respectively). Nets repurposed for use indoors were used as curtains or to cover the ceiling or bathrooms (Figure 3A). Similarly in the lowlands, most nets (69.8%) were repurposed for use outdoors, primarily to cover trees (22%), as chicken coops (20%), or as fences (25%) (Figure

1A). Nets repurposed for use indoors covered bathrooms or were cut up and used as washing sponges. Of the repurposed nets, 7 (10.4%) were classified as beneficial repurposing and 60

(89.6%) were classified as neutral repurposing in the highlands. In the lowlands, 28 repurposed nets (9.0%) were repurposed beneficially and 284 nets (91.0%) were neutral.

Twenty-six of highlands households and 127 of lowlands households with repurposed nets reported why they repurposed nets. In both sites, the vast majority of nets were repurposed because households determined them to be either torn, old and worn out, or both torn and old

(Table 7).

Table 7. Household-reported reasons for repurposing bed nets Highlands Lowlands Representative Quote n (%) N=26 n (%) N=127 households households Net was torn/too many “The [net] had many holes that could not holes 10 (38.5) 23 (18.1) be sewn up” “[The net] had lasted a long period of time...more than 5 years therefore [we] Net was old and worn out 5 (19.2) 78 (61.4) decided to use it as a garden fence” Net was old and had “[The net] had many holes and [was] holes 7 (26.9) 14 (11.0) old” Damage by outside “Bed net was eaten by a rat” factors 1 (3.8) 3 (2.4) “[The net was used] to prevent hens Needed to protect crops 2 (7.7) 3 (2.4) from attacking the vegetables” Did not want to throw “[It is] better that way instead of away 1 (3.8) 2 (1.6) throwing away after it was torn” “After those nets are torn most people Other 0 4 (3.1) use them for other purposes”

49

Reports of net repurposing in the community

Though more prevalent in the lowlands than the highlands (94.3% vs. 70.6% of households, respectively), it was very common for households to report seeing other households in their community repurpose bed nets in both sites (Table 6). Similarly to the observed net repurposing, more households reported noticing community members repurposing bed nets in

Chepsonoi (86.0% of households) compared to Kiborgok (49.0% of households). In the lowlands, more households reported observing net repurposing in their community in Kabar

Central (97.8% of households) compared to Kabar West (88.8% of households). Households in the study population reported observing similar uses for repurposed nets that were observed by the study team, with uses in chicken coops, as fences, and as washing sponges among the most common (Figure 3B). However, households reported that others also use nets for fishing, as decoration, as ropes, and as sieves, while these particular uses were not observed by the study team among the participant households (Figure 3B).

50

Figure 3: Various uses of repurposed bed nets as observed by the study team (3A) and as observed by study participants (3B) in the highlands and lowlands of western Kenya *Images from publicdomainvectors.org 3A

70

60

50

40

30

20

10 Percentof repurposed bednets

0 Covering Chicken Fence Cover Cover Ceiling Curtains Washing Fishing trees coop bathroom vegetables sponge

Highlands (N=67 nets) Lowlands (N=312 nets)

3B 90

80

70

60

50

40

30 Percentof households 20

10

0

Highlands (n=643 households) Lowlands (n=574 households)

51

Bed net misuse

Repurposed nets that are fully intact (could effectively be used for preventing mosquito bites) are considered to be misused nets. Though we were unable to determine the insecticidal efficacy of repurposed nets to know if they were still efficacious in repelling and killing mosquitoes, 1 of the 67 repurposed nets in the highlands and 3 of the 312 repurposed nets in the lowlands had no holes, suggesting only 1.8% and 1.1% of repurposed nets in the highlands and lowlands respectively were misused (Figure 4). Although we are unable to determine whether holes appeared before or after repurposing, the vast majority of repurposed nets in both sites had many holes (Figure 4).

Bed net misuse can also occur among nets inside the home that are generally used for sleeping. Among 874 functional nets inside homes in the highlands, only 7 nets (0.8%) were self- reported to have ever been used for something other than mosquito protection while sleeping.

Similarly, this was reported for only 6 nets (0.7%) of 884 functional nets in the lowlands.

Figure 4. Observed number of holes among repurposed bed nets in the highlands and lowlands of western Kenya 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Highlands (n=67)

Lowlands (n=312)

None Few Moderate Many

52

Impact of bed net repurposing on malaria prevention efforts

We tested 4 hypotheses to assess the impact of net repurposing on malaria prevention efforts and malaria outcomes. In both sites, there was no association between observed net repurposing and household (1) net ownership, (2) net use, (3) malaria in the last year, or (4) fever in the last two weeks (Figure 5). There were 27 households in the highlands (4.2%) and 13 households in the lowlands (2.3%) that had repurposed nets and at least one household member who did not use a net. Other factors were associated with observed net repurposing including animal ownership and social factors. Animal ownership was positively associated with repurposing in the lowlands, where households with repurposed nets had 1.53 times the odds of owning goats and 2.74 times the odds of owning chickens compared to households with no repurposed nets (Figure 5). Though not statistically significant, there was a trend of increasing presence of repurposing with decreasing net use in the household in the highlands (Figure 5). As households report lower proportions of members sleeping under nets, they also had more repurposed bed nets. Notably, in both sites, households that repurposed nets had significantly greater odds of reporting that they have observed other community households repurposing nets

(OR 7.27 95% CI 2.22, 23.79 highlands, OR 3.59 95% CI 1.23, 10.51 lowlands) compared to households with no repurposed nets (Figure 5).

53

Figure 5. Univariate logistic regression results (odds ratios and 95% confidence intervals) assessing factors associated with the presence of bed net repurposing at the household level, adjusted for sublocation

54

D. Specific Aim 3: Systematic review

To perform a systematic review of the efficacy of educational interventions targeting bed net longevity practices.

Hypothesis: The efficacy of education campaigns is modified by the specific longevity practices targeted and by the degree to which the intervention applied educational and health behavior theories.

1. Methods

We conducted a systematic literature review with the objective of assessing whether education interventions increase knowledge of or adherence to recommended bed net care and repair practices. We evaluated net durability or longevity as a potential secondary outcome.

We searched for interventions targeting either ITNs or LLINs. Though ITNs are no longer distributed, interventions focusing on ITNs were included because we were interested in the change in knowledge or practices, and because many of the recommendations for ITNs and

LLINs overlap. We hypothesized that the efficacy of education campaigns is modified by the specific care and repair practices targeted and by the degree to which the intervention applied educational and health behavior theories.

Search strategy

The protocol was registered through PROSPERO International Prospective Register of

Systematic Reviews on 30 April 2019 (registration number CRD42019123932). We identified studies through a literature search using controlled vocabulary terms (MeSH-medically controlled vocabulary) and key word combinations for “education intervention” terms,

55

” terms and malaria (Table 8, Appendix H). The search was conducted on February

13, 2019 in MEDLINE PubMed (1946-2019, 13 February), Embase (1947-2019, 13

February), Cochrane Controlled Register of Trials CENTRAL (1966-2019, 13 February),

Bibliomap (indexing date not reported – 2019, February 13), ProQuest Dissertations & Theses

(1982-2019, 13 February), Trials Register of Promoting Health Interventions TRoPHI (2004-

2019, 13 February) and Clarivate Web of Science Core Collection (1900-2019, 13 February) databases (Table 9). Citations were imported into EndNote X8 (Thomas Reuters) for deduplication and management.

Table 8. Search terms adapted for use in database searches for peer-reviewed studies Category Search Terms Bed nets “mosquito nets” [Mesh*] “LLIN” “LLINs” “long lasting insecticide treated nets” “long lasting insecticide treated bednets” “long lasting insecticide treated bed nets” “long lasting insecticidal nets” “long-lasting insecticide-treated nets” “ITN” “ITNs” “Insecticide treated nets” “insecticide treated bednets” “insecticide treated bed nets” “Icon Life” Yorkool Olyset Netprotect Permanet Malaria “Malaria” [Mesh] Malaria Education interventions “health education” [Mesh] “health behavior” [Mesh] “intervention” “education intervention” “malaria education” “education campaign” “behavior change” “behaviour change” *MeSH includes the medically controlled vocabulary for use in the PubMed database

56

The searches identified 6507 studies prior to removing duplicates. Deduplication in

EndNote X8 (Thomas Reuters) resulted in 5927 studies. Studies were reviewed following the protocol set forth by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses

(PRISMA) guidelines. Initial screening of titles and abstracts was completed by 2 reviewers (ES,

CS) with a third reviewer (KE) assessing discrepancies, according to the following study inclusion criteria: malaria education intervention, communication campaign, or behavior change campaign pertaining to bed net care and repair practices for ITNs or LLINs. We managed data in

Google Sheets (Google, Mountain View, CA) with individual reviewer columns hidden to other reviewers to maintain a blinded screening process.

After initial screening, full text articles for 446 studies were retrieved and screened by 2 reviewers (ES, CS) with a third reviewer (KE) assessing discrepancies through the same blinded process as initial screening. Studies were assessed according to the same inclusion criteria as in the initial screening process with the application of the following exclusion criteria (Figure 6): malaria intervention that does not pertain to bed net care and repair, education intervention not relevant to malaria, review papers, conference abstracts, and commentary/opinion pieces.

Grey literature was identified through relevant organizations that support bed nets for malaria prevention (Table 9). Contents of each of the organization websites including all tabs within the websites were searched between 12-20 March, 2019. All records related to malaria education and/or bed net care and repair practices were retrieved and reviewed to be summarized.

57

Table 9. Databases and malaria organizations searched for malaria education intervention records Peer-Reviewed Literature Databases Grey Literature: Malaria Organizations BiblioMap African Leaders Malaria Alliance Cochrane Central Register of Controlled Against Malaria Foundation Trials (CENTRAL) Embase Bill & Melinda Gates Foundation ProQuest Dissertations and Theses Global The Global Fund PubMed Imagine No Malaria Trials Register of Promoting Health Kenya NGOs Alliance Against Malaria Interventions (TRoPHI) Web of Science JSTOR Medicines for Malaria Venture National Institute of Malaria Research Roll Back Malaria South African Malaria Initiative United States Agency for International Development World Health Organization

Data abstraction

The resulting 43 peer-reviewed studies were included for data abstraction. We collected details of the study location, population, and years of intervention along with the reported malaria epidemiology for context. Additionally, we abstracted data on the intervention purpose and description, the care and repair practices targeted, the results from intervention and comparison groups, the types and descriptions of education activities, and the details of health behavior theories and/or instructional models described. Additionally, 289 records from the grey literature search were summarized.

Quality Assessment

Due to the variety of study designs including quantitative and qualitative designs, a single existing quality screening tool was not appropriate to assess the quality of the included studies in relation to the study question. We developed a 6-item quality criteria tool (Table 10) and scored

58

the studies that had education interventions as a main component. Studies received either 1

(contains the criteria item) or 0 (does not contain the criteria item) points for each quality criteria, resulting in a potential score range between 0-6. Articles containing 1-2 criteria received

1 star, those containing 3-4 received 2 stars, and those containing 5-6 received 3 stars.

Table 10 . Education intervention quality criteria tool Quality Criteria 1 point 0 points 1 Was there a well-defined comparison group? Yes (1) No (0)

2 Was the comparison group selected appropriately? Yes (1) No or answer to question 1=0 (0) 3 Was there an explicit link to appropriate health behavior theory? (An Yes (1) No (0) appropriate theory is one that targets the appropriate level of audience of individual, community, children, or adults) 4 Was there an adequate description of educational activities? (An Yes (1) No (0) adequate description provides replicable details of the educational activities)

5 Were the educational activities equitable for diverse learners? Yes (1) No (0) (Equitable activities include the implementation of multiple types of activities such as audio, visual, interpersonal, etc) 6 Was there hypothesis testing of the differences between groups Yes (1) No (0) pertaining to net care knowledge and/or practices?

2. Results

We identified 6,507 peer-reviewed records from the database search (Table 2) and removed duplicates in EndNote X7 (Clarivate Analytics, Philadelphia, PA), leaving 5,927 records for title and abstract screening (Figure 6). We excluded 5,481 records through title and abstract review where 1,292 records were duplicates not caught during the initial deduplication process and 4,189 were deemed not relevant in initial screening. This resulted in 446 full text articles that were reviewed, and from this process we excluded 391 articles using the inclusion and exclusion criteria. Most articles were excluded because they were not malaria education interventions (n=206) or because they were education interventions, but were not related to bed net care and repair practices in any way (n=87) (Figure 6). Others were found to be conference

59

abstracts (n=7), commentary pieces (n=46), reviews (n=29), or were excluded or other reasons

(n=16). Of the 43 remaining articles, we abstracted data from 29, excluding 14 in the process because they lacked care or repair results (n=6) or because the education was too general, meaning we found no evidence that education on net care and repair practices was provided

(n=8).

After searching multiple malaria organizations (Table 9) we identified 387 records pertaining to malaria education (Figure 6) in the grey literature search. We screened all of these records for eligibility to be summarized, excluding 98 in the process. We excluded records if they did not mention anything about malaria health education or bed net care and repair (n=85) or if they were toolkits or guides, hence not an activity that included education on care and repair

(n=13). This resulted in 289 grey literature records to be summarized (Figure 6).

60

Figure 6. PRISMA scheme of record selection and exclusion criteria

61

Peer Reviewed Literature

Malaria education as a main focus

An education intervention was the main activity of 16 (55.2%) of the 29 included peer- reviewed articles (Table 13), and a relatively minor activity of the remaining 13 articles (44.8%)

(Table 14). Of the 16 articles with education intervention as a main activity, 12 were studies conducted in sub-Saharan Africa and the remaining 4 conducted in Latin America, the Middle

East, and Southeast Asia (Table 13). Regarding the main outcomes measured in Table 13, 6 studies focused on increasing ITN or LLIN use [55, 56, 57, 58, 59, 60], 6 focused on generally improving malaria knowledge, attitudes, and/or practices [18, 57, 59, 60, 61, 62], and 5 focused more specifically on outcomes related to net durability or care and repair [6, 63, 64, 65, 66].

Concerning the care and repair practices addressed in the articles, 9 provided education of ITN treatment or retreatment with insecticide [18, 56, 57, 60, 61, 62, 63, 67, 68], 4 educated about net storage during the day [6, 55, 65, 66], 2 about appropriate soap types to use in washing nets [6,

66], 1 about recommended drying practices [6], and 1 about the manner of washing nets gently

[66].

Key findings of education on net care and repair knowledge and/or practices are described for each article in Table 13. We assessed whether studies were informed by or utilized health behavior and/or instructional theories or models (Table 11). Of the 16 articles that had health education as a main activity, 10 were informed by, applied, or built health behavior theories [6, 18, 55, 56, 61, 64, 65, 66, 67, 68] while 6 mentioned no grounding in theory (Table

13) [57, 58, 59, 60, 62, 63]. Of those mentioning theory, 6 were informed by theory, 2 applied theory, and 2 built or built upon theory. Based on the key bed net care and repair findings, we summarized whether there was evidence that the education interventions led to improvements in

62

care and repair knowledge, practices, and/or net durability. There were 8 studies that reported overall improvements [18, 55, 58, 60, 62, 66, 67, 68], 4 that had mixed results [6, 57, 61, 64], and 4 that were unclear in whether improvements occurred [56, 59, 63, 65]. There is no apparent pattern in care and repair conclusions in regard to whether theory was mentioned or how theory was used (Table 11). Of those articles not grounded in theory (n=6), 50% reported positive care and repair outcomes, while 5 (50%) of the articles grounded in theory (n=10) reported positive care and repair outcomes.

Table 11. Care and repair conclusions in relation to theory use of education interventions Care and repair conclusions How theory was Positive Negative Mixed Unclear used No theory [58], [60], [62] [57] [59], [63] Informed by theory [18], [67], [68] [64], [6] [56] Applied theory [61] [65] Built or built upon [55], [66] theory

Table 12 describes the care and repair conclusions in relation to the types of education strategies implemented. Education type includes mass or interpersonal education, the latter which involves small-scale instruction or communication (household or individual level). Mass education may be passive or active. Passive mass education includes interventions where no personal contact is made between those implementing and the recipients of the education in contrast to active mass education where there is personal contact between education implementers and recipients. There were 6 interventions that used passive mass education, which included activities such as posters, slogans, print and audiovisual media, pamphlets, t-shirts, radio programs, newspapers, maintenance labels, and social marketing [6, 18, 56, 60, 64, 66].

Active mass education activities were conducted by 12 interventions and included dramas, music, poetry, art, demonstrations, community events and workshops, lectures, small group flip-

63

chart presentations, and advocacy by community health workers [6, 18, 55, 56, 60, 61, 62, 63,

64, 66, 67, 68]. Interpersonal education involves activities such as education through house-to- house visits, establishing village bed net committees that talk with community members, dialogues and conversations, debates, training local health leaders, and individual flip-chart education. There were 8 interventions that used interpersonal methods [6, 18, 55, 57, 58, 62, 65,

66]. Discerning between mass and interpersonal education methods, there were 6 interventions that used mass strategies only, and 3 interventions that used interpersonal strategies only (Table

12). Care and repair conclusions of those 6 that used only mass education were 2 positive, 1 negative, 2 unclear, and 2 mixed results (Table 12). There was a positive, unclear, and mixed result each from those 3 articles using only interpersonal strategies. Stronger results were observed from interventions that combined at least one mass education strategy with interpersonal methods, where 4 interventions had positive care and repair results and 1 had mixed results (Table 12).

Table 12. Care and repair conclusions in relation to education methods [citation]  Care and repair Mass education Interpersonal conclusions education

Passive mass Active mass education education

[55]  Positive X X

[63]  Unclear X

[56]  Unclear X X

[57]  Mixed X

[61]  Negative X

[64]  Mixed X X

[62]  Positive X X

[67, 68]  Positive X

64

[18]  Positive X X X

[58]  Positive X

[59]  Unclear

[60]  Positive X X

[65]  Unclear X

[6]  Mixed X X X

[66]  Positive X X X Star quality rating

Figure 7 depicts the care and repair outcomes through the interaction between education type (mass vs. interpersonal vs. both) and theory use. The only clear pattern in outcomes exists among interventions that used both mass and interpersonal education methods, regardless of theory use, where 4 articles reported positive results and 1 reported mixed results. The article

[62] that used both education methods but did not use theory reported positive net care and repair results. A variety of results was found among articles that used a single method of education and did or did not use theory (Figure 7).

65

Figure 7. Care and repair conclusions in relation to education methods and theory use

66

Quality Assessment

Education intervention quality was assessed for the 16 articles that had health education as a main activity (Table 13) using the rubric detailed above (Table 10, Appendix F). We classified the education intervention component of 3 articles as single star [57, 59, 63], of which

2 had unclear results and one had mixed results of bed net care and repair measurements. Four articles had two stars for their education intervention components [56, 65, 67, 68]. Two of these articles had unclear care and repair conclusions, while improved knowledge was measured in the other two. However, the two articles reporting improved knowledge were describing the same study [67, 68]. There were 9 articles that received 3 stars, 6 of which reported improvements in care and repair knowledge and/or practices [18, 55, 58, 60, 62, 66], and 3 of which reported mixed results of care and repair knowledge or practices [6, 61, 64]. None of the one star quality- rated education interventions, all of the two star interventions, and 6 of the 3 star interventions mentioned theoretical framework.

Malaria education as a minor element

Of the 13 articles which contained an element of malaria education, but in which education was not a main focus, 9 were studies conducted in sub-Saharan Africa, and the remaining 4 in South and Southeast Asia and South America (Table 14). The main outcomes studied ranged from malaria indices [69, 70, 71], aspects of bed net care and repair [15, 72, 73,

74, 75], program evaluation [71, 76, 77], and bed net coverage and/or use [72, 74, 75, 78, 79].

The bed net care and repair practices addressed by the articles in table 4 include washing frequency [15, 69, 70, 74, 75, 78, 79, 80], ITN treatment and/or retreatment with insecticide [70,

72, 73, 74, 76, 77, 80], net storage during the day [70], bed net repair [15, 70, 72, 80],

67

appropriate soap type [75], proper drying practices [75, 79], and general care messages [71, 73].

Only one of the 13 articles in table 4 mentioned a theoretical framework. Snow et al. (1999) used social marketing strategies to increase net ITN retreatment and found that house to house retreatment schemes can be highly effective in increasing retreatment, but that the delivery mechanism is more important than social marketing [73]. Overall, 5 articles reported improvements in bed net care and repair knowledge and/or practices [70, 75, 76, 77, 80], 3 articles had mixed results [72, 73, 79], 4 had unclear results [69, 71, 74, 78], and 1 reported negative results following education efforts [15] (Table 14).

The number of bed net care and repair messages varied between the 29 peer-reviewed studies. More than half of interventions provided education on only one care and repair recommendation (15 studies), 4 articles on 2 recommendations, 6 articles on three recommendations, and 3 studies on 4 or more recommendations (Tables 13 & 14).

68

Table 13. Characteristics, key findings, and quality scores of peer-reviewed articles with a malaria education main focus Author, Country, Main Care and repair Key care and repair findings Theoretical Care and repair Year  Intervention outcomes practices framework, conclusions [citation] years, Sample addressed level of use size

Marsh, Kenya, 1993- ITN use Storage, washing Knowledge of bed net messages among children Community Improved 1996 [55] 1994, 200 frequency, repair increased from 58.2% pre-education to 80.5% 3 Research & knowledge  mothers & 50 months post-education. Post-intervention, between Health Education household heads 33-50% of parents reported hearing messages from Framework, children, and specific bed net messages could be Built upon recalled by 30% of mothers.

Mfaume, Tanzania, 1993, ITN Retreatment Post intervention 52-98% of households regularly None Unclear 1997 [63 ] 44 mosque retreatment retreated their nets  attendees

Minja, Tanzania, 1996- ITN Treatment/ Pre-IEC campaign 3% of survey respondents had Social Unclear 2001 [56] 1997, 26 focus ownership retreatment treated nets and 34% had knowledge of retreatment. Marketing,  groups, 39 in- and use Post-IEC campaign, ITN ownership and use Informed by depth interviews, increased to >50% 204 survey respondents

Rhee, Mali, 2003, 133 ITN use, Treatment/ Education intervention had no effect on household None Mixed. Improved 2005 [57] households malaria retreatment net treatment/retreatment (OR 1.6, 95% CI 0.8,3.3) practices at  knowledge individual, but not household level

Alvarado, Colombia, 1995- Malaria Treatment/ Intervention led to no change in net retreatment (OR Precede-Proceed Mixed. Did not 2006 [61] 1997, 450 knowledge, retreatment 0.88, 96% CI 0.57,1.86). Model, Applied improve  respondents from attitudes, practices, 20 villages and Decrease in malaria incidence (OR 0.58, 95% CI reduced malaria practices, 0.39,0.87) incidence malaria incidence

Panter- The Gambia, Bed net Repair Mean % holes repaired increased from 27-41% Social Ecology Mixed. Improved

69

Brick, 2003, 92 repair (p<0.001) Model, Informed practices, did not 2006 [64] households by improve net  No improvement in net quality quality

Nonaka, Laos, 2004-2005, Malaria Treatment/ Difference in mean Malaria Attitudes score of 1.6 None Improved 2008 [62] 130 knowledge, retreatment was 1 points (p<0.001) between intervention and control attitudes  schoolchildren, attitudes, of 7 points in the groups 103 guardians of practices Malaria Attitudes schoolchildren, score & 130 married women without children

Widmar, Tanzania, 2007- LLIN Retreatment, Knowledge that LLINs do not require retreatment Theory of Improved 2009 & 2008, 344 coverage repair increased from 23.5-39.4% (p<0.001) and Planned knowledge Ho, 2012 households knowledge to repair nets increased from 37.1-90.4% Behavior, [67,68] from pre-intervention to 15 month follow up Informed by 

Ayi, 2010 Ghana, 2007- Malaria Retreatment Net retreatment increased from 21.5-50.0% Health Improved [18]  2008, 186 school knowledge, (p<0.001) in the intervention group and from 25.3- Promoting practices children & parasite 30.5 (p>0.05) in the control group from pre- School Concept, 383 adults prevalence education to follow up Participatory Learning and Action Instructional Model, Informed by

Rickard, Ghana, 2008- ITN use Washing, repair, Knowledge of net repair increased from 54.7-81.3% None Improved 2011 [58] 2009, 61 general (p<0.001) from baseline to 1 year post intervention knowledge  households maintenance

Ahmadi, Iran, 2009, 2563 LLIN use, Washing 76.3% of untreated net households washed every 6 None Unclear 2012 [59] households malaria frequency months compared to 68.9% of LLIN households.  perceptions 23.7% of untreated net households washed every and practices year compared to 31.1% of LLIN households

Amoran, Nigeria, ~2010, ITN Retreatment From pre-intervention to follow up, increase in None Improved

70

2012 400 nursing knowledge knowledge of when to retreat ITNs from 39-76% knowledge [60]  mothers and use (p<0.001) among intervention group compared to 17.0-17.8% (p>0.05) among the control group

Harvey, Peru, 2000-2001, Treatment, Storage during In phase 1, 33% of households stored their nets Trials of Unclear 2013 [65] 15 households net care day properly during the day at least 80% of the time Improved  compared to 47% of households in phase 2 Practices, Applied

Helinski, Uganda, 2013- LLIN Storage during No difference between intervention and comparison Net Care and Mixed. Improved 2015 [6] 2014, 870 durability day, washing group in the increase in net repairs from baseline to Repair knowledge, did  households frequency, soap follow up (5.6-56.4% intervention group, 3.4-48% Conceptual not improve net type, drying, comparison group) Framework, durability repair Informed by Knowledge at follow up: Handle nets with care 39.9% int, 28.3% comp (p=0.059) Keep away from fire 16.6% int, 17.6% comp (p=0.800) Keep away from rats 21.3% int, 11.4% comp (p=0.007) Keep away from children 29.3% int, 15.5% comp (p=0.003) Roll/tie up when not in use 57.0% int, 51.1% comp (p=0.215) Wash less often 16.7% int, 31.9% comp (p=0.001) Wash with ordinary soap 62.0% int, 73.9% comp (p=0.014) Dry in shade 61.2% int, 52.7% comp (p=0.080) Repair holes 58.5% int, 10.3% comp (p<0.001)

Koenker, Nigeria, 2012- LLIN care Storage during No difference between intervention and control P-Process, Built Improved 2015 [66] 2014, 2170 and repair day, washing group in the increase of observed repairs at follow knowledge,  households behaviors, frequency, soap up (10.5-26.5% intervention, 10.3-17.8% controls improved net condition type, washing p>0.05) practices manner Nets tied up during the day were more likely to be in serviceable condition at follow up (OR 2.70 95% CI 1.50,4.86)

71

50.1% of nets survived to follow up among households that recalled care and repair messages compared to 30.9% of nets among households that could not recall messages

Recall of messages at follow up Handle carefully 61% int, 35% cont (p<0.05) Avoid fire 22% int, 12% cont (p<0.05) Repair early 26% int, 12% cont (p<0.05) Tie/fold up when not in use 36% int, 30% cont (p>0.05) Don’t soil with food 22% int, 11% cont (p<0.05) Wash only when dirty 21% int, 11% cont (p<0.05) Use mild soap 46% int, 27% cont (p<0.05) Wash gently 41% int, 29% cont (p>0.05)

Reported practices at follow up Keep away from children 49% int, 31% cont (p<0.05) Keep away from pests 23% int, 12% cont (p<0.05) Roll/tie up when not in use 43% int, 30% cont (p>0.05) Handle with care 38% int, 25% cont (p<0.05) Don’t soil with food 27% int, 13% cont (p<0.05) Keep away from fire 15% int, 10% cont (p>0.05) Wash gently 46% int, 30% cont (p<0.05) Wash only when dirty 42% int, 36% cont (p<0.05) Inspect regularly for holes 11% int, 5% cont (p>0.05) Repair quickly 12% int, 4% cont (p<0.05)

72

Table 14. Characteristics and key findings of peer-reviewed articles with a minor malaria education focus Author, Country, Main Care and repair Key care and repair findings Theoretical Care and repair Year Intervention outcomes practices framework conclusions [citation] years, Sample addressed

Alonso, The Gambia, Malaria Washing On average, nets in intervention group washed None Unclear 1993 [69] 1989, 12 villages morbidity frequency every 2 months vs. every month in control group and mortality

Brieger, Nigeria, 1992- Parasite Retreatment, Knowledge scores of bed net use and care increased None Improved 1996 [70] 1993, 1347 prevalence, washing from 3.8-5.2 (possible max score of 9) from the first knowledge households entomologic frequency, storage monitoring visit to the last. al indicators during day, repair

Cham, The Gambia, Program Treatment 5 months following campaign, 83.3% of nets had None Improved 1996 [76] 1992, 360 evaluation been treated compared to 0 at baseline practices compounds

Kachur, Kenya, 1997, 60 ITN Retreatment, At follow up all hanging nets had been repaired, and None Mixed 1999 [72] households and maintenance repair 87% were in good or fair condition. 16 focus groups & use No nets had been retreated

Snow, Kenya, 1993- Retreatment Retreatment, House to house retreatment attained 98% Social marketing Mixed. Delivery 1999 [73] 1997, 5700 general care retreatment coverage compared to 64% in the cost method more households free sentinel dipping method and 7% in cost important than retrievel sentinel dipping method social marketing

Alaii, Kenya, 1996 & Net and Retreatment, Net care knowledge in intervention group vs. None Improved 2003 [80] 1999, 600 malaria washing control group: knowledge households prevention frequency, repair Retreat every 6 months (98% vs. 75.8%, p=0.001) perceptions Wash every 6 months prior to retreatment (48.1% vs. 26.3%, p=0.001) Repair if torn (36.7% vs. 24.2%, p=0.001)

Shirayam Laos, 2003, 240 LLIN Washing History of participation in malaria education None Did not improve a, 2007 households condition & frequency, programs not associated with any maintenance practices [15] maintenance regeneration behaviors

73

behavior process, repair Of households surveyed, 38.2% followed washing frequency recommendations, 83% followed regeneration process, and 32.7% of hous eholds with torn nets had repair

Wacira, Kenya, 2002, ITN Treatment/retreat Practices in employer-based retreatment services None Unclear 2007 [74] 2095 households ownership, ment, washing group vs. community-based services group: use, frequency Nets washed > 3 times in 6 month period (40% vs. retreatment 50%) Nets washed whenever appear dirty (25% vs. 25%) Treat/retreat rate (58.1% vs 46.9%, p<0.001) Treat/retreat in last 6 months (25.1% vs. 13.5%, p<0.05)

Fernando Sri Lanka, 2007- LLIN Washing Significant correlation between knowledge None Improved , 2008 2008, 2467 perceptions, frequency, soap regarding LLINs and maintenance practices. practices [75] households practices, type, drying and use Differences between cultural groups (Sinhalese, Tamil, Muslim) Knowledge that LLINs do not require retreatment (16.7% vs. 17.5% vs. 9.5%) Mean net care practices scores (11.77 ± 0.59, 7.52 ± 10.57, 8.18 +0.63) (p<0.05)

Gerstl, Sierra Leone, LLIN Washing Washing frequency practices among different None Unclear 2010 [78] 2008, 900 coverage frequency aged nets (Nets acquired 0-10 months, 1-1.5 households, 4997 and use years, >2 years prior to survey): individuals Never washed (31.8% vs. 5.6% vs. 3.9%) Washed 1-4 times (63.1% vs. 77.8% vs. 39.4%) Washed 5-20 times (3.8% vs. 14.4% vs. 48.7%) Washed >20 times (0.6% vs. 0.6% vs. 3.2%)

Ahmed, Bangladesh, ITN Washing Net care knowledge in 2011 vs 2008: None Mixed 2011 [79] 2008 & 2011, coverage frequency, Correct washing frequency (28.3% vs. 30.6% of SE 11655 and use washing manner, households and 35.6% vs. 12.0% of NE households) households drying Don’t wash in pond or canal water (32.8% vs. 31.6% of SE households and 58% vs. 11.8% of NE

74

households)

Correct drying knowledge (39.5% vs. 20.4% of SE households and 58.3% vs. 7.7% of NE households)

Hiwat, Suriname, 2005- Program Treatment/retreat 327 nets treated/retreated (54.5% of target goal) None Improved 2012 [77] 2009, Medical evaluation ment practices, but did Mission Malaria 58 communities reached by malaria media not reach targets Program awareness campaign (41.4% of target goal) Evaluation

610 people reached through BCC (19.2% of target goal)

Khadka, Angola & Program General The program reduced odds of fever by 54% over the None Unclear 2018 [71] Namibia, 2014- evaluation, maintenance 2 year intervention period 2016, 2126 malaria households morbidity

75

Grey Literature

Of the 289 grey literature records included for summarization (Figure 6), 286 (99%) were from the Against Malaria Foundation [81]. These records encompass 136 net distributions completed by various organizations between 2006 and 2017, across 26 countries in sub-

Saharan Africa, 5 countries in South and Southeast Asia, and 3 countries in Latin America. Each distribution included a pre-distribution report and at least one distribution report detailing the plans and completed actions of the distributions including education components. We summarized the specific bed net care and repair education messages reported through these distribution reports (Figure 3). More than half (80, 58.8%) of the distributions did not mention any aspects of bed net care and repair, while 38 (27.9%) of the distributions included messaging around general net care defined as stating only “maintenance” or “general care instructions”.

There were 26 (19.1%) of distributions addressed specific care and repair practices including washing (15.4%), drying (6.6%), avoiding damage (2.2%), appropriate storage during the day

(2.2%), repair (1.5%), and appropriate soap types (0.7%) (Figure 8). These do not add to 100% because some reports included both general care and at least one specific care practice.

76

Figure 8. Bed net care and repair messages reported among Against Malaria Foundation-funded bed net distributions

Bed net care and repair messages reported among Against Malaria Foundation funded bed net distributions

Soap type 0.7

Repair 1.5

Storage during day 2.2

Avoiding damage 2.2

Retreating 5.9

Drying 6.6

Washing 15.4

General care 27.9

None 58.8

0 5 10 15 20 25 30 35 40 45 50 55 60 65 Percent of distributions

The remaining 3 grey literature records included were from Imagine No Malaria (1 record) and Malaria Consortium (2 records). Imagine No Malaria did not detail a particular education intervention, but rather reported that the organization supports bed net distributions by training community health workers to educate households on proper net use and care of nets during bed net delivery and installation activities [82].

Malaria Consortium included 2 particular projects that contained components of bed net care and repair education. The Pioneer Project conducted from 2009-2014 in Uganda promoted the message to “Dry your net under the shade after washing it” though net care was not a main focus of Behavior Change Communication ( BCC) activities [83]. Interviews and focus groups discussions from The Malaria Prevention and Control Project in Mozambique from 2011-2017 discussed exposure to net care and repair messaging during a LLIN distribution 6 months

77

previously [84]. Participants had high awareness that net care and repair is important and good recall of messages to repair torn or damaged nets, but the accurate recall of other care recommendations was inconsistent [84]. Specifically, respondents did not know appropriate net washing locations (in the river vs. a washtub or bucket) or whether nets should be dried in the shade or sun [84]. None of the grey literature records mentioned theory and were contained sparse instructional details.

78

CHAPTER 3

DISSERTATION CONCLUSIONS

A. Specific Aim 1: Bed net care adherence

Net care adherence scores were generally low (adherent) in both the highlands and lowlands of western Kenya. Nets had median adherence scores of 2 and 4, respectively, with 0 being the optimal score out of a possible 14. However, there are a few key care behaviors that could be targeted for improvement in the lowlands. Net storage, washing frequency, and drying are the key areas for interventions to address.

Tying up nets or storing them in a safe space while not in use during the day prevents physical damage. This practice may also increase the amount of time a net appears clean, which may lead to less frequent washing. Focus group participants in Senegal reported that tying up nets during the day is a best practice for preventing damage, and removing the net and storing it in a safe space is also a good practice particularly when children are likely to pull on hanging nets [39]. A ‘trials of improved practices’ study in the Peruvian Amazon region found that some households that did not remove their nets during the day developed ways of tying the nets and covering them with plastic so they could remain hanging during the day and still be protected from damage [65]. In our study, variables associated with leaving a net hanging rather than tied up during the day were varied, and span attitudes and environmental factors in both the highlands and lowlands. This suggests that there is no one factor that influences net storage behavior, but a combination of various factors do. Harvey and others reported 15 different reasons a household left a net hanging, the most common of which included lack of time, someone rests under the net during the day, feeling lazy, or forgetting to remove the net [65], indicating that it is inconvenient and time consuming to remove nets during the day. Tying up nets during the day is

79

less labor and time intensive, and prevents damage compared to nets left hanging over sleeping spaces during the day.

Frequent washing can degrade both the physical and chemical integrity of LLINs, so current recommendations are to wash LLINs at most every 3 months. Many nets are washed more frequently than the current recommendations. In the lowlands nearly 80% of nets were washed < 3 months prior to the survey. Other studies have found variable washing frequencies spanning from 1.5 washes per year on average in Uganda [85] to up to 8 washes per month in

Mali [86]. Additionally, 31% of nets were washed more frequently than the recommended practice in a Kenyan study conducted among the southern coastal districts [87]. These differing wash frequencies across studies make generalizability difficult, however the reasons for doing so seem to be less related to knowledge of preferred practices and more related to social and structural factors. Nets simply get dirty quickly, and households (due to personal or social acceptability reasons) prefer to wash nets when they are dirty [15, 41, 42, 65, 86, 88]. Similarly in our study, washing frequency was associated with environmental factors, and not with knowledge and attitudes like net care knowledge, perception of malaria risk, or confidence in ability to prevent malaria. Although there is great variation across studies, it is important to monitor the practice and understand why households perform variable washing practices due to the significant implications for net longevity.

Soap type and net brand may be modifying factors with washing frequency depending on the pH level of the soap. Bar soaps and mild detergents cause less damage to insecticide levels on nets compared to bleach. In this study, no nets were reported to be washed with bleach, though we were unable to determine brands or pH levels of soaps and detergents used. A washing trial measuring insecticide residual after set time intervals of 6 or 12 months following

80

net distribution in Uganda found that nearly all (99.6%) of nets in the region were washed with mild soap (pH 9-10) and the remaining nets washed with detergent [89] while we found a greater proportion of nets were washed with detergent (21.2% of nets in the highlands and 53.5% of nets in the lowlands). With respect to adherent soap type practices, the Ugandan study found that insecticide integrity differed by net brand and net generation within net brands [89]. Washing nets with more stringent detergents rather than mild soap may impact insecticide integrity. A study in western Kenya tested insecticide effectiveness by measuring how many mosquitos were able to feed through nets and how many mosquitos died after net exposure after 5, 10, and 15 washes with a detergent commonly used in rural areas [90]. Even after only 5 washes with detergent, up to 17.6% of mosquitos were able to feed through nets and up to 83.3% of mosquitos remained alive 24 hours after net exposure depending on net brand [90]. Though this study did not compare mosquito ability to feed or mosquito mortality after exposure to nets washed without detergent, this suggests that while net brand also plays a role, insecticide integrity is noticeably compromised even after 5 washes with detergent [90]. For consistent messaging, recommendations should continue to emphasize the use of no or very mild soaps for net washing, regardless of net brand.

Drying nets in the sun is not recommended because direct sunlight degrades insecticides

[91]. As many nets are dried in the sun in the lowlands, proper net drying is a key net care component to be addressed. Another study in Kenya found that 64.9% of nets were dried in the sun [92], and nets were observed drying in direct sunlight in Uganda as well [93]. Our results suggest that knowledge and attitudes, particularly a lack of knowledge of appropriate net drying influence this particular care practice. Learning net care from a community event was associated

81

with better net drying adherence, suggesting education campaigns geared toward community settings may be effective in improving net drying practices in the lowlands sites.

Implications for interventions

Net washing frequency, use of soap, and drying location affect net longevity [42, 85, 93,

94, 95, 96, 97]. Other studies have found that LLINs do not last as long as intended in the field, often in poor condition after 1-2 years of use [98, 99]. While there is variation among studies in net longevity following distribution campaigns [100], many nets are lost to attrition or are damaged before the universal 3 year life span [91].

Because LLINs involve multiple care and repair practices, and each practice is influenced by multidimensional factors spanning attitudes and environmental factors, intervention strategies should be variable and informed by context- and location-specific care and repair practices [101].

Because of the varied and at times non-existent associations between knowledge of net care items and adherence to care practices, education interventions alone are unlikely to markedly improve bed net care and repair. Social factors are likely the most difficult to change as they are related to cultural and personal belief systems. Thus, successful solutions aimed at improving care and repair should particularly involve environmental interventions. Some possibilities to explore include providing materials to assist with tying nets up during the day if household structures do not easily permit net folding or tying. A recommendation by Harvey and others is to design nets to be easily pulled up or so they can remain hanging, yet be more protected from damage [65]. Because we also found that many nets are left hanging during the day, our results support that this suggestion could prevent physical degradation without necessitating behavior change. Furthermore, consistent and complete care and repair messaging provided from any

82

place where a net is distributed (particularly clinics and distribution events) is recommended.

Demonstrations of proper care and repair activities may be particularly effective.

B. Specific Aim 2: Bed net repurposing and misuse

Observation of repurposed bed nets

We observed that 8.1% and 33.0% of households had repurposed nets in the highlands and lowlands, respectively. Repurposing bed nets is reportedly a common occurrence throughout

Kenya [102]. Bed nets were repurposed for multiple uses both indoors and outdoors, particularly for fencing, chicken coops, to protect crops, to cover bathrooms or serve as curtains, or cut up as cleaning rags. Of these repurposed nets, 89.6% and 91.0% of nets in the highlands and lowlands, respectively, are classified as neutral as their new purpose does not protect household members from mosquitoes [43]. Many of these neutral nets were repurposed to help store, dry, or otherwise secure food. Generally, once nets are deemed too old or worn to be used for sleeping, they are repurposed for food security purposes. Few nets were beneficially repurposed (still offered protection against mosquitoes). Households that use bed nets for sleeping while also beneficially repurposing old nets as screens or curtains may have better mosquito protection than households that dispose of old nets or repurpose them for uses that do not protect against mosquitos. Kibe et al. suggest the development of net repurposing and disposal guidelines to promote beneficial repurposing (curtains, window screens, etc) [102]. Many repurposed nets were used for similar purposes to those reported in other studies. Participants in focus groups in

Timor-Leste mentioned repurposing bed nets for protecting vegetables/crops and for use as sieves [103], and others have reported that nets are used for drying fish [104] and various other purposes including as curtains, ceiling covers, and decorations [102, 105, 106].

83

Reported reasons for bed net repurposing

The reasons reported for repurposing are nearly ubiquitous in that households deemed the nets too old and/or too damaged to be used for sleeping under anymore. It was apparent from focus groups and interviews in Ethiopia that bed nets are repurposed when they are no longer useful for sleeping due to wear and tear [107]. Focus group participants in Kenya explained that nets are repurposed because they are strong and durable for other uses, and because they do not know a better way of disposing of the nets once they are no longer useful for sleeping under

[102]. While repurposing almost always occurs when nets are no longer usable, there is variation among households’ definition of usability [107, 108]. Some households may consider nets with only a few holes to be too damaged for use [106, 109], while other households repurposed nets that were older than 2 years [42]. In a longitudinal study following LLINs post-distribution in

Zambia, the mean net age of discarded or repurposed nets was 18 months [110]. Though we could not determine the temporality of whether holes appeared in repurposed nets before or after they were repurposed, nearly all repurposed nets in both the highlands and lowlands had many holes. Efforts to promote net repair may help nets continue to be used for mosquito protection longer before they are discarded or repurposed.

Reports of net repurposing in the community

Another level of repurposing examined includes the reported observations by study participants of net repurposing frequency among their community members. Viewing repurposing as socially acceptable, or the normalization of repurposing, may influence whether households repurpose nets, as we found households were significantly more likely to have repurposed nets if they reported observing other households in the community repurposing nets.

Net repurposing is also often concentrated in communities [105], thus as the practice becomes

84

normalized in a community, more households are likely to repurpose their bed nets. In Senegal, study participants reported seeing bed nets used for alternative purposes in some communities, but not others [106], emphasizing the clustered nature of net repurposing. Social acceptability is associated with other aspects of bed net use, particularly the desire to keep nets from appearing dirty [39, 41], so the more socially acceptable a practice is, the more prevalent it becomes.

Rather than throwing old, worn nets away, households observe others finding innovative uses for them, see that the practice is acceptable, and find ways to make their nets useful for their household.

Bed net misuse

The third level of repurposing we assessed was bed net misuse. Using an intact, functional net for a purpose other than mosquito protection while sleeping was very rare. In our study sites, only 1 repurposed net in the highlands and 3 nets in the lowlands had no holes, though it is possible these nets were still considered old and worn by the household.

Additionally, among intact nets inside the home, fewer than 1% of functioning bed nets have ever been alternatively used in both sites. Others also report rare misuse [9] and suggest misuse occurs when alternatively using a bed net is economically beneficial [46, 111], particularly where nets are misused for fishing [11, 103, 104]. We observed no repurposed nets used for fishing, though nearly 50% of households in the lowlands reported seeing others using nets for fishing. There are no large bodies of water near the study sites, and fishing is not an economic activity in the area. Most areas where bed nets are misused for fishing are heavy fishing communities [11, 42, 104, 112, 113] near large bodies of water including coastal areas, the

African Great Lakes, and large rivers [114]. Our study sites, like others where non-fishing activities drive the economy, report that misuse is rare [9, 46, 105].

85

Due to the environmental harm caused by fishing with ITNs or LLINs [115, 116, 117,

118], misusing nets for fishing is a serious problem that should be addressed [43]. While misusing nets for fishing is an environmental problem, it is not clear if misuse in general is a problem for malaria prevention efforts including net ownership, use, and malaria outcomes. For example, it is not always clear whether households that use nets for fishing also use nets to sleep under [104].

Impact of net repurposing on malaria prevention efforts

Though common, bed net repurposing only poses a problem to malaria prevention efforts if households that repurpose nets do not meet ownership recommendations or if household members do not use nets. We found no association between household net repurposing and net ownership, use, or malaria outcomes. While there were more households with repurposed nets in the lowlands than in the highlands, more household members also slept under nets in the lowlands, supporting reports that repurposing is unlikely detrimental for household net use for mosquito prevention [9]. Few households had repurposed nets and household members who did not sleep under a net in both sites. Though sleeping under a torn, old net may still offer some protection against mosquitos, most households with repurposed nets also used nets for sleeping.

Even in a study where bed net misuse for fishing was very common, more than 80% of individuals using nets for fishing reported they also used bed nets in the household [104]. As more net distributions occur, households are likely to attain newer nets to replace worn nets that are then repurposed in economically beneficial ways to the household [102]. However, in contrast to the lowlands, we observed a non-significant, though interesting trend suggesting households with fewer individuals sleeping under nets are more likely to have repurposed nets in the highlands than households where nearly everyone slept under a net. Even if households in the

86

lowlands do not own enough nets, several household members crowd under a single net, while this occurs infrequently in the highlands [119]. This suggests there may be differences in net use and repurposing behaviors between areas of year-round and seasonal Plasmodium transmission.

C. Specific Aim 3: Systematic review of malaria education interventions

Education interventions resulted in improvements of bed net care and repair knowledge and/or practices in less than half of all included peer-reviewed articles (n=13, 44.8%) and in 8

(50%) of the articles with an education intervention as a main study component. Interventions that received 3 stars in the quality assessment were more likely to include an appropriately selected comparison group, be grounded in theory, and use equitable instructional activities

(Appendix F). Evaluation of education intervention effectiveness is discussed assessing theoretical grounding, education strategies and methods, and the implications on net durability and longevity.

Theoretical grounding

It is thought that theory based health education interventions are more effective than non- theory based interventions [120] because they outline the rationale and process for how the intervention will result in change [121]. However, we found no differences in success between interventions grounded in theory and those that did not mention theory. This was further confirmed by the interaction between education method type and theory use (Figure 7), which showed that theory does not appear to modify care and repair outcomes in relationship to the method of education. Rather, it seems that using a combination of mass education and interpersonal education methods shows more evidence for intervention effectiveness than grounding in theory, though this requires further exploration for confirmation.

87

Interventions that were informed by or applied theory achieved varied net care and repair results. Though there were only 2 interventions that built or built upon theory, both resulted in positive care and repair outcomes. Building or building upon theory may be the most successful way to implement education interventions because it requires extremely context-specific knowledge of the target populations, and care and repair practices are known to vary by community [98]. However, a larger evidence base is needed to determine how utilizing health behavior theory influences the success of education interventions aimed at improving bed net care and repair practices. Potentially clouding the effect of theory on intervention effectiveness is the wide variety of theories utilized. No two articles used the same theory. This is not surprising given the large number of health behavior theories available. As of the 1990s, 66 health behavior theories and models were found among 536 articles to inform education intervention designs

[122], and more theories and models are continually developed for differing health contexts such as the net care and repair conceptual framework [66] utilized by Helinski et al [6].

Education

Comparison groups

Studies that do not adequately describe the selection of appropriate comparison groups makes it difficult to synthesize the overall effectiveness of education interventions on changing knowledge or behaviors. Without knowing whether comparison groups could be contaminated with intervention messaging, or without directly measuring this contamination, it is difficult to know if exposure misclassification affects intervention outcomes. Some studies purposefully selected control or comparison groups to limit the possibility of contamination. Amoran et al. randomly selected intervention and control sites in geographically separated areas [60]. Helinksi et al. chose culturally similar, but geographically separated sites where the populations spoke

88

different languages [6], and Koenker et al. chose a control site that was culturally similar to the intervention site, but that was outside of the range of broadcasting services [66]. Furthermore, studies that use passive mass education methods survey respondents on exposure to specific intervention slogans to measure the level of exposure misclassification [6, 66]. Furthermore, interventions that can provide evidence of temporality, dose-response relationships of behavior adoption and exposure to education activities, explanations for how change occurred, and methods to minimize effects of non-randomization will help build evidence of intervention effectiveness [123].

Descriptions of instructional strategies

Interventions receiving 3 stars in the quality assessment detailed replicable descriptions of the instructional strategies that were implemented, however still lacked a framework for the actual provision of the instructional events. Health education interventions are inherently multidisciplinary, often requiring expertise in public and global health, health promotion, and anthropology, yet they appear to lack involvement of experts in education – those that understand and apply learning theories and instructional frameworks. Of the articles with malaria education as a main focus (Table 13), only 3 indicate author affiliations with individuals outside of public health, medical, and anthropology realms including one author from a humanities department [61] and authors from a center for communication programs [6, 66]. In 2005, Kinzie linked health behavior theories and instructional frameworks specifically to provide instructional design strategies for health educators [21], but this resource was not cited by any of the articles in Table 3. Though not related to net care and repair, one intervention aimed at improving malaria knowledge among travelers explicitly utilized learning theory to develop an educational game, resulting in significant knowledge improvements [124]. More emphasis should be placed

89

on the learning process for differing target audiences in the design and implementation of malaria education interventions, particularly interventions that educate through interpersonal communication.

Types of instructional strategies

Though not explicitly defined as such, the interventions receiving 3 stars employed equitable educational activities, meaning activities were accessible to diverse learners. This follows a principle of universal design for instruction which is often discussed for post- secondary education [125], but is applicable to all instructional settings. Offering various ways of delivering educational content allows information to be accessible to multiple types of learners [125]. In the context of BCC campaigns, this means using various messaging tactics including visual, auditory, and interpersonal communication strategies to send a unified message.

For example, one intervention employed only one mode of education (auditory messages during sermons) [63] while other interventions combined multiple instructional activities such as flipbooks, radio messaging, posters, and drama performances [6,18, 55, 56, 58, 60, 61, 62, 64,

65, 66]. A review of malaria social and behavior change communication campaigns concluded that interventions using multiple activities, coupling mass media messaging with community- tailored strategies increased the likelihood of obtaining behavior change [126]. Providing multiple types of educational activities is particularly important in areas with disparities in literacy rates. Written materials along with other visual and auditory methods can ensure that populations with varying literacy levels are exposed to the same messaging in a way they can best learn.

90

Overall educational methods

Interventions reached target populations using mass education, interpersonal education, or a combination of the two. It is likely using the combination of mass and interpersonal strategies may be most effective, as all 5 interventions that used a combination reported positive

(4 studies) or mixed (1 study) care and repair results [6, 18, 55, 62, 66]. There was no discernable pattern of care and repair conclusions among interventions that used only mass or interpersonal education methods, however, these conclusions are tenuous due to the small sample size of interventions.

Comparing care and repair results across interventions is further complicated because implementation of education exposure differs between passive mass, active mass, and interpersonal education methods. Passive mass education is the easiest to implement consistently, while more variation dependent on individuals providing education is expected with active mass and interpersonal education. Rhee et al implemented interpersonal education to increase the proportion of retreated ITNs, but obtained mixed results where more individuals who received education retreated ITNs, but there was no difference between intervention and control households in the number that retreated nets [57]. A potential explanation for weak results was found due to interaction in retreatment outcomes depending on the field guide who implemented the education even though all guides received the same training [57]. Education methods using individuals to provide the instruction should have some training in education theory to obtain the greatest possible effectiveness and highest possible degree of consistency between educators. For example, a school based education intervention that improved ITN retreatment practices in Ghana [18] specifically trained instructors in Participatory Learning and

91

Action (PLA), an evidence-based facilitated approach that actively engages participants in the targeted interventions [127].

Comprehensive versus single message education

There is a lack of comprehensive care and repair messaging among both education interventions and bed net distributions. Ten (63%) of the articles with malaria education as a main component addressed only a single care and repair practice, 3 (18.8%) educated on 2-3 practices, and only 3 (18.8%) interventions educated on 4 or more practices. It is important to note that very few of the included interventions were focused on improving care and repair knowledge and/or practices. Most interventions were concerned with other outcomes such as improving net use, thus care and repair education was a relatively minor priority. There was also a serious lack of specificity in care education of net distributions funded by the Against Malaria

Foundation where 86.7% of distribution reports described no care and repair messaging or general messaging defined as only stating “net care” or “net maintenance” in the education description (Figure 8). To achieve maximal improvements in net longevity and durability, comprehensive education interventions dedicated to improving care and repair practices are needed, and net distribution campaigns should ensure that care and repair messaging is consistent and comprehensive. One such intervention that focused on net care and repair knowledge and practices incorporated comprehensive messaging and resulted in improved care and repair knowledge, practices, and net longevity [66], suggesting that education interventions can lead to promising results, though more evidence is needed.

Impacts on bed net durability and longevity

While education interventions led to marginal improvements in knowledge and practices, there is little evidence that these improvements improved net durability and longevity, largely

92

because only three studies measured bed net durability as an outcome. One found that education increased care and repair knowledge and attitudes, but did not improve net durability [6]. One found that repair practices significantly increased, but did not improve net condition [64].

Another study found that nets tied up while not in use were more likely to be in serviceable condition at follow up [66]. More education interventions of bed net care and repair that measure net durability as an outcome are needed. Future interventions should be conducted through LLIN distributions before nets are used to demonstrate whether bed net care and repair education improves net durability and longevity. Furthermore, more emphasis on net care education should be placed on net distributions funded by non governmental organizations (NGOs) such as the

Against Malaria Foundation (AMF) to ensure all net recipients are receiving complete and accurate net care recommendations. When an organization wishes to implement a net distribution funded by AMF, they are required to provide plans for educational components of the distribution [81]. However, this requirement is vague and does not specify particular malaria or malaria prevention content that should be conveyed, resulting in more than half of the 136 completed distributions that did not mention any education plans or education conducted regarding net care or repair (Figure 3).

Conclusions and recommendations

It is difficult to make strong conclusions regarding the efficacy of education interventions on improving net care and repair knowledge, practices, or durability as most of the included interventions educated on care and repair as a secondary or minor objective. We did not find evidence that interventions grounded in health behavior theory achieved more positive results than those not grounded in theory. Interventions that employed both mass and interpersonal education showed promising results that warrant further investigation.

93

For increased rigor and comparability of future education interventions, we recommend the following:

1. Because education interventions on care and repair practices must be highly context-specific, a focus on training instructors in pertinent education theory may help improve education quality and consistency to yield stronger results.

2. Interventions should be planned, monitored, evaluated, and reported using guidance put forth by the RBM working group. There is a wealth of material freely available including standardized indicators, survey development guidance, and checklists to ensure complete reporting [121, 128,]

129].

D. Limitations

Specific Aim 1:

Net care and repair practices are likely quite different between communities, regions, and countries, making comparisons and generalizability between sites and other studies difficult.

This highlights the importance of LLIN monitoring after distributions in different locations to track and respond to local trends of net ownership, use, and maintenance. The President’s

Malaria Initiative (PMI) hosts a LLIN durability monitoring repository to track LLIN field durability among different locations [5] and is an excellent resource for comparing results across studies.

“Net condition” was classified based on standard images provided in the survey forms

(Appendix C), but this was not a standard method of measuring net condition. Using a proportional hole index would have been a standard and more comparable method of collecting

94

this information, but was unavailable for these analyses. Additionally, we did not measure net repair practices, so this could not be included in the net care adherence score.

To describe net storage during the day, we categorized nets that were removed during the day as a moderately adherent practice. However, we do not know where the nets were placed during the day, so there is potential for misclassification if these nets are removed to a space that makes them prone to damage rather than a safe storage space.

It is possible that the person reporting for the household may not be the same person who cares for the nets, which could lead to misclassification. However, there were options for “don’t know” in survey items, hopefully limiting this occurrence. Additionally, social acceptability bias is possible as individuals might report better care practices than are actually performed, particularly if they have high knowledge of net care. Due to lack of data, we approximated net repair behaviors by measuring the number of holes in each net, as nets with holes had not been repaired. Using holes as a proxy could be related to overall net condition and distort the observed association. Another limitation is the inability to distinguish between activities conducted at community events and net distributions. There may be misclassification if some individuals consider a distribution to be a community event. This is unlikely to change the results, however, as in the lowlands almost no one reported receiving education during a distribution and both community events and distributions in the highlands were associated with better adherence.

Specific Aim 2:

While we were able to directly observe repurposed nets, the age or condition of the net when first repurposed is unknown. It is possible that nets were first repurposed when they were still viable for intended use. However, nearly all participants discussed repurposing nets once

95

they were too old or damaged for intended use. Direct measurement of net misuse was difficult.

We defined misuse as repurposed nets that were in physically good condition with no holes or as nets used as intended in the household but reported to have ever been used for an alternative purpose. To obtain more nuanced and accurate measurements of true net misuse, future studies should inquire about the age and physical condition of bed nets when first repurposed.

Specific Aim 3:

Due to the variety of study designs and bed net care and repair practices addressed by education interventions, we were unable to pool results for an overall summary. The search terms

“Plasmodium ” and “ Anopheles ” were not included in the search strategies, and thus could have caused us to miss some records. However, this is likely a minor issue considering we assessed a large number of records from a wide variety of databases. Due to the variety of study designs of the included interventions, the quality assessment used was not a standardized or validated assessment tool. Rather, it is a way to highlight important elements of education interventions and should be evaluated as a description rather than a judgement. We transparently reported the scoring of each included study so that readers can evaluate our decision and make their own judgments (Appendix F).

96

REFERENCES

1. World Malaria Report 2018. Geneva: World Health Organization; 2018. Licence: CC BY-NC-SA 3.0

2. Lengeler C. Insecticide-treated bed nets and curtains for preventing malaria. Cochrane Database Syst Rev. 2009; doi:10.1002/14651858.CD000363.pub2.

3. Augustincic Polec L, Petkovic J, Welch V, Ueffing E, Tanjong Ghogomu E, Pardo Pardo J, Grabowsky M, Attaran A, Wells GA, Tugwell P. (2015) Strategies to increase the ownership and use of insecticide- treated bednets to prevent malaria.Cochrane Database of Systematic Reviews 3, Art. No.: CD009186. DOI: 10.1002/14651858.CD009186.pub2.

4. President’s Malaria Initiative. LLIN durability monitoring: toolkit and data repository for PMI-funded LLIN durability monitoring studies. 2018; https://www.durabilitymonitoring.org Accessed 6 Dec 2018.

5. Kilian A, Koenker H, Obi E, Selby RA, Fotheringham M, Lynch M. Field durability of the same type of long-lasting insecticidal net varies between regions in Nigeria due to differences in household behaviour and living conditions. Mal J. 2015; doi:10.1186/s12936-015-0640-4

6. Helinski MH, Namara G, Koenker H, Kilian A, Hunter G, Acosta A., et al. Impact of a behaviour change communication programme on net durability in eastern Uganda. Mal J. 2015; doi:10.1186/s12936-015-0899-5

7. President’s Malaria Initiative. ITN care and repair: improving net lifespan through behavior change communication. Networks Summary Series. 2015; https://www.k4health.org/sites/default/files/networks_summary_net_care_and_repair_2015-jan_0.pdf Accessed 5 Dec 2018.

8. K4Health. Care of mosquito nets toolkit. 2016; https://www.k4health.org/toolkits/care-repair-LLIN Accessed 6 Dec 2018.

9. Eisele TP, Thwing J, Keating J. Claims about the misuse of insecticide-treated mosquito nets: are these evidence-based? PLoS Med, 2011; doi: 10.1371/journal.pmed.1001019

10. Keating J, Miller JM, Bennett A, Moonga HB, Eisele TP (2009) Plasmodium falciparum parasite infection prevalence from a household survey in Zambia using microscopy and a rapid diagnostic test: implications for monitoring and evaluation. Acta Trop 112 , 277–282.

11. Mclean, K. A., Byanaku, A., Kubikonse, A., Tshowe, V., Katensi, S., & Lehman, A. G. (2014). Fishing with bed nets on Lake Tanganyika: a randomized survey - Kate_A_McLean.pdf, 1–5. Retrieved from http://ihi.eprints.org/2937/1/Kate_A_McLean.pdf

12. Becker-Dreps SI, Biddle AK, Pettifor A, Musuamba G, Imbie DN, Meshnick S, Behets F. (2009) Cost-effectiveness of adding bed net distribution for malaria prevention to antenatal services in Kinshasa, Democratic Republic of the Congo. Am J Trop Med Hyg 81 (3), 496-502.

13. Fernando SD, Abeyasinghe RR, Galappaththy GNL, Gunawardena N, Ranasinghe ACR, Rajapaksa LC (2009) Sleeping arrangements under long-lasting impregnated mosquito nets: differences during low and high malaria transmission seasons. Trans R Soc Trop Med Hyg 103 , 1204-1210.

97

14. Krezanoski PJ, Campbell JI, Santorino D, Bangsberg DR (2017) Objective monitoring of insecticide- treated bednet use to improve malaria prevention: SmartNet development and validation. PLoS ONE 12 (2), e0168116.

15. Shirayama Y, Phompida S, Kuroiwa C, Miyoshi M, Okumura J, Kobayashi J (2007) Maintenance behaviour and long-lasting insecticide-treated nets (LLITNs) previously introduced into Bourapar district, Khammouane province, Lao PDR. J Public Health 121 , 122-129.

16. De La Cruz N, Crookston B, Gray B, Alder S, Dearden K (2009) Microfinance against malaria: impact of Freedom from Hunger’s malaria education when delivered by rural banks in Ghana. Trans R Soc Trop Med Hyg 103 , 1229-1236.

17. Onyango-Ouma W, Aagaard-Hansen J, Jensen BB (2005) The potential of schoolchildren as health change agents in rural western Kenya. Soc Sci Med 61 , 1711-1722.

18. Ayi I, Nonaka D, Adjovu JK, Hanafusa S, Jimba M, Bosompem KM, et al. (2010) School-based participatory health education for malaria control in Ghana: engaging children as health messengers. Malar J 9,98.

19. Michie S, Abraham C, Eccles MP, Francis JJ, Hardeman W, Johnston M (2011) Strengthening evaluation and implementation by specifying components of behaviour change interventions: a study protocol. Implement Sci 6,10.

20. Michie S, van Stralen MM, West R (2011) The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci 6, 42.

21. Kinzie MB (2005) Instructional design strategies for health behaviour change. Patient Ed Counsel 56 , 3-15.

22. Roberts LS, Janovy J (2009) Foundations of Parasitology. 8 th ed. McGraw-Hill. New York, NY.

23. Jelinek T (2010) Vector-borne transmission: malaria, dengue, and Yellow Fever. Pgs 381-394. In modern infectious disease epidemiology: concepts, methods, mathematical models, and public health. Springer Science+Business Media, LLC. New York, NY.

24. Reyburn H, Mbatia R, Drakeley C, Bruce J, Carneiro H, Olomi R, et al. (2005) Association of transmission intensity and age with clinical manifestations and case fatality of severe Plasmodium falciparum malaria. JAMA 293(12), 1461-1470

25. Tangpukdee N, Duangdee C, Wiliaratana P, Krudsood S (2009) Malaria diagnosis: a brief review. Korean J Parasitol 47(2), 93-102

26. Wangai LN, Karou MG, Njiruh PN, Sabah O, Kimani FT, Magoma G, Kiambo N (2011) Sensitivity of microscopy compared to molecular diagnosis of P. falciparum: implications on malaria treatment in epidemic areas of Kenya. Afr J Infect Dis 5(1), 1-6

27. World Health Organization (1951) Report on the malaria conference in equatorial Africa. WHO Technical Report Series 38, 1-72

98

28. Nkumama IN, O’Meara WP, Osier FHA (2017) Changes in malaria epidemiology in Africa and new challenges for elimination. Trends Parasitol 33(2), 128-140

29. World Health Organization (2015) Kenya malaria profile. Global Health Observatory Data. Accessed 12 May 2019

30. Hay SI, Rogers DJ, Shanks GD, Myers MF, Snow RW (2001) Malaria early warning in Kenya. Trends Parasitol 17(2), 95-99

31. Ernst KC, Hayden MH, Olsen H, Cavanaugh JL, Ruberto I, Agawo M, Munga S (2016) Comparing ownership and use of bed nets at two sites with differential malaria transmission in western Kenya. Malar J 15:217 doi: 10.1186/s12936-016-1262-1

32. Malaria (2013) IARC monographs on the evaluation of carcinogenic risks to humans. 104, 141-119

33. Centers for Disease Control and Prevention (2018) Malaria biology. Global Health, Division of Parasitic Diseases and Malaria Accessed 11 May 2019

34. Miller LH, Good MF, Milon G (1994) Malaria pathogenesis. Science 264, 1878-1883.

35. World Health Report 1999. WHO, Geneva, Switzerland

36. Greenwood BM, Collins FH, Duffy PE (2008) Malaria: progress, perils, and prospects for eradication. J Clin Invest 118(4) 1266-1276

37. World Health Organization. Global technical strategy for malaria 2016-2030. Global Malaria Programme. 2015; World Health Organization. http://doi.org/ISBN: 978 92 4 156499 1. Accessed 15 Sep 2018.

38. World Health Organization (2012) Updated WHO policy recommendation: intermittent preventive treatment of malaria in pregnancy using sulfadoxine-pyrimethane (IPTp-SP). Geneva, Switzerland

39. Loll DK, Berthe S, Faye SL, Wone I, Arnold B, Koenker H., et al. “You need to take care of it like you take care of your soul”: perceptions and behaviours related to mosquito net damage, care, and repair in Senegal. Mal J. 2014; doi:10.1186/1475-2875-13-322

40. Hunter GC, Scandurra L, Acosta A, Koenker H, Obi E, Weber R. We are supposed to take care of it: A qualitative examination of care and repair behaviour of long-lasting, insecticide-treated nets in Nasarawa State, Nigeria. Mal J. 2014; doi:10.1186/1475-2875-13-320

41. Miller JE, Jones COH, Ndunguru S, Curtis V, Lines J. A new strategy for treating nets. Part 2: users’ perceptions of efficacy and washing practices and their implications for insecticide dosage. Trop Med Int Health. 1999;4:167–174.

42. Mutuku FM, Khambira M, Bisanzio D, Mungai P, Mwanzo I, Muchiri EME., et al. (2013). Physical condition and maintenance of mosquito bed nets in Kwale County, coastal Kenya. Mal J. 2013; doi:10.1186/1475-2875-12-46

99

43. Roll Back Malaria Partnership (2018) Consensus statement on repurposing ITNs : applications for BCC messaging and actions at the country level

44. De Oliveria AM, Wolkon A, Krishnamoorthy R, Erskine M, Crenshaw DP, Roberts J, Saute F (2010) Ownership and usage of insecticide-treated bed nets after free distribution via a voucher system in two provinces of Mozambique. Malar J 9:222

45. Ye Y, Patton E, Kilian A, Dovey S, Exkert E (2012) Can universal insecticide-treated net campaigns achieve equity in coverage and use? The case of northern Nigeria. Malar J 11:32

46. Bennett A, Smith SJ, Yambasu S, Jambai A, Alemu W, Kabano A, Eisele TP (2012) Household possession and use of insecticide-treated mosquito nets in Sierra Leone 6 months after a national mass distribution campaign. PLoS One 7(5) e37927

47. Zhou G, Li JS, Ototo EN, Atieli HE, Githeko AK, Yan G (2014) Evaluation of universal coverage of insecticide-treated nets in western Kenya: field surveys. Malar J 13:351

48. Kohler LN, Hibler EA, Harris RB, Oren E, Roe DJ, Jurutka PW, Jacobs ET. Greater adherence to cancer prevention guidelines is associated with higher circulating concentrations of vitamin D metabolites in a cross-sectional analysis of pooled participants from 2 chemoprevention trials. J Nutr. 2017; doi:10.3945/jn.116.243352

49. McCullough ML, Patel AV, Kushi LH, Patel R, Willett WC, Doyle C, et al. Following cancer prevention guidelines reduces risk of cancer, cardiovascular disease, and all-cause mortality. Cancer Epid Biomarkers Prev. 2011; doi:10.1158/1055-9965.EPI-10-1173

50. Glanz K, Rimer BK, Lewis FM. Health behavior and health education: theory, research, and practice: 3rd Ed. 2002.

51. Homish GG, Edwards EP, Eiden RD, Leonard KE. Analyzing family data: a GEE approach for substance use researchers. Addict Behav. 2010; doi: 10.1016/j.addbeh.2010.01.002

52. Liang KY, Zeger SL. Regression analysis for correlated data. Annu Rev Publ Health. 1993; doi:10.1146/annurev.pu.14.050193.000355

53. Lumley T. Generalized estimating equations for ordinal data : a note on working correlation structures. Biometrics. 1996;52:354-361.

54. Dillip A, Mboma ZM, Greer G, Lorenz LM. Understanding roles of men and women in net care and repair in southern Tanzania. Mal J. 2018; doi:10.1186/s12936-018-2608-7

55. Marsh VM, Mutemi W, Some ES, Haaland A, Snow RW (1996) Evaluating the community education programme of an insecticide-treated bed net trial on the Kenyan coast. Health Policy Plan, 11 (3), 280-291

56. Minja H, Schellenberg JA, Mukasa O, Nathan R, Abdulla S, Mponda H, Obrist B (2001) Introducing insecticide-treated nets in the Kilombero Valley, Tanzania: the relevance of local knowledge and practice for an information, education and communication (IEC) campaign. Trop Med Int Health, 6 (8), 614-623

100

57. Rhee M, Sissoko M, Perry S, McFarland W, Parsonnet J, Doumbo O. (2005) Use of insecticide- treated nets (ITNs) following a malaria education intervention in Piron, Mali: a control trial with systematic allocation of households. Malar J, 4 , 35. doi:10.1186/1475-2875-4-35

58. Rickard DG, Dudovitz RN, Wong MD, Jen HC, Osborn RD, Fernandez HE, Donkor CI (2011) Closing the gap between insecticide treated net ownership and use for the prevention of malaria. Prog Community Health Partnersh, 5 (2), 123-131. doi:10.1353/cpr.2011.0018

59. Ahmadi M, Vatandoost H, Shaeghi M, Raeisi A, Abedi F, Eshraghian M, et al . (2012) Effects of educational intervention on long-lasting insecticidal nets use in a malarious area, southeast Iran. Acta Med Iran, 50 (4), 279‐287

60. Amoran OE, Fatugase KO, Fatugase OM, Alausa KO (2012) Impact of health education intervention on insecticide treated nets uptake among nursing mothers in rural communities in Nigeria. BMC Research Notes 5:444

61. Alvarado BE, Gomez E, Serra M, Carvajal R, Carrasquilla G. (2006). Evaluación de una estrategia educativa en malaria aplicada en localidades rurales del Pacífico colombiano. Biomedica, 26 (3), 342-352.

62. Nonaka D, Kobayashi J, Jimba M, Vilaysouk B, Tsukamoto K, Kano S, et al (2008) Malaria education from school to community in Oudomxay province, Lao PDR. Parasitol International 57 , 76-82

63. Mfaume MS, Winch PJ, Makemba AM, Premji Z (1997). Mosques against malaria. World Health Forum, 18 (1), 35-38

64. Panter-Brick C, Clarke SE, Lomas H, Pinder M, Lindsay SW (2006) Culturally compelling strategies for behaviour change: a social ecology model and case study in malaria prevention. Soc Sci Med, 62 (11), 2810-2825. doi:10.1016/j.socscimed.2005.10.009

65. Harvey SA, Olórtegui MP, Leontsini E, Asayag CR, Scott K, Winch PJ. Trials of improved practices (TIPs): A strategy for making long-lasting nets last longer? Am J Trop Med Hyg. 2013; doi:10.4269/ajtmh.12-0641

66. Koenker H, Kilian A, Hunter G, Acosta A, Scandurra L, Fagbemi B, Lynch M (2015) Impact of a behaviour change intervention on long-lasting insecticidal net care and repair behaviour and net condition in Nasarawa State, Nigeria. Malar J, 14 , 18. doi:10.1186/s12936-014-0538-6

67. Widmar M, Nagel CJ, Ho DY, Benziger PW, Hennig N (2009) Determining and addressing obstacles to the effective use of long-lasting insecticide-impregnated nets in rural Tanzania. Malar J, 8 , 315. doi:10.1186/1475-2875-8-315

68. Ho D. (2012) An approach to achieve sustainable bed-net use in rural Tanzania. (1509713), Mount Sinai School of Medicine, Ann Arbor

69. Alonso PL, Lindsay SW, Armstrong Schellenberg JR, Gomez P, Hill AG, David PH, Greenwood BM (1993) A malaria control trial using insecticide-treated bed nets and targeted chemoprophylaxis in a rural area of The Gambia, west Africa. 2. Mortality and morbidity from malaria in the study area. Trans R Soc Trop Med Hyg, 87 Suppl 2 , 13-17.

101

70. Brieger W, Onyido A, Sexton J, Ezike V, Breman J, Ekanem O (1996) Monitoring community response to malaria control using insecticide-impregnated bed nets, curtains and residual spray at Nsukka, Nigeria. Health education research, 11 (2), 133‐145. Retrieved from https://www.cochranelibrary.com/central/doi/10.1002/central/CN-00183839/full

71. Khadka A, Perales NA, Wei DJ, Gage AD, Haber N, Verguet S, et al . (2018) Malaria control across borders: quasi-experimental evidence from the Trans-Kunene malaria initiative (TKMI). Malar J, 17 (1), 224. doi:10.1186/s12936-018-2368-4

72. Kachur SP, Phillips-Howard PA, Odhacha AM, Ruebush TK, Oloo AJ, Nahlen BL (1999) Maintenance and sustained use of insecticide-treated bednets and curtains three years after a controlled trial in western Kenya. Trop Med Int Health, 4 (11), 728-735.

73. Snow RW, McCabe E, Mbogo CN, Molyneux CS, Some ES, Mung'ala, VO, Nevill CG (1999) The effect of delivery mechanisms on the uptake of bed net re-impregnation in Kilifi District, Kenya. Health Policy Plan, 14 (1), 18-25.

74. Wacira DG, Hill J, McCall PJ, Kroeger A (2007) Delivery of insecticide-treated net services through employer and community-based approaches in Kenya. Tropical medicine & international health, 12 (1), 140-149. doi:10.1111/j.1365-3156.2006.01759.x

75. Fernando SD, Abeyasinghe RR, Galappaththy GN, Gunawardena N, Rajapakse LC (2008) Community factors affecting long-lasting impregnated mosquito net use for malaria control in Sri Lanka. Trans R Soc Trop Med Hyg, 102 (11), 1081-1088. doi:10.1016/j.trstmh.2008.06.007

76. Cham MK, D'Alessandro U, Todd J, Bennett S, Fegan G, Cham BA, Greenwood BM (1996) Implementing a nationwide insecticide-impregnated bednet programme in The Gambia. Health Policy Plan, 11 (3), 292-298.

77. Hiwat H, Hardjopawiro LS, Takken W, Villegas L (2012) Novel strategies lead to pre-elimination of malaria in previously high-risk areas in Suriname, South America. Malar J, 11 , 10. doi:10.1186/1475- 2875-11-10

78. Gerstl S, Dunkley S, Mukhtar A, Maes P, De Smet M, Baker S, Maikere J (2010) Long-lasting insecticide-treated net usage in eastern Sierra Leone - the success of free distribution. Trop Med Int Health, 15 (4), 480-488. doi:10.1111/j.1365-3156.2010.02478.x

79. Ahmed SM, Hossain S, Kabir MM, Roy S (2011) Free distribution of insecticidal bed nets improves possession and preferential use by households and is equitable: findings from two cross-sectional surveys in thirteen malaria endemic districts of Bangladesh. Malar J, 10 , 357. doi:10.1186/1475-2875-10-357

80. Alaii J, Van DBH, Kachur S, Mwenesi H, Vulule J, Hawley W, et al . (2003) Perceptions of bed nets and malaria prevention before and after a randomized controlled trial of permethrin-treated bed nets in western Kenya. American journal of tropical medicine and hygiene, 68 (4 Suppl), 142‐148

81. Against Malaria Foundation (2019). Distributions. < https://www.againstmalaria.com/Distributions.aspx?MapID=1&StatusID=7> Accessed 11 May 2019

82. Caviness C, Gish A. Five ways The United Methodist Church has impacted malaria. Imagine No

102

Malaria. Accessed 14 March 2019

83. Malaria Consortium. Pioneer Project 2009-2014: a holistic systems strengthening approach towards malaria control in mid-western Uganda. Successes and lessons.

84. Malaria Consortium (2018) Nets bring good health: a qualitative inquiry into mosquito net use and care practices in two northern provinces of Mozambique. London, United Kingdom Accessed 14 March 2019

85. Kilian A, Byamukama W, Pigeon O, Gimnig J, Atieli F, Koekemoer L, et al. Evidence for a useful life of more than three years for a polyester-based long-lasting insecticidal mosquito net in western Uganda. Mal J. 2011; doi:10.1186/1475-2875-10-299

86. Leonard L, Diop S, Doumbia S, Sadou A, Mihigo J, Koenker H, et al. Net use, care and repair practices following a universal distribution campaign in Mali. Mal J. 2014; doi:10.1186/1475-2875-13- 435

87. Khambira, M. Insecticide treated bednets ownership, use and maintenance behaviour in Kwale, Msambweni and Kinango districts in Kenya. http://ir- library.ku.ac.ke/bitstream/handle/123456789/7046/Maureen Khambira.pdf;sequence=1. 2013; Accessed 6 Dec 2018.

88. Erlanger T. Field issues related to insecticide-treated nets in Tanzania. MSc Thesis. University of Basel and Swiss Tropical Institute. 2002; 153–160.

89. Kilian A, Byamukama W, Pigeon O, Atieli F, Duchon S, Phan C. Long-term field performance of a polyester-based long-lasting insecticidal mosquito net in rural Uganda. Mal J. 2008; doi:10.1186/1475- 2875-7-49

90. Atieli FK, Munga SO, Ofulla AV, Vulule JM. The effect of repeated washing of long-lasting insecticide-treated nets (LLINs) on the feeding success and survival rates of Anopheles gambiae. Mal J. 2010; doi:10.1186/1475-2875-9-304

91. World Health Organization. Guidelines for laboratory and field-testing of long-lasting insecticidal nets. 2015; ISBN 978 92 4 1505277.

92. Mejía P, Teklehaimanot HD, Tesfaye Y, Teklehaimanot A. Physical condition of Olyset® nets after five years of utilization in rural western Kenya. Mal J. 2013; doi:10.1186/1475-2875-12-158

93. Scandurra L, Acosta A, Koenker H, Kibuuka D, Harvey S. “It is about how the net looks”: a qualitative study of perceptions and practices related to mosquito net care and repair in two districts of eastern Uganda. Mal J. 2014;13:504.

94. Allan R, O’Reilly L, Gilbos V, Kilian A. An observational study of material durability of three world health organization-recommended long-lasting insecticidal nets in Eastern Chad. Am J Trop Med Hyg. 2012; doi:10.4269/ajtmh.2012.11-0331

103

95. Smith SC, Joshi UB, Grabowsky M, Selanikio J, Nobiya T, Aapore T. (2007). Evaluation of bednets after 38 months of household use in northwest Ghana. Am J Trop Med Hyg. 2007;77(6):243-248.

96. Norris LC, Norris DE. Efficacy of long-lasting insecticidal nets in use in Macha, Zambia, against the local Anopheles arabiensis population. Mal J. 2011; doi:10.1186/1475-2875-10-254

97. Atiele H, Zouh G, Yaw A, Mwanzo I, Githeko A, Yan G. Insecticide-treated net (ITN) ownership, usage, and malaria transmission in the highlands of western Kenya. Parasit Vectors. 2011; doi:10.1186

98. Gnanguenon V, Azondekon R, Oke-Agbo F, Beach R, Akogbeto M. Durability assessment results suggest a serviceable life of two, rather than three, years for the current long-lasting insecticidal (mosquito) net (LLIN) intervention in Benin. BMC Infect Dis. 2014; doi:10.1186/1471-2334-14-69

99. Hakizimana E‚ Cyubahiro B‚ Rukundo A‚ Kabayiza A‚ Mutabazi A‚ Beach R‚ et al. Monitoring long- lasting insecticidal net (LLIN) durability to validate net serviceable life assumptions‚ in Rwanda. Malar J. 2014;13:344

100. Wills AB, Smith SC, Anshebo GY, Graves PM, Endeshaw T, Shargie EB, et al. Physical durability of permanet 2.0 long lasting insecticidal nets over three to 32 months of use in Ethiopia. Mal J. 2013; doi:10.1186/1475-2875-12-242

101. Briët OJT, Penny MA. Repeated mass distributions and continuous distribution of long-lasting insecticidal nets: modelling sustainability of health benefits from mosquito nets, depending on case management. Mal J. 2013; doi:10.1186/1475-2875-12-401

102. Kibe LW, Kamau AW, Gachigi JK, Habluetzel A, Mbogo CM. A formative study of disposal and re- use of old mosquito nets by communities in Malindi, Kenya. MWJ, 2015;6,9

103. Lover AA, Sutton BA, Asy AJ, Wilder-Smith A. An exploratory study of treated-bed nets in Timor- Leste : patterns of intended and alternative usage. Malar J, 2011:10,199

104. Minakawa N, Dida GO, Sonye GO, Futami K, Kaneko S. Unforeseen misuses of bed nets in fishing villages along Lake Victoria. Malar J, 2008; doi: 10.1186/1475-2875-7-165

105. Baume CA, Reithinger R, Woldehanna S. Factors associated with use and non-use of mosquito nets owned in Oromia and Amhara Regional States, Ethiopia. Malar J, 2009; doi:10.1186/1475-2875-8-264

106. Loll DK, Berthe S, Faye SL, Wone I, Koenker H, Arnold B, Weber R. User-determined end of net life in Senegal: A qualitative assessment of decision-making related to the retirement of expired nets. Malar J, 2013; doi:10.1186/1475-2875-12-337

107. Doda Z, Solomon T, Loha E, Gari T, Lindtjørn B. A qualitative study of use of long-lasting insecticidal nets (LLINs) for intended and unintended purposes in Adami Tullu, East Shewa Zone, Ethiopia. Malar J, 2018; doi:10.1186/s12936-018-2209-5

108. Koenker H, Kilian A, De Beyl CZ, Onyefunafoa EO, Selby RA, Abeku T, et al. What happens to lost nets: a multi-country analysis of reasons for LLIN attrition using 14 household surveys in four countries. Malar J, 2014; doi:10.1186/1475-2875-13-464

104

109. Hoibak S (2010) Presentation of preliminary findings: long lasting insecticide treated net coverage and durability surveys, Dadaab Refugee Camps; 2010

110. Tan KR Coleman J Smith B Hamainza B Katebe-Sakala C Kean C Kowal A Vanden Eng J Parris TK Mapp CT Smith SC Wirtz R Kamuliwo M Craig AS A longitudinal study of the durability of long- lasting insecticidal nets in Zambia. Malar J. 2016 Feb 19;15:106 Janz NK, Becker MH. The health belief model: a decade later. Health Education Quarterly 1984; 11(1) 1- 47

111. Honjo, K., & Satake, A. (2014). N-player mosquito net game: Individual and social rationality in the misuse of insecticide-treated nets. Journal of Theoretical Biology , 342 , 39–46. http://doi.org/10.1016/j.jtbi.2013.11.001

112. Habonimana, D. (2017). Improving Long-Lasting Insecticidal Nets use in Kayange community of North-Western Burundi : A Quality Improvement Study A research report submitted to the Faculty of Health Science in partial fulfillment of the requirements for the degree of Master of Science ( MSc ) in Epidemiology - Implementation Science.

113. Larsen, D. A., Welsh, R., Mulenga, A., & Reid, R. (2018). Widespread mosquito net fishing in the Barotse floodplain: Evidence from qualitative interviews. PLoS ONE , 13 (5), 1–16. http://doi.org/10.1371/journal.pone.0195808

114. Short, R., Gurung, R., Rowcliffe, M., & Hill, N. (2018). The use of mosquito nets in fisheries : A global perspective, 1–14.

115. Public Health Service (2003) Toxicological profile for pyrethrins and pyrethroids. U.S. Department of Health and Human Services. Atlanta, Georgia

116. Bush, E. R., Short, R. E., Milner-Gulland, E. J., Lennox, K., Samoilys, M., & Hill, N. (2017). Mosquito Net Use in an Artisanal East African Fishery. Conservation Letters , 10 (4), 450–458. http://doi.org/10.1111/conl.12286

117. Haya, K. (1989). Toxicity of pyrethroid insecticides to fish. Environmental Toxicology and Chemistry , 8(5), 381–391. http://doi.org/10.1002/etc.5620080504

118. Hill, I. R. (1989). Aquatic organisms and pyrethroids. Pesticide Science , 27 (4), 429–457. http://doi.org/10.1002/ps.2780270408

119. Coalson J, Santos EM, Little AC, Anderson E, Stroupe N, Agawo M, Munga S, Ernst KC (2019) Inadequate ratio of long-lasting insecticidal nets (LLIN) to household members challenges universal coverage in western Kenya: a cross-sectional study Under Review

120. Glanz K, Bishop DB (2010) The role of behavioral science theory in development and implementation of public health interventions. Ann Rev Public Health 31 , 399-418.

121. The RBM Partnership to End Malaria (2018) The strategic framework for malaria social and behavior change communication 2018-2030. Venier, Switzerland: RBM

122. Glanz K, Lewis FM, Rimer BK (1997) Linking theory, research, and practices. In: Glanz K, Lewis

105

FM, Rimer BK, editors. Health behavior and health education. 2 nd ed. San Francisco: Jossey-Bass; 3-18

123. RBM Partnership to End Malaria (2019) The art of asking questions: priority research areas and approaches for malaria social and behavior change programs. Venier, Switzerland: RBM.

124. Hartjes LB (2010) The development and evaluation of a malaria risk-reduction game for travelers. (3448913), The University of Wisconsin - Madison, Ann Arbor.

125. Burgstahler S (2009) Universal design for instruction (UDI): definition, principles, guidelines, and examples. University of Washington

126. The Health Communication Capacity Collaborative (HC3). (2017) Malaria SBCC evidence literature review. Baltimore, Maryland: Johns Hopkins Center for Communication Programs

127. Thomas S (2004) What is participatory learning and action (PLA): an introduction. Centre for International Development and Training (CIDT); London

128. RBM Partnership to End Malaria (2017) Malaria social and behavior change communication indicator reference guide: second edition. Venier, Switzerland: RBM

129. RBM Partnership to End Malaria (2019) Developing M&E plans for malaria social and behavior change programs: a step-by-step guide. Venier, Switzerland: RBM

106

APPENDIX A: DETERMINATION OF HUMAN RESEARCH

107

APPENDIX B: MANUSCRIPT 1

Bed net care practices and associated factors in western Kenya

Ellen M Santos* 1 [email protected] Corresponding Author

Jenna E Coalson 1 [email protected]

Elizabeth T Jacobs 1 [email protected]

Yann C Klimentidis 1 [email protected]

Stephen Munga 3 [email protected]

Maurice Agawo 3 [email protected]

Elizabeth Anderson 1 [email protected]

Nancy Stroupe 1 [email protected]

Kacey C Ernst 1 [email protected]

1 Mel and Enid Zuckerman College of Public Health, University of Arizona, 1295 N Martin Ave, Tucson, AZ 85724

3 Centre for Global Health Research, Kenya Medical Research Institute, PO Box 1578, 40100 Kisumu, Nyanza, Kenya

Abstract

Background: Insecticide treated nets (ITNs) and long-lasting insecticidal nets (LLINs) are effective for malaria prevention and are designed to provide nearly five years of mosquito protection. However, many ITNs and LLINs become damaged and ineffective for mosquito bite

108

prevention within one to two years in field conditions. Non-adherence to recommended bed net care and repair practices may partially explain this shortened net longevity. Methods: Using data from a cross-sectional study, we developed a net care adherence score and described adherence to net care practices in two regions of western Kenya. We then measured the relationships between attitudes and environmental factors that influence net longevity with adherence to bed net care practices. Results: While overall care practices are highly adherent particularly in the highlands, practices related to daily storage, washing frequency, and drying location need improvement in the lowlands. Seventy-seven percent of nets in the lowlands were washed <3 months prior to the survey compared to 23% of nets in the highlands. More nets were dried in the sun in the lowlands (32% of nets) compared to the highlands (4% of nets). Different elements of care are influenced by various malaria attitudes and environmental factors, highlighting the complexity of factors associated with net care. For example, households that learned about net care from community events, that share a sleeping structure with animals, and that have nets used by adult males tend to adhere to washing frequency recommendations. Conclusions: In western Kenya, many nets are cared for in accordance to recommended practices, particularly in the highlands sites. In the lowlands, demonstrating methods at community events to tie nets up during the day coupled with messaging to emphasize infrequent washing and drying nets in the shade may be an appropriate intervention. As illustrated by differences between the highlands and lowlands sites in the present study, should interventions to improve adherence to bed net care practices be necessary, they should be context-specific.

Key Words

109

Malaria Kenya LLIN ITN Bed nets Care and repair

Background

Insecticide treated nets (ITNs) and long-lasting insecticidal nets (LLINs) are effective malaria prevention tools. Notably, ITNs have reduced malaria incidence by half in children under 5 in areas of stable malaria transmission [1]. The primary strategy for malaria prevention in sub-Saharan Africa, according to the WHO, has therefore become the distribution of LLINs

[2]. To achieve their optimal effectiveness, bed nets must be available as a resource (supply), accessible to households (ownership and universal access), used regularly (use) and well maintained to maximize efficacy (care and repair).

While some LLIN distribution programs have shifted to combinations of continuous and mass distribution models [3], many malaria control programs still distribute LLIN every 3-5 years, the anticipated optimal lifespan of a LLIN [4]. Initiated in 2005 under President George

W. Bush, the President’s Malaria Initiative (PMI) developed a network of monitoring sites to measure realistic LLIN durability and lifespan in home conditions [5,6]. Studies from these monitoring sites find variable serviceable lifespans generally between 2-3 years, though significant damage noticed from as little as 6 months to 4.5 years, with differences by net brand and variations in insecticidal activity and attrition [5,7-15] These studies also suggest that inadequate bed net maintenance behaviors contribute to this shortened lifespan [8].

In an effort to standardize terminology for research into bed net maintenance attitudes and practices, the NetWorks project established the phrase “care and repair”. The project

110

developed a conceptual framework of the knowledge, attitudes, intentions, social norms, and behaviors related to care and repair practices [16]. Proper care and repair of LLINs increases their effective longevity, maintaining higher and more stable coverage of vector control between mass community distributions [5,8,9]. Recommended LLIN care practices include hanging nets up while not in use, washing gently and infrequently and with no soap or mild soap, drying indoors or in the shade, and repairing holes [17]. Reported adherence to these practices has been mixed, with high adherence to some practices and low adherence to others. A study in Laos examined maintenance behaviors 2-3 years after LLIN distribution and found that 38.2% of households followed the recommended washing frequency, 83% followed drying recommendations, and 32.7% of households with torn LLINs had repaired them [18]. However, the results showed weak associations between maintenance behaviors and malaria knowledge and prevention attitudes [18], obfuscating how educational interventions may be designed to improve adherence. Other studies qualitatively describe how caring for the family and saving money motivate bed net care and repair, though few LLINs in the study area were actually repaired [19,20]. Perceptions of social acceptability and color of the LLIN have also been associated with washing frequency, possibly due to social implications of having a dirty net

[21,22].

These previous studies demonstrate that while social norms and individual perceptions and knowledge influence how LLIN care and repair is conducted, these relationships are complex and vary among communities. Additionally, the NetWorks conceptual framework includes environmental factors such as household attributes that may influence the practicality of adhering to optimal care and repair practices [16]. In this study we examine associations between care and repair behaviors and attitudes and environmental factors from the NetWorks conceptual

111

framework [16] in western Kenya (Figure 1). Understanding factors related to appropriate net care and repair behaviors will clarify what messaging or other intervention strategies are needed, and where/how they should be targeted to extend the real-world longevity of LLINs.

Figure 1. Attitudes and environmental factors related to adherence to net care and repair behaviors, adapted from the NetWorks conceptual framework [16].

Methods

Overview: We describe bed net care practices from a cross-sectional study administered in two regions of western Kenya with differing Plasmodium transmission patterns. Though the vast majority (95.3%) of nets were LLINs, the term ‘bed net’ or ‘net’ was chosen because we could not verify the brand or type of all nets in the sample. We developed a bed net care adherence score and examined the relationship between overall adherence to care recommendations and net condition. We then examined the relationship between adherence and attitudes and environmental elements of the NetWorks conceptual model [16]

Study design and data collection: Household surveys were administered during a cross- sectional study in western Kenya. Surveys were administered between June and August of 2015 in two sites (highlands and lowlands) of differing Plasmodium transmission patterns. Data collected included household, individual, and bed net items encompassing demographic information, malaria history, knowledge and perceptions, and bed net attributes, including ownership, universal access, care, and use. Bed net and household level data were used for the present analyses.

112

Study sites : Kapkangani location encompasses the rural highland sublocations of

Chepsonoi, Kiborgok, and Tindinyo at an altitude range of 1600 – 2100 m above sea level and seasonal Plasmodium transmission following the rainy season from April-May. In this region, the population is primarily comprised of Kalenjin and Luhya ethnic groups. The primary occupation is rural subsistence agriculture, while some work as casual laborers in the commercial tea industry. Miwani is located on the Kano Plain and includes the rural lowland sublocations of

Kabar Central and Kabar West at an altitude of 1200 m above sea level. Transmission is holo- endemic and year-round with seasonal increases. People are primarily of the Luo ethnic group and are subsistence farmers with some employed by area commercial sugarcane and rice growers.

Household selection: The study site was mapped and enumerated prior to sample selection. Households were randomly selected from an enumerated list. Households were oversampled by 20% and then randomly ordered. Visits were made to each household in subsequent order until the required sample size for each village had been met. Male or female heads of household were approached in person by field team staff for recruitment. Households were included if they had resided in the study area for a period of at least one month. Households of all sizes and configurations were included. A household was defined as individuals who regularly eat meals together. Data from 1,217 households (n = 640 in Kapkangani and n = 572 in

Miwani) were used in the present study.

Ethical considerations: The current analysis involved secondary analyses of the de- identified data from the primary study. The primary study was approved by the Kenyan Medical

Research Institute (KEMRI) (SSC #2810) and the University of Arizona ethical review boards with deferral of primary oversight from the University of Arizona to KEMRI. Informed consent

113

was obtained from all household heads. Parental consent was obtained for children less than 7 years of age, and assent was obtained for those 8 to 17 years of age.

Bed Net Care Adherence Score: Data on each net owned by the household were collected in addition to overall net care practices reported by the household head. A 7-item care scoring system was developed from factors including net storage, washing, and drying,. Each net received a score between 0-2 for each item and the scores were totaled for an overall net care adherence score, ranging from 0-14 in which lower scores indicate better adherence (Table 1).

This scoring system was chosen to be more flexible and nuanced than a strict binary system of good-inadequate care. Similar scoring systems are used in cancer prevention literature [24,25].

We measured the association between net care adherence score with overall net condition. Field staff classified each net as excellent, good, fair, or poor condition in comparison to standard images provided on the survey forms (See Additional File 1).

Individual net care elements with ≥ 20% poor adherence (scores of 2) were then selected for further analysis, as these indicate the elements that may need improvement in the study population. We measured the associations of attitudes and environmental factors with adherence to these selected individual net care elements to inform focused interventions.

Table 1. Elements of the bed net care adherence scoring system. Scores of 0 indicate best practice, scores of 1 indicate moderately adherent practice, and scores of 2 indicate non-adherent practice. Wash Where Drying Ever Score Storage Soap Type Wash Manner Freq Washed Location Retreat None; Never if 0 Washtub or Not scrubbed/ Indoors; Never LLIN; ever Best Tied up Never bucket; beaten; Never Never Wash or if ITN ≥ 6 Practice Never wash wash wash Bar soap months 1 Remove ≥ 3 Outdoors Ever if Moderate Lake Detergent _____ d months in shade LLIN Practice

114

2 Never if Non- Leave < 3 Stream or Scrubbed or ITN if net Bleach In the sun adherent as is months river beaten age ≥ 6 Practice months

Attitudes and Environmental Factors: Knowledge of bed net care is defined as the interviewee’s (household’s) knowledge of particular net care and repair practices taken from an open-ended survey question. Participants were asked what things they should do to take care of their bed net. Whether participants mentioned a particular net care practice indicated knowledge of that practice. Risk perception operates both intra and inter-personally [26]. Perception of malaria risk was measured from participant’s perception of the seriousness of malaria for their family (intrapersonal) and for their community (interpersonal). Both of these items are scored on a 5-point likert scale where 1 indicates ‘not at all serious’, and 5 ‘extremely serious’. We added these items to obtain an overall risk perception score (range 2-10). Confidence in the ability to prevent malaria indicates how well participants felt they are able to protect their family from malaria from a 5-point likert scale where 1 indicates ‘not at all’ and 5 ‘extremely well’.

Household structural factors include both family and physical structures. Family structure includes the bed net user , which is a series of binary variables indicating which household members slept under the net. Net user variables measured in years include: child < 5, child 6-18, adult male 19-50, adult female 19-50, adult male 50+, and adult female 50+. Source of Net Care

Knowledge includes a series of binary variables indicating where the interviewee learned net care. These variables include: clinic staff, community education session, or during a net distribution. Self reported ability to pay for a bed net is a binary variable where ‘yes’ indicates the household could afford a bed net if one was not provided for free. The ability to pay for a net may influence how well nets are cared for depending if it is easy or difficult to obtain a new net.

115

Physical structural factors include universal access , defined as households that meet the

WHO recommendation of one net for every two household members, amount of shade in the house compound , defined as no shade, little, half-shaded, or very shaded, and animals sleep in structure at night , a binary variable where ‘yes’ indicates an animal shares the sleeping structure space at night.

Statistical Analyses: Stata 12.1 (College Station, TX) was used for all analyses with

α=0.05 at the bed net level. Due to the clustered, dependent nature of the data (multiple nets per household), Generalized Estimating Equations (GEEs) were used to measure the association between net condition and overall net care adherence score, and to measure the associations between the identified key independent variables and binary bed net care adherence elements.

Binary care adherence elements were used to assess associations with non-adherent practices, thus best and moderate practices were grouped together to serve as the reference category. GEEs are flexible and ideal for analyzing dependent data that do not meet the assumptions of classical regression techniques [27,28].

These models can also be used for ordinal discrete data [29]. The GEE parameters used for adherence score analyses were adjusted for bed net age and included Gaussian family, identity link function, exchangeable correlation structure, and robust standard errors. The parameters used for individual net care element analyses included binomial family, identity link function, exchangeable correlation structure, and robust standard errors. GEE models assessing washing frequency were adjusted for net age because this element was measured as “when the net was last washed”. Because the highlands and lowlands differed on nearly all demographic characteristics, level of malaria endemicity, and history of LLIN distributions and community education programs, GEEs were run separately for each site

116

Results

Net ownership was greater in the lowlands where 98% of households owned at least one net compared to 68.6% of households in the highlands (Table 2). Most households had one sleeping structure, though households in the highlands tended to have both more individuals in the household as well as more sleeping structures than in the lowlands, where 29.6% vs. 10.2% had more than one structure, respectively. Females are primarily responsible for net care [30], thus it is important to understand attributes of female household heads. In both sites, most female household heads had some primary education or have completed primary education

(Table 2).

Table 2: Household characteristics of study sample by site Highlands Lowlands (643 households) (574 households) N (%) N (%) People in household 1-3 241 (37.5) 382 (66.6) 4-6 296 (46.0) 169 (29.4) ≥ 7 106 (16.5) 23 (4.0)

Sleeping structures in household 1 453 (70.5) 516 (89.9) 2 157 (24.4) 50 (8.7) 3 27 (4.2) 7 (1.3) 4 5 (0.8) 1 (0.2) 5 1 (0.2) 0

Nets per household 0 203 (31.6) 11 (1.9) 1 167 (26.0) 318 (55.4) 2 154 (24.0) 185 (32.2) 3 86 (13.4) 45 (7.8) 4 28 (4.4) 14 (2.4) ≥ 5 5 (0.8) 1 (0.2) Education of the female household head None 55 (8.6) 22 (3.8) Some Primary 261 (40.6) 200 (34.8) Completed Primary 152 (23.6) 190 (33.1) Some Secondary 66 (10.3) 57 (9.9) Completed Secondary 30 (4.7) 39 (6.8) Any Tertiary 20 (3.1) 8 (1.4) No female household head 42 (6.5) 50 (8.7)

117

LLIN Care Adherence : Bed nets in the highlands had better overall care adherence scores compared to the lowlands. Adherence in the highlands was quite good, where nets had a median adherence score of 2, compared to scores in the lowlands, that while still good, had a lower adherence score of 4 (possible range 0-14) (Table 3). When each care element was measured individually, 2 elements in the highlands and 3 in the lowlands had inadequate adherence (<80% of nets cared for according to recommendations) (Figure 2). Adherence to washing location (bucket vs. river or stream), washing manner (gently washed or scrubbed/beaten), soap type, and retreating practices were adequate in both sites and excluded from further analyses. However <80% of nets met adherence recommendations for daily storage, washing frequency, and drying location in at least one of the sites (highlands or lowlands).

Frequent net washing is not recommended, and nets in the lowlands were washed more frequently (77% <3 months) than the highlands (23% <3 months). In the lowlands, 32% of nets were inappropriately dried in the sun compared to only 4% in the highlands (Figure 2).

Figure 2. Individual bed net care adherence score components by site. Scores for components in the highlands are the left-sided, bold colored bars, and the scores for components in the lowlands are the right-sided, pale colored bars.

Attitudes and Environmental Factors: Households in both sites were most knowledgeable about proper drying practices (drying nets out of the sun). In the highlands,

59.6% of households specifically mentioned nets should not be dried in the sun compared to

49.0% of households in the lowlands (Table 3). This knowledge translated to practice in the highlands (4% of nets dried in the sun), but less so in the lowlands (32% of nets dried in the sun)

(Figure 2). Notably in both sites only 25 (3.9%) of households in the highlands and 44 (7.7%) in the lowlands mentioned they should not wash their nets or wash them infrequently, suggesting

118

that messaging regarding proper washing frequency has not been well communicated in these sites (Table 3). Perception of malaria risk and reported confidence in ability to prevent malaria were both greater in the lowlands compared to the highlands (Table 3).

Nearly a quarter of nets were used by at least one child <5 years in both sites. The distribution of net users is similar in both sites, though more nets were used by males 19-50 years in the lowlands compared to the highlands. Significantly more households in the lowlands

(87.1%) reported the ability to afford a bed net compared to households in the highlands (68.8%)

(Table 3).

Most households learned about net care and repair practices from clinics in both sites.

Approximately 20% of households learned from community events and nearly 20% from net distributions in the highlands, in contrast with the lowlands where 38.8% learned from community events and only 0.34% learned from a net distribution.

Factors Associated with Net Care Adherence: There was no association between net condition and overall net care adherence score in the highlands (Table 4). However in the lowlands, poor net condition was associated with worse overall net care adherence score after adjustment for net age. (Table 4).

Table 4: GEE results of association between net condition and overall net care adherence score, adjusted for net age

Highlands Lowlands Overall Net Condition Coefficient 95% CI Coefficient 95% CI Excellent Ref Ref Good 0.48 -0.34, 1.29 0.46 -0.45, 1.36 Fair 0.61 -0.16, 1.38 0.64 0.42, 0.86 Poor -0.27 -1.75, 1.22 0.96 0.50, 1.41 Associations measured using GEE models with net care adherence as an ordinal variable. Adjusted for net age.

119

Different attitudes and environmental factors were associated with elements of net care in different ways (Table 5). Knowledge of net care, and nets used by children <5 years, children 6-

18 years, adult males 19-50 years, and adult males >50 years were associated with better adherence (lower scores). Nets used by adult females >50 years were associated with worse adherence (higher scores). Many independent variables had positive associations with some elements of the net care adherence score, and negative associations with others, illustrating the complexity of the factors that influence net care behaviors.

As compared to those properly stored during the day, nets that remained hanging were more likely to belong to a household with high confidence to prevent malaria and less likely to have been used by a child <5 years in both sites (Table 5). This suggests that households with a sense of self-efficacy to prevent malaria were less likely to properly store bed nets, and nets used by young children were stored properly during the day. Other factors were positively and negatively associated with storage practices, but different across sites (Table 5).

Households that learned about net care from community events, that shared a sleeping structure with animals, and that had nets used by adult males tended to adhere to washing frequency recommendations (Table 5). We hypothesized this difference could be due to housing structural factors where dustier environments may be associated with more frequent washing.

However, no association was found between having dirt floors and washing behaviors in either site (Table 5).

There was a remarkable difference in drying practices between sites. In the lowlands,

32% of nets were dried in the sun compared to only 4% in the highlands (Figure 2). Drying in the shade may not be feasible for individuals depending upon the environment surrounding their compound, so we examined whether amount of shade was associated with drying practices, but

120

found no association, though households in the lowlands had less shade compared to households

in the highlands (Table 3). Interestingly, nets improperly dried in the sun were less likely to

belong to a household with high knowledge of proper drying practices, indicating that increasing

knowledge may positively influence this practice. Source of net care knowledge was similarly

associated, suggesting that community events were effective in their messaging regarding proper

drying practices (Table 5).

Table 5: Odds ratios (95% Confidence Intervals) of univariate GEE models measuring the associations between attitudes and environmental factors with binary net care practices

Highlands Lowlands

Attitudes and Wash Dry Storage Wash Freq A Storage Environmental Factors Freq A Location Knowledge of 0.86 (0.80, 0.92) 0.86 (0.80, appropriate care 0.92)

Perception of malaria risk 1.03 (1.02, 1.05)

Confidence to prevent 1.06 (1.02, 1.11) 1.08 (1.04, 1.13) malaria

Net User Child <5 0.92 (0.86, 0.99) 0.93 (0.87, 0.98) Child 6-18 0.91 (0.85, 0.97) Adult male 19-50 Adult female 19-50 Adult male >50 0.93 (0.91, 0.96) Adult female >50 1.27 (1.14, 1.40)

1.10 (1.01, 1.20) 1.15 (1.04, Ability to buy net 1.29)

Source of Net Care Knowledge 0.87 (0.82, 0.93) 1.10 (1.02, Clinic 1.19) 0.87 (0.81, 0.93) 0.93 (0.90, 0.96) 1.24 (1.17, 1.33) 1.07 (1.01, 0.87 (0.81, Community event 1.13) 0.94) Net distribution 1.20 (1.09, 1.33)

Universal Access (1 net: 2 1.11 (1.03, 1.19) people)

Number of trees around N/A N/A N/A N/A

121

compound

Sleeping Structure Attributes 1.07 (1.01, 1.14) 1.14 (1.04, Earth floors 1.26) Animals occupy sleeping 1.13 (1.03, 1.24) 0.95 (0.92, 0.97) 0.92 (0.86, 0.99) structure A Adjusted for net age; Blue color indicates associations with better adherence, orange color indicates associations with worse adherence, White indicates no associaton

Discussion

Net care adherence scores were generally low (adherent) in both the highlands and

lowlands of western Kenya. Nets had median adherence scores of 2 and 4, respectively, with 0

being the optimal score out of a possible 14. However, there are a few key care behaviors that

could be targeted for improvement in the lowlands. Net storage, washing frequency, and drying

are the key areas for interventions to address.

Tying up nets or storing them in a safe space while not in use during the day prevents

physical damage. This practice may also increase the amount of time a net appears clean, which

may lead to less frequent washing. Focus group participants in Senegal reported that tying up

nets during the day is a best practice for preventing damage, and removing the net and storing it

in a safe space is also a good practice particularly when children are likely to pull on hanging

nets [19]. A ‘trials of improved practices’ study in the Peruvian Amazon region found that some

households that did not remove their nets during the day developed ways of tying the nets and

covering them with plastic so they could remain hanging during the day and still be protected

from damage [31]. In our study, variables associated with leaving a net hanging rather than tied

up during the day were varied, and span attitudes and environmental factors in both the highlands

and lowlands. This suggests that there is no one factor that influences net storage behavior, but a

combination of various factors do. Harvey and others reported 15 different reasons a household

122

left a net hanging, the most common of which included lack of time, someone rests under the net during the day, feeling lazy, or forgetting to remove the net [31], indicating that it is inconvenient and time consuming to remove nets during the day. Tying up nets during the day is less labor and time intensive, and prevents damage compared to nets left hanging over sleeping spaces during the day.

Frequent washing can degrade both the physical and chemical integrity of LLINs, so current recommendations are to wash LLINs at most every 3 months. Many nets are washed more frequently than the current recommendations. In the lowlands nearly 80% of nets were washed < 3 months prior to the survey. Other studies have found variable washing frequencies spanning from 1.5 washes per year on average in Uganda [32] to up to 8 washes per month in

Mali [33]. Additionally, 31% of nets were washed more frequently than the recommended practice in a Kenyan study conducted among the southern coastal districts [34]. These differing wash frequencies across studies make generalizability difficult, however the reasons for doing so seem to be less related to knowledge of preferred practices and more related to social and structural factors. Nets simply get dirty quickly, and households (due to personal or social acceptability reasons) prefer to wash nets when they are dirty [18,21,22,31,33,35]. Similarly in our study, washing frequency was associated with environmental factors, and not with knowledge and attitudes like net care knowledge, perception of malaria risk, or confidence in ability to prevent malaria. Although there is great variation across studies, it is important to monitor the practice and understand why households perform variable washing practices due to the significant implications for net longevity.

Soap type and net brand may be modifying factors with washing frequency depending on the pH level of the soap. Bar soaps and mild detergents cause less damage to insecticide levels

123

on nets compared to bleach. In this study, no nets were reported to be washed with bleach, though we were unable to determine brands or pH levels of soaps and detergents used. A washing trial measuring insecticide residual after set time intervals of 6 or 12 months following net distribution in Uganda found that nearly all (99.6%) of nets in the region were washed with mild soap (pH 9-10) and the remaining nets washed with detergent [36] while we found a greater proportion of nets were washed with detergent (21.2% of nets in the highlands and 53.5% of nets in the lowlands). With respect to adherent soap type practices, the Ugandan study found that insecticide integrity differed by net brand and net generation within net brands [36]. Washing nets with more stringent detergents rather than mild soap may impact insecticide integrity. A study in western Kenya tested insecticide effectiveness by measuring how many mosquitos were able to feed through nets and how many mosquitos died after net exposure after 5, 10, and 15 washes with a detergent commonly used in rural areas [37]. Even after only 5 washes with detergent, up to 17.6% of mosquitos were able to feed through nets and up to 83.3% of mosquitos remained alive 24 hours after net exposure depending on net brand [37]. Though this study did not compare mosquito ability to feed or mosquito mortality after exposure to nets washed without detergent, this suggests that while net brand also plays a role, insecticide integrity is noticeably compromised even after 5 washes with detergent [37]. For consistent messaging, recommendations should continue to emphasize the use of no or very mild soaps for net washing, regardless of net brand.

Drying nets in the sun is not recommended because direct sunlight degrades insecticides

[4]. As many nets are dried in the sun in the lowlands, proper net drying is a key net care component to be addressed. Another study in Kenya found that 64.9% of nets were dried in the sun [38], and nets were observed drying in direct sunlight in Uganda as well [39]. Our results

124

suggest that knowledge and attitudes, particularly a lack of knowledge of appropriate net drying influence this particular care practice. Learning net care from a community event was associated with better net drying adherence, suggesting education campaigns geared toward community settings may be effective in improving net drying practices in the lowlands sites.

Implications for interventions

Net washing frequency, use of soap, and drying location affect net longevity

[22,32,39,40,41,42,43]. Other studies have found that LLINs do not last as long as intended in the field, often in poor condition after 1-2 years of use [10,14]. While there is variation among studies in net longevity following distribution campaigns [44], many nets are lost to attrition or are damaged before the universal 3 year life span [4].

Because LLINs involve multiple care and repair practices, and each practice is influenced by multidimensional factors spanning attitudes and environmental factors, intervention strategies should be variable and informed by context- and location-specific care and repair practices [45].

Because of the varied and at times non-existent associations between knowledge of net care items and adherence to care practices, education interventions alone are unlikely to markedly improve bed net care and repair. Social factors are likely the most difficult to change as they are related to cultural and personal belief systems. Thus, successful solutions aimed at improving care and repair should particularly involve environmental interventions. Some possibilities to explore include providing materials to assist with tying nets up during the day if household structures do not easily permit net folding or tying. A recommendation by Harvey and others is to design nets to be easily pulled up or so they can remain hanging, yet be more protected from damage [31]. Because we also found that many nets are left hanging during the day, our results support that this suggestion could prevent physical degradation without necessitating behavior

125

change. Furthermore, consistent and complete care and repair messaging provided from any place where a net is distributed (particularly clinics and distribution events) is recommended.

Demonstrations of proper care and repair activities may be particularly effective.

Limitations: Net care and repair practices are likely quite different between communities, regions, and countries, making comparisons and generalizability between sites and other studies difficult. This highlights the importance of LLIN monitoring after distributions in different locations to track and respond to local trends of net ownership, use, and maintenance. The

President’s Malaria Initiative (PMI) hosts a LLIN durability monitoring repository to track LLIN field durability among different locations [5] and is an excellent resource for comparing results across studies.

“Net condition” was classified based on standard images provided in the survey forms

(See Additional File 1), but this was not a standard method of measuring net condition. Using a proportional hole index would have been a standard and more comparable method of collecting this information, but was unavailable for these analyses. Additionally, we did not measure net repair practices, so this could not be included in the net care adherence score.

To describe net storage during the day, we categorized nets that were removed during the day as a moderately adherent practice. However, we do not know where the nets were placed during the day, so there is potential for misclassification if these nets are removed to a space that makes them prone to damage rather than a safe storage space.

It is possible that the person reporting for the household may not be the same person who cares for the nets, which could lead to misclassification. However, there were options for “don’t know” in survey items, hopefully limiting this occurrence. Additionally, social acceptability bias is possible as individuals might report better care practices than are actually performed,

126

particularly if they have high knowledge of net care. Due to lack of data, we approximated net repair behaviors by measuring the number of holes in each net, as nets with holes had not been repaired. Using holes as a proxy could be related to overall net condition and distort the observed association. Another limitation is the inability to distinguish between activities conducted at community events and net distributions. There may be misclassification if some individuals consider a distribution to be a community event. This is unlikely to change the results, however, as in the lowlands almost no one reported receiving education during a distribution and both community events and distributions in the highlands were associated with better adherence.

Conclusions

We found high adherence to recommended net care practices in western Kenya. Net care and repair is influenced by various attitudes and environmental factors. There is little evidence that behavior change is needed in the highlands sites of western Kenya. In the lowlands, reducing the need for frequent washing by promoting methods for tying up nets during the day, coupled with messaging to emphasize infrequent washing and drying nets in the shade may be an appropriate intervention. If there are concerns about net longevity following a distribution, the need for behavior change interventions should be assessed in local contexts to target specific practices and the driving forces underlying those practices.

Table 3: Distributions of net care adherence and attitudes and environmental factors Highlands CDE Lowlands CDE Median (Range) Median (Range) or # (%) or # (%) p-value A Overall net care adherence score F 2 (0-8) 4 (0-9) <0.001*

Overall net condition <0.001* G Excellent 522 (59.9) 594 (67.2) Good 227 (26.1) 162 (18.3) Fair 98 (11.3) 85 (9.6) Poor 18 (2.1) 23 (2.6)

127

Knowledge and Attitudes What kinds of things should you do to take care of your bed net? C Tie up in the morning 177 (27.6) 94 (16.4) <0.001* Don’t wash or wash rarely 25 (3.9) 44 (7.7) 0.005* Don’t use detergent 135 (21.1) 37 (6.5) <0.001* Don’t beat on the rocks 149 (23.2) 48 (8.4) <0.001* Don’t hang in the sun 382 (59.6) 281 (49.0) <0.001* Re-treat with insecticide 213 (33.2) 271 (47.2) <0.001* Tie up all holes 166 (25.9) 418 (72.8) <0.001*

Perception of malaria risk C (2=low, 10=high) 6 (2, 10) 7.5 (2, 10) <0.001* B

Confidence in ability to prevent malaria C ( Likert scale) 2 (1, 5) 3 (1, 5) <0.001* Environmental Factors Household Family Structure: Net User (years) D Child < 5 219 (25.0) 232 (26.2) 0.386 Child 6-18 332 (38.0) 312 (35.3) 0.387 Adult Male 19-50 169 (19.3) 218 (24.7) 0.003* Adult Female 19-50 332 (38.0) 337 (38.1) 0.386 Adult Male > 50 61 (7.0) 58 (6.6) 0.811 Adult Female > 50 113 (12.9) 101 (11.4) 0.416

Ability to pay for a net C 441 (68.8) 494 (87.1) <0.001*

Source of net care knowledge C Clinic 328 (37.5) 372 (42.1) 0.127 Community event 176 (20.1) 340 (38.5) <0.001* Net distribution 179 (20.5) 3 (0.34) <0.001*

Universal Access (1 net:2 members) 239 (37.2) 414 (72.1) <0.001*

Amount of shade in house compound C,G <0.001* No shade 60 (9.3) 106 (18.5) Little shade 321 (49.9) 255 (44.4) Half shaded 229 (35.6) 159 (27.7) Very shaded 27 (4.2) 38 (6.6)

Sleeping structure has Earth floor E <0.001* Earth floors 916 (73.5) 893 (87.0) No earth floors 331 (26.5) 134 (13.0)

128

Animals present in sleeping structure at night E 0.004* Animals present 164 (13.2) 179 (17.4) No animals present 1083 (86.8) 848 (82.6)

A Two sample test of proportions B Nonparametric equality of medians test C Household level data (Highlands n=641, Lowlands n=574) D Bed net level data (Highlands n=874, Lowlands n=884) E Sleeping structure level data (Highlands n=1247, Lowlands n=1027) *Statistically significant at alpha 0.05; F Lower scores indicate better net care and repair adherence practices; G Pearson chi-square test

List of Abbreviations

ITNs: Insecticide treated nets

LLINs: Long lasting insecticide treated nets

PMI: President’s Malaria Initiative

KEMRI: Kenya Medical Research Institute

GEEs: Generalized estimating equations

Declarations

Ethics approval and consent to participate:

The current analysis involved secondary analyses of the de-identified data from a primary cross- sectional study. The primary study was approved by the Kenyan Medical Research Institute

(KEMRI) (SSC #2810) and the University of Arizona ethical review boards with deferral of primary oversight from the University of Arizona to KEMRI. Informed consent was obtained from all household heads. Parental consent was obtained for children less than 7 years of age, and assent was obtained for those 8 to 17 years of age.

Consent for publication:

Not Applicable

129

Availability of data and material:

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

The present analyses were prepared with no funding sources. The primary study was funded via

NIH/NIAID R15 Al100118-01.

Authors’ contributions

ES analyzed and interpreted the data, and drafted the manuscript. JE, EA, NS, MA, and KE assisted with interpreting the results. EJ, YK, LH, and KE offered scientific guidance in preparation of the analyses, and assisted in the manuscript preparation. SM and MA reviewed the results and conclusions and offered contextual support. MA and JE performed data cleaning. All authors read and approved the final manuscript.

Acknowledgements

We thank Lindsay Hansen for her writing guidance and editing contributions to this manuscript.

References

1. Lengeler C. Insecticide-treated bed nets and curtains for preventing malaria. Cochrane Database Syst Rev. 2009; doi:10.1002/14651858.CD000363.pub2.

2. World Health Organization. Global technical strategy for malaria 2016-2030. Global Malaria Programme. 2015; World Health Organization. http://doi.org/ISBN: 978 92 4 156499 1. Accessed 15 Sep 2018.

3. World Health Organization. Continuous distribution of long-lasting insecticidal nets in Africa through antenatal and immunization services : a joint statement by the Roll Back Malaria Working Groups on Malaria in Pregnancy and Vector Control and the Alliance for Malaria Prevention. 2015.

4. World Health Organization. Guidelines for laboratory and field-testing of long-lasting insecticidal nets. 2015; ISBN 978 92 4 1505277.

130

5. President’s Malaria Initiative. LLIN durability monitoring: toolkit and data repository for PMI-funded LLIN durability monitoring studies. 2018; https://www.durabilitymonitoring.org Accessed 6 Dec 2018.

6. President’s Malaria Intitiative. 2013; http://www.pmi.gov Accessed 6 Dec 2018.

7. Randriamaherijaona S Raharinjatovo J Boyer S. Durability monitoring of long-lasting insecticidal (mosquito) nets (LLINs) in Madagascar: physical integrity and insecticidal activity Parasites & Vectors 2017; 10:564

8. Kilian A, Koenker H, Obi E, Selby RA, Fotheringham M, Lynch M. Field durability of the same type of long-lasting insecticidal net varies between regions in Nigeria due to differences in household behaviour and living conditions. Mal J. 2015; doi:10.1186/s12936-015-0640-4

9. Helinski MH, Namara G, Koenker H, Kilian A, Hunter G, Acosta A., et al. Impact of a behaviour change communication programme on net durability in eastern Uganda. Mal J. 2015; doi:10.1186/s12936-015-0899-5

10. Gnanguenon V, Azondekon R, Oke-Agbo F, Beach R, Akogbeto M. Durability assessment results suggest a serviceable life of two, rather than three, years for the current long-lasting insecticidal (mosquito) net (LLIN) intervention in Benin. BMC Infect Dis. 2014; doi:10.1186/1471-2334-14-69

11. Azondekon R‚ Gnanguenon V‚ Oke-Agbo F‚ Houevoessa S‚ Green M‚ Akogbeto M. A tracking tool for long-lasting insecticidal (mosquito) net intervention following a 2011 national distribution in Benin. Parasit Vector. 2014;7:6

12. Morgan J‚ Abílio AP‚ do Rosario Pondja M‚ Marrenjo D‚ Luciano J‚ Fernandes G; et al. Physical durability of two types of long-lasting insecticidal nets (LLINs) three years after mass LLIN distribution campaign in Mozambique‚ 2008-2011. Am J Trop Med Hyg. 2015;92:286-293

13. Hakizimana E‚ Cyubahiro B‚ Rukundo A‚ Kabayiza A‚ Mutabazi A‚ Beach R‚ et al. Monitoring long- lasting insecticidal net (LLIN) durability to validate net serviceable life assumptions‚ in Rwanda. Malar J. 2014;13:344

14. Craig AS, Muleba M, Smith SC, Katebe-Sakala C, Chongwe G, Hamainza B, et al. Long-lasting insecticidal nets in Zambia: A cross-sectional analysis of net integrity and insecticide content. Mal J. 2015; doi:10.1186/s12936-015-0754-8

15. Tan KR Coleman J Smith B Hamainza B Katebe-Sakala C Kean C Kowal A Vanden Eng J Parris TK Mapp CT Smith SC Wirtz R Kamuliwo M Craig AS A longitudinal study of the durability of long-lasting insecticidal nets in Zambia. Malar J. 2016 Feb 19;15:106

16. President’s Malaria Initiative. ITN care and repair: improving net lifespan through behavior change communication. Networks Summary Series. 2015; https://www.k4health.org/sites/default/files/networks_summary_net_care_and_repair_2015-jan_0.pdf Accessed 5 Dec 2018.

17. K4Health. Care of mosquito nets toolkit. 2016; https://www.k4health.org/toolkits/care-repair-LLIN Accessed 6 Dec 2018.

131

18. Shirayama Y, Phompida S, Kuroiwa C, Miyoshi M, Okumura J, Kobayashi J. Maintenance behaviour and long-lasting insecticide-treated nets (LLITNs) previously introduced into Bourapar district, Khammouane province, Lao PDR. Pub Health. 2007; doi:10.1016/j.puhe.2006.06.016

19. Loll DK, Berthe S, Faye SL, Wone I, Arnold B, Koenker H., et al. “You need to take care of it like you take care of your soul”: perceptions and behaviours related to mosquito net damage, care, and repair in Senegal. Mal J. 2014; doi:10.1186/1475-2875-13-322

20. Hunter GC, Scandurra L, Acosta A, Koenker H, Obi E, Weber R. We are supposed to take care of it: A qualitative examination of care and repair behaviour of long-lasting, insecticide-treated nets in Nasarawa State, Nigeria. Mal J. 2014; doi:10.1186/1475-2875-13-320

21. Miller JE, Jones COH, Ndunguru S, Curtis V, Lines J. A new strategy for treating nets. Part 2: users’ perceptions of efficacy and washing practices and their implications for insecticide dosage. Trop Med Int Health. 1999;4:167–174.

22. Janz NK, Becker MH. The health belief model: a decade later. Health Education Quarterly 1984; 11(1) 1-47

23. Mutuku FM, Khambira M, Bisanzio D, Mungai P, Mwanzo I, Muchiri EME., et al. (2013). Physical condition and maintenance of mosquito bed nets in Kwale County, coastal Kenya. Mal J. 2013; doi:10.1186/1475-2875-12-46

24. Kohler LN, Hibler EA, Harris RB, Oren E, Roe DJ, Jurutka PW, Jacobs ET. Greater adherence to cancer prevention guidelines is associated with higher circulating concentrations of vitamin D metabolites in a cross-sectional analysis of pooled participants from 2 chemoprevention trials. J Nutr. 2017; doi:10.3945/jn.116.243352

25. McCullough ML, Patel AV, Kushi LH, Patel R, Willett WC, Doyle C, et al. Following cancer prevention guidelines reduces risk of cancer, cardiovascular disease, and all-cause mortality. Cancer Epid Biomarkers Prev. 2011; doi:10.1158/1055-9965.EPI-10-1173

26. Glanz K, Rimer BK, Lewis FM. Health behavior and health education: theory, research, and practice: 3rd Ed. 2002.

27. Homish GG, Edwards EP, Eiden RD, Leonard KE. Analyzing family data: a GEE approach for substance use researchers. Addict Behav. 2010; doi: 10.1016/j.addbeh.2010.01.002

28. Liang KY, Zeger SL. Regression analysis for correlated data. Annu Rev Publ Health. 1993; doi:10.1146/annurev.pu.14.050193.000355

29. Lumley T. Generalized estimating equations for ordinal data : a note on working correlation structures. Biometrics. 1996;52:354-361.

30. Dillip A, Mboma ZM, Greer G, Lorenz LM. Understanding roles of men and women in net care and repair in southern Tanzania. Mal J. 2018; doi:10.1186/s12936-018-2608-7

31. Harvey SA, Olórtegui MP, Leontsini E, Asayag CR, Scott K, Winch PJ. Trials of improved practices (TIPs): A strategy for making long-lasting nets last longer? Am J Trop Med Hyg. 2013; doi:10.4269/ajtmh.12-0641

132

32. Kilian A, Byamukama W, Pigeon O, Gimnig J, Atieli F, Koekemoer L, et al. Evidence for a useful life of more than three years for a polyester-based long-lasting insecticidal mosquito net in western Uganda. Mal J. 2011; doi:10.1186/1475-2875-10-299

33. Leonard L, Diop S, Doumbia S, Sadou A, Mihigo J, Koenker H, et al. Net use, care and repair practices following a universal distribution campaign in Mali. Mal J. 2014; doi:10.1186/1475-2875-13- 435

34. Khambira, M. Insecticide treated bednets ownership, use and maintenance behaviour in Kwale, Msambweni and Kinango districts in Kenya. http://ir- library.ku.ac.ke/bitstream/handle/123456789/7046/Maureen Khambira.pdf;sequence=1. 2013; Accessed 6 Dec 2018.

35. Erlanger T. Field issues related to insecticide-treated nets in Tanzania. MSc Thesis. University of Basel and Swiss Tropical Institute. 2002; 153–160.

36. Kilian A, Byamukama W, Pigeon O, Atieli F, Duchon S, Phan C. Long-term field performance of a polyester-based long-lasting insecticidal mosquito net in rural Uganda. Mal J. 2008; doi:10.1186/1475- 2875-7-49

37. Atieli FK, Munga SO, Ofulla AV, Vulule JM. The effect of repeated washing of long-lasting insecticide-treated nets (LLINs) on the feeding success and survival rates of Anopheles gambiae. Mal J. 2010; doi:10.1186/1475-2875-9-304

38. Mejía P, Teklehaimanot HD, Tesfaye Y, Teklehaimanot A. Physical condition of Olyset® nets after five years of utilization in rural western Kenya. Mal J. 2013; doi:10.1186/1475-2875-12-158

39. Scandurra L, Acosta A, Koenker H, Kibuuka D, Harvey S. “It is about how the net looks”: a qualitative study of perceptions and practices related to mosquito net care and repair in two districts of eastern Uganda. Mal J. 2014;13:504.

40. Allan R, O’Reilly L, Gilbos V, Kilian A. An observational study of material durability of three world health organization-recommended long-lasting insecticidal nets in Eastern Chad. Am J Trop Med Hyg. 2012; doi:10.4269/ajtmh.2012.11-0331

41. Smith SC, Joshi UB, Grabowsky M, Selanikio J, Nobiya T, Aapore T. (2007). Evaluation of bednets after 38 months of household use in northwest Ghana. Am J Trop Med Hyg. 2007;77(6):243-248.

42. Norris LC, Norris DE. Efficacy of long-lasting insecticidal nets in use in Macha, Zambia, against the local Anopheles arabiensis population. Mal J. 2011; doi:10.1186/1475-2875-10-254

43. Atiele H, Zouh G, Yaw A, Mwanzo I, Githeko A, Yan G. Insecticide-treated net (ITN) ownership, usage, and malaria transmission in the highlands of western Kenya. Parasit Vectors. 2011; doi:10.1186

44. Wills AB, Smith SC, Anshebo GY, Graves PM, Endeshaw T, Shargie EB, et al. Physical durability of permanet 2.0 long lasting insecticidal nets over three to 32 months of use in Ethiopia. Mal J. 2013; doi:10.1186/1475-2875-12-242

133

45. Briët OJT, Penny MA. Repeated mass distributions and continuous distribution of long-lasting insecticidal nets: modelling sustainability of health benefits from mosquito nets, depending on case management. Mal J. 2013; doi:10.1186/1475-2875-12-401

Figure titles and legends

Figure 1. Attitudes and environmental factors related to adherence to net care and repair behaviors

Adapted from the NetWorks conceptual framework [16].

Table 1. Elements of the bed net care adherence scoring system.

Scores of 0 indicate best practice, scores of 1 indicate moderately adherent practice, and scores of 2 indicate non- adherent practice.

Table 2: Household characteristics of study sample by site

Table 3: Distributions of net care adherence and attitudes and environmental factors

A Two sample test of proportions B Nonparametric equality of medians test C Household level data (Highlands n=641, Lowlands n=574) D Bed net level data (Highlands n=874, Lowlands n=884) E Sleeping structure level data (Highlands n=1247, Lowlands n=1027) *Statistically significant at alpha 0.05; F Lower scores indicate better net care and repair adherence practices; G Pearson chi-square test

Table 4: GEE results of association between net condition and overall net care adherence score

Adjusted for net age. Associations measured using GEE models with net care adherence as an ordinal variable. Adjusted for net age.

Table 5: Univariate GEE models measuring associations between attitudes and environmental factors with binary net care practices

Odds ratios (95% Confidence Intervals) A Adjusted for net age; Blue color indicates associations with better adherence, orange color indicates associations with worse adherence, White indicates no association

134

APPENDIX C: ADDITIONAL FILE 1

135

APPENDIX D: MANUSCRIPT 2

“After those nets are torn most people use them for other purposes” : An examination of bed net repurposing and misuse in western Kenya

Ellen M Santos* 1 [email protected] Corresponding Author

Jenna E Coalson 1 [email protected]

Elizabeth T Jacobs 1 [email protected]

Yann C Klimentidis 1 [email protected]

Stephen Munga 2 [email protected]

Maurice Agawo 2 [email protected]

Kacey C Ernst 1 [email protected]

1 Mel and Enid Zuckerman College of Public Health, University of Arizona, 1295 N Martin Ave, Tucson, AZ 85724

2 Centre for Global Health Research, Kenya Medical Research Institute, PO Box 1578, 40100 Kisumu, Nyanza, Kenya

Abstract

Background: The consequences of using bed nets for purposes other than sleeping protection from mosquito bites on malaria prevention efforts are poorly understood. Often termed misuse, there is controversy about how often intact bed nets are used for alternative purposes as compared to how often older worn-out nets are repurposed. Methods: We reported three measures of repurposing among two sites (highlands and lowlands) from a cross-sectional study

136

of 1,217 households in western Kenya: direct observations around the home, self-report of alternative uses of nets inside the home, and participant reporting of community-level repurposing. We measured bed net misuse. We then assessed the potential impact of repurposing on malaria prevention efforts. We hypothesized that compared to households without repurposed nets, households with repurposed nets (1) have higher net ownership, (2) have a greater percent of members sleeping under nets, (3) report fewer malaria epidsodes in the last year, and (4) report fewer fevers in the last two weeks. Results: Of 643 households in the highlands, we observed 52 (8.1%) with repurposed nets, while 184 (33.0%) of the 574 lowlands households were observed to have repurposed nets. Among the observed repurposed nets, 77.6% of in the highlands and 69.8% in the lowlands were repurposed for use outdoors, commonly serving as chicken coops, fences, and tree covers. Repurposed nets found indoors served functions including curtains, covering bathrooms, and cut up and used as washing sponges. In both sites, the vast majority of nets were repurposed because households determined them to be either torn, old and worn out, or both torn and old. We found no impact of net repurposing on malaria prevention efforts. In both sites, there was no association between observed net repurposing and household (1) net ownership, (2) net use, (3) malaria in the last year, or (4) fever in the last two weeks. Net misuse was rare. Of 379 repurposed nets, 4 (1.06%) were in good condition with no holes. Of 1,758 functional nets inside homes, 13 (0.74%) were reported to have ever been used for an alternative purpose. Conclusions: Though bed net repurposing is common, particularly in the lowlands, repurposing may not be detrimental to malaria prevention efforts. Future studies should use consistent definitions of bed net repurposing and misuse to better understand the practice of using bed nets for alternative purposes.

137

Introduction

The distribution of insecticide treated bed nets (ITNs) and long lasting insecticidal nets

(LLINs) is the primary malaria prevention strategy in sub-Saharan Africa. The use of ITNs or

LLINs (from here, referred to as bed nets or nets) for purposes other than protecting individuals from malaria has been reported, typically described as misuse, but not rigorously investigated across multiple contexts. Observations of ‘misuse’ have been generalized to larger populations and taken out of context by the media, [1,2] and further amplified in the malaria literature [3].

There is a need to understand and define bed net misuse in contrast with repurposing.

Different studies report various accounts of bed net misuse. Eisele and others assert that very little evidence exists to support claims of widespread misuse [4]. In Zambia, prevalence of misuse of nets for fishing was only 3% in the Luangwa District [4], an area reliant on fishing. In

Sierra Leone, only 5.3% of households reported using nets for anything other than protection against mosquitos [5]. Using predictive modeling, Honjo and others explain that misuse is more likely to occur in contexts of extreme poverty where alternatively using nets is more helpful for the livelihood of the household [6]. Baume and others found misuse to be uncommon and concentrated in a small minority of communities in Ethiopia [7].

It is important to understand what does not constitute misuse before defining true bed net misuse. Repurposing inactive nets (nets no longer useful for sleeping under) is not misuse [8].

Repurposing can be beneficial (nets still act as a barrier against mosquitos in some way) or neutral (use of nets for other household purposes). True misuse is defined as the alternative use of an intact bed net, or the use of a net that causes environmental harm as in instances where the insecticides from nets damage ecosystems during fishing [8]. Misuse is typically measured from self-report survey items such as “is this bed net ever used for purposes other than preventing

138

mosquito bites?” This is problematic for two reasons; it is not always clear if these nets are in serviceable condition to properly classify as misuse or repurposing, and self-report likely introduces misclassification due to social-acceptability bias [9].

It is important to properly distinguish between net misuse and repurposing to understand the consequences of each phenomenon for malaria prevention efforts. In particular, it is unclear whether bed net repurposing impacts malaria prevention efforts. There is evidence to suggest that bed nets are repurposed when households deem them to be old and/or torn [10]. We therefore seek to better understand the context of net repurposing and misuse, and to measure whether households that have repurposed nets also use nets for mosquito prevention and have lower occurrence of malaria compared to households with no repurposed nets.

We hypothesize that LLIN repurposing is more likely a reflection of re-use of old nets rather than misuse of nets. We report three measures of repurposing: direct observations around the home, self-report of alternative uses of nets inside the home, and participant reporting of community-level repurposing. Qualitatively, we describe observed uses of repurposed nets and self-reported reasons for net repurposing. To quantitatively measure the impact of repurposing on malaria prevention efforts, we hypothesize that compared to households without repurposed nets, (1) more households with repurposed nets meet the WHO recommended net ownership threshold of 1 net: 2 household members, (2) the percent of household members who slept under a net is greater among households with repurposed nets, (3) there are fewer reports of malaria diagnosis within the last year of an individual within a household and (4) there are fewer reports of fever in the last two weeks of an individual within a household. We use data from a cross- sectional study administered in two regions with differing malaria transmission in western Kenya

139

to measure these hypotheses. Details of the parent cross-sectional study were previously described [11].

Methods

Definitions

Household surveys were conducted in 1,217 households among three sublocations of the highlands (Chepsonoi, Kiborgok, and Tindinyo) and among two sublocations of the lowlands

(Kabar West and Kabar Central) from March to December, 2015. Analyses were conducted separately for the highlands and lowlands because the two sites differ in terms of malaria transmission intensity, which can impact individual malara prevention behaviors. Because the majority of surveys were conducted during the spring and summer months, households surveyed from March to May are classified as early surveys, those from June to July are mid surveys, and those from August to December are late surveys. Households where the study team observed any bed nets being used for purposed other than covering sleeping areas were determined to have net repurposing.

Household heads were asked about various malaria perceptions including how serious a problem malaria is for their family (Likert scale: not at all serious to extremely serious) and the likelihood a household member is to contract malaria within the next month (Likert scale: not at all likely to extremely likely). To get a sense of socioeconomic status, households were categorized into wealth quartiles based on an inverse frequency of ownership-weighted asset- based wealth index. Additionally, households were asked whether they would be able to afford a bed net if one was not provided for free, measured as a binary Yes/No response.

Households that met the WHO-recommended net ownership threshold included those with at least one (non-repurposed) net for every two household members. The percent of

140

household members who slept under a net was calculated using the number of household members who reported to have slept under a bed net the night prior to the household survey.

Malaria diagnosis within the last year or fever within the last two weeks is defined as at least one household member who was reported to have been diagnosed with malaria within one year or at least one household member with self-reported fever within two weeks prior to the household survey, respectively. Malaria diagnosis and fever were self-reported or reported by the household head regarding other household members. We indicated binary household ownership of animals including cattle, sheep, goats, chickens, and dogs. We also measured whether households reported observing other households in their community repurpose bed nets with a binary

Yes/No variable.

Statistical Analyses

We used the Pearson Chi-Square test to determine whether the distribution of various household characteristics and malaria perceptions differed between households with and without repurposed bed nets in the highlands and lowlands. To test the four stated hypotheses and to measure the magnitude of associations of structural and social factors with household bed net repurposing, we used univariate logistic regression models. Logistic regression assumptions were tested where appropriate, including independent observations and linearity in the log-odds.

Because a household’s local community likely influences both the selected independent variables and bed net repurposing, we adjusted for sublocation. We used Stata v12 (College Station, TX) and α=0.05 for all analyses.

Results

Household characteristics and malaria perceptions

141

There were no statistically significant differences in the distributions of any household characteristics between households with and without repurposed nets in both the highlands and lowlands (Table 1). Most households in the highlands have 4-6 members, 1-2 sleeping structures,

≤ 2 bed nets, a female household head that has some or completed primary education, and fall into the second wealth quartile. Most households in the lowlands have 1-3 members, 1 sleeping structure, 1-2 bed nets, a female household head that has some or completed primary education, and fall into the fourth wealth quartile. Additionally, 69.1% of households in the highlands and

87.1% in the lowlands reported the ability to afford a bed net if one was not provided for free

(Table 1).

Most households reported that malaria is a slightly or somewhat serious problem for their family in the highlands, and a slightly-to-serious problem in the lowlands, with no significant differences between households with and without repurposed bed nets. In the highlands and lowlands, most (91.9% and 86.6%, respectively) households report having no problems in hanging bed nets or other difficulties with net use (Table 1). A higher proportion of repurposed nets were observed earlier in the study period with a decreasing trend during the mid and late study period in both sites, though the distribution of households in each study period with and without repurposed nets was not significantly different in either site.

Observation of repurposed bed nets

Fewer households repurposed nets in the highlands compared to the lowlands. Of 643 households in the highlands, we observed 52 (8.1%) with repurposed nets, while 184 (33.0%) of the 574 lowlands households had repurposed nets observed by the study team. Some households repurposed more than one net, indicating clustering of net repurposing within households, as there were 67 repurposed nets among the 52 highlands households and 312 repurposed nets

142

among the 184 lowlands households that had repurposed nets. Clustering of net repurposing may also occur at the community level, so we described clustering of net repurposing observations within sublocations (Table 1). In the highlands, Chepsonoi had the highest prevalence of observed repurposing and Kiborgok had the lowest prevalence (10.0% and 5.9% of households respectively). The distribution of repurposed nets between the highland sublocations was significantly different. In the lowlands, the prevalence of repurposing was similar between Kabar

Central and Kabar West (33.5% and 30.2%, respectively) (Table 1).

Most nets in the highlands were repurposed for use outdoors (77.6%) (Figure 1A). The most common functions of observed repurposed nets were as chicken coops or fences (33% and

37%, respectively). Nets repurposed for use indoors were used as curtains or to cover the ceiling or bathrooms (Figure 1A). Similarly in the lowlands, most nets (69.8%) were repurposed for use outdoors, primarily to cover trees (22%), as chicken coops (20%), or as fences (25%) (Figure

1A). Nets repurposed for use indoors covered bathrooms or were cut up and used as washing sponges. Of the repurposed nets, 7 (10.4%) were classified as beneficial repurposing and 60

(89.6%) were classified as neutral repurposing in the highlands. In the lowlands, 28 repurposed nets (9.0%) were repurposed beneficially and 284 nets (91.0%) were neutral.

Twenty-six of highlands households and 127 of lowlands households with repurposed nets reported why they repurposed nets. In both sites, the vast majority of nets were repurposed because households determined them to be either torn, old and worn out, or both torn and old

(Table 2).

Table 2: Household-reported reasons for repurposing bed nets Highlands Lowlands Representative Quote n (%) N=26 n (%) N=127 households households Net was torn/too many “The [net] had many holes that could not holes 10 (38.5) 23 (18.1) be sewn up” Net was old and worn out 5 (19.2) 78 (61.4) “[The net] had lasted a long period of

143

time...more than 5 years therefore [we] decided to use it as a garden fence” Net was old and had holes 7 (26.9) 14 (11.0) “[The net] had many holes and [was] old” Damage by outside “Bed net was eaten by a rat” factors 1 (3.8) 3 (2.4) “[The net was used] to prevent hens from Needed to protect crops 2 (7.7) 3 (2.4) attacking the vegetables” Did not want to throw “[It is] better that way instead of throwing away 1 (3.8) 2 (1.6) away after it was torn” “After those nets are torn most people use Other 0 4 (3.1) them for other purposes”

Reports of net repurposing in the community

Though more prevalent in the lowlands than the highlands (94.3% vs. 70.6% of households, respectively), it was very common for households to report seeing other households in their community repurpose bed nets in both sites (Table 1). Similarly to the observed net repurposing, more households reported noticing community members repurposing bed nets in

Chepsonoi (86.0% of households) compared to Kiborgok (49.0% of households). In the lowlands, more households reported observing net repurposing in their community in Kabar

Central (97.8% of households) compared to Kabar West (88.8% of households). Households in the study population reported observing similar uses for repurposed nets that were observed by the study team, with uses in chicken coops, as fences, and as washing sponges among the most common (Figure 1B). However, households reported that others also use nets for fishing, as decoration, as ropes, and as sieves, while these particular uses were not observed by the study team among the participant households (Figure 1B).

Figure 1: Various uses of repurposed bed nets as observed by the study team (1A) and as observed by study participants (1B) in the highlands and lowlands of western Kenya *Images from publicdomainvectors.org

144

Bed net misuse

Repurposed nets that are fully intact (could effectively be used for preventing mosquito bites) are considered to be misused nets. Though we were unable to determine the insecticidal efficacy of repurposed nets to know if they were still efficacious in repelling and killing mosquitoes, 1 of the 67 repurposed nets in the highlands and 3 of the 312 repurposed nets in the lowlands had no holes, suggesting only 1.8% and 1.1% of repurposed nets in the highlands and lowlands respectively were misused (Figure 2). Although we are unable to determine whether holes appeared before or after repurposing, the vast majority of repurposed nets in both sites had many holes (Figure 2).

Bed net misuse can also occur among nets inside the home that are generally used for sleeping. Among 874 functional nets inside homes in the highlands, only 7 nets (0.8%) were self- reported to have ever been used for something other than mosquito protection while sleeping.

Similarly, this was reported for only 6 nets (0.7%) of 884 functional nets in the lowlands.

Figure 2: Observed number of holes among repurposed bed nets

Impact of bed net repurposing on malaria prevention efforts

We tested 4 hypotheses to assess the impact of net repurposing on malaria prevention efforts and malaria outcomes. In both sites, there was no association between observed net repurposing and household (1) net ownership, (2) net use, (3) malaria in the last year, or (4) fever in the last two weeks (Figure 3). There were 27 households in the highlands (4.2%) and 13 households in the lowlands (2.3%) that had repurposed nets and at least one household member who did not use a net. Other factors were associated with observed net repurposing including animal ownership and social factors. Animal ownership was positively associated with repurposing in the lowlands, where households with repurposed nets had 1.53 times the odds of

145

owning goats and 2.74 times the odds of owning chickens compared to households with no repurposed nets (Figure 3). Though not statistically significant, there was a trend of increasing presence of repurposing with decreasing net use in the household in the highlands (Figure 3). As households report lower proportions of members sleeping under nets, they also had more repurposed bed nets. Notably, in both sites, households that repurposed nets had significantly greater odds of reporting that they have observed other community households repurposing nets

(OR 7.27 95% CI 2.22, 23.79 highlands, OR 3.59 95% CI 1.23, 10.51 lowlands) compared to households with no repurposed nets (Figure 3).

Figure 3: Univariate logistic regression results (odds ratios and 95% confidence intervals) assessing factors associated with the presence of bed net repurposing at the household level, adjusted for sublocation

Discussion

Observation of repurposed bed nets

We observed that 8.1% and 33.0% of households had repurposed nets in the highlands and lowlands, respectively. Repurposing bed nets is reportedly a common occurrence throughout

Kenya [12]. Bed nets were repurposed for multiple uses both indoors and outdoors, particularly for fencing, chicken coops, to protect crops, to cover bathrooms or serve as curtains, or cut up as cleaning rags. Of these repurposed nets, 89.6% and 91.0% of nets in the highlands and lowlands, respectively, are classified as neutral as their new purpose does not protect household members from mosquitoes [8]. Many of these neutral nets were repurposed to help store, dry, or otherwise secure food. Generally, once nets are deemed too old or worn to be used for sleeping, they are repurposed for food security purposes. Few nets were beneficially repurposed (still offered protection against mosquitoes). Households that use bed nets for sleeping while also beneficially repurposing old nets as screens or curtains may have better mosquito protection than households that dispose of old nets or repurpose them for uses that do not protect against mosquitos. Kibe

146

and others suggest the development of net repurposing and disposal guidelines to promote beneficial repurposing (curtains, window screens, etc) [12]. Many repurposed nets were used for similar purposes to those reported in other studies. Participants in focus groups in Timor-Leste mentioned repurposing bed nets for protecting vegetables/crops and for use as sieves [13], and others have reported that nets are used for drying fish [14] and various other purposes including as curtains, ceiling covers, and decorations [7,12,15].

Reported reasons for bed net repurposing

The reasons reported for repurposing are nearly ubiquitous in that households deemed the nets too old and/or too damaged to be used for sleeping under anymore. It was apparent from focus groups and interviews in Ethiopia that bed nets are repurposed when they are no longer useful for sleeping due to wear and tear [16]. Focus group participants in Kenya explained that nets are repurposed because they are strong and durable for other uses, and because they do not know a better way of disposing of the nets once they are no longer useful for sleeping under [12].

While repurposing almost always occurs when nets are no longer usable, there is variation among households’ definition of usability [10,16]. Some households may consider nets with only a few holes to be too damaged for use [15,17], while other households repurposed nets that were older than 2 years [18]. In a longitudinal study following LLINs post-distribution in Zambia, the mean net age of discarded or repurposed nets was 18 months [19]. Though we could not determine the temporality of whether holes appeared in repurposed nets before or after they were repurposed, nearly all repurposed nets in both the highlands and lowlands had many holes.

Efforts to promote net repair may help nets continue to be used for mosquito protection longer before they are discarded or repurposed.

147

Reports of net repurposing in the community

Another level of repurposing examined includes the reported observations by study participants of net repurposing frequency among their community members. Viewing repurposing as socially acceptable, or the normalization of repurposing, may influence whether households repurpose nets, as we found households were significantly more likely to have repurposed nets if they reported observing other households in the community repurposing nets.

Net repurposing is also often concentrated in communities [7], thus as the practice becomes normalized in a community, more households are likely to repurpose their bed nets. In Senegal, study participants reported seeing bed nets used for alternative purposes in some communities, but not others [15], emphasizing the clustered nature of net repurposing. Social acceptability is associated with other aspects of bed net use, particularly the desire to keep nets from appearing dirty [20,21], so the more socially acceptable a practice is, the more prevalent it becomes. Rather than throwing old, worn nets away, households observe others finding innovative uses for them, see that the practice is acceptable, and find ways to make their nets useful for their household.

Bed net misuse

The third level of repurposing we assessed was bed net misuse. Using an intact, functional net for a purpose other than mosquito protection while sleeping was very rare. In our study sites, only 1 repurposed net in the highlands and 3 nets in the lowlands had no holes, though it is possible these nets were still considered old and worn by the household.

Additionally, among intact nets inside the home, fewer than 1% of functioning bed nets have ever been alternatively used in both sites. Others also report rare misuse [4] and suggest misuse occurs when alternatively using a bed net is economically beneficial [5,22], particularly where nets are misused for fishing [13,14,23]. We observed no repurposed nets used for fishing, though

148

nearly 50% of households in the lowlands reported seeing others using nets for fishing. There are no large bodies of water near the study sites, and fishing is not an economic activity in the area.

Most areas where bed nets are misused for fishing are heavy fishing communities

[14,18,23,24,25] near large bodies of water including coastal areas, the African Great Lakes, and large rivers [26]. Our study sites, like others where non-fishing activities drive the economy, report that misuse is rare [4,5,7].

Due to the environmental harm caused by fishing with ITNs or LLINs [27,28,29,30], misusing nets for fishing is a serious problem that should be addressed [8]. While misusing nets for fishing is an environmental problem, it is not clear if misuse in general is a problem for malaria prevention efforts including net ownership, use, and malaria outcomes. For example, it is not always clear whether households that use nets for fishing also use nets to sleep under [14].

Impact of net repurposing on malaria prevention efforts

Though common, bed net repurposing only poses a problem to malaria prevention efforts if households that repurpose nets do not meet ownership recommendations or if household members do not use nets. We found no association between household net repurposing and net ownership, use, or malaria outcomes. While there were more households with repurposed nets in the lowlands than in the highlands, more household members also slept under nets in the lowlands, supporting reports that repurposing is unlikely detrimental for household net use for mosquito prevention [4]. Few households had repurposed nets and household members who did not sleep under a net in both sites. Though sleeping under a torn, old net may still offer some protection against mosquitos, most households with repurposed nets also used nets for sleeping.

Even in a study where bed net misuse for fishing was very common, more than 80% of individuals using nets for fishing reported they also used bed nets in the household [14]. As more

149

net distributions occur, households are likely to attain newer nets to replace worn nets that are then repurposed in economically beneficial ways to the household [12]. However, in contrast to the lowlands, we observed a non-significant, though interesting trend suggesting households with fewer individuals sleeping under nets are more likely to have repurposed nets in the highlands than households where nearly everyone slept under a net. Even if households in the lowlands do not own enough nets, several household members crowd under a single net, while this occurs infrequently in the highlands [11]. This suggests there may be differences in net use and repurposing behaviors between areas of year-round and seasonal Plasmodium transmission.

Limitations

While we were able to directly observe repurposed nets, the age or condition of the net when first repurposed is unknown. It is possible that nets were first repurposed when they were still viable for intended use. However, nearly all participants discussed repurposing nets once they were too old or damaged for intended use. Direct measurement of net misuse was difficult.

We defined misuse as repurposed nets that were in physically good condition with no holes or as nets used as intended in the household but reported to have ever been used for an alternative purpose. To obtain more nuanced and accurate measurements of true net misuse, future studies should inquire about the age and physical condition of bed nets when first repurposed.

Conclusions

Bed net misuse is very rare in our study sites, while net repurposing is more common.

Bed nets are repurposed for various outdoor and indoor purposes such as fencing for chicken coops, or cut up into washing sponges. The presence of repurposing was not associated with bed net ownership, household net use, or malaria outcomes. Thus, net repurposing does not appear to be detrimental for malaria prevention efforts. Future assessments of bed net misuse and

150

repurposing should describe each occurrence according to the Roll Back Malaria Consensus

Statement [8] to ensure the same definitions are used for comparability between studies and

regions.

Table 1: Household characteristics and malaria perceptions comparing households with and without repurposed bed nets A Highlands Lowlands Any No Any No Repurposing Repurposing P- Repurposing Repurposing P- Total (n= 52) (n=587) Value A Total (n=184) (n=374) Value A People per household 0.629 0.183 1-3 240 21 (8.8) 219 (91.3) 372 120 (32.3) 252 (67.7) 4-6 293 25 (8.5) 268 (91.5) 164 53 (32.3) 111 (67.7) ≥ 7 106 6 (5.7) 100 (94.3) 22 11 (50.0) 11 (50.0) # Sleeping structures 0.537 0.349 1 450 38 (8.4) 412 (91.6) 501 167 (33.3) 334 (66.7) 2 156 14 (9.0) 142 (91.0) 49 15 (30.6) 34 (69.4) 3 27 0 27 (100) 7 1 (14.3) 6 (85.7) 4 5 0 5 (100) 1 1 (100) 0 5 1 0 1 (100) 0 0 0 Nets per household 0.236 0.470 0 202 11 (5.4) 191 (94.6) 10 2 (20.0) 8 (80.0) 1 167 17 (10.2) 150 (89.8) 309 95 (30.7) 214 (69.3) 2 152 15 (9.9) 137 (90.1) 179 65 (36.3) 114 (63.7) 3 85 7 (8.2) 78 (91.8) 45 16 (35.6) 29 (64.4) 4 28 1 (3.6) 27 (96.4) 14 5 (35.7) 9 (64.3) ≥5 5 1 (20.0) 4 (80.0) 1 1 (100) 0 Education level female household head 0.497 0.076 None 55 5 (9.1) 50 (90.9) 22 10 (45.5) 12 (54.5) Some primary 261 25 (9.6) 236 (90.4) 193 54 (28.0) 139 (72.0) Completed primary 148 10 (6.8) 138 (93.2) 187 77 (41.2) 110 (58.8) Some secondary 66 6 (9.1) 60 (90.9) 54 16 (29.6) 38 (70.4) Completed secondary 30 0 30 (100) 39 10 (25.6) 29 (74.4)

151

Any tertiary 20 3 (15.0) 17 (85.0) 8 1 (12.5) 7 (87.5) No female household head 42 1 (2.4) 41 (97.6) 47 14 (29.8) 33 (70.2) How serious a problem malaria is for the family 0.383 0.938 0.273 Not at all serious 22 4 (18.2) 18 (81.8) 5 2 (40.0) 3 (60.0) Slightly serious 357 26 (7.3) 331 (92.7) 143 43 (30.1) 100 (69.9) Somewhat serious 189 16 (8.5) 173 (91.5) 100 33 (33.0) 67 (67.0) Serious 64 5 (7.8) 59 (92.2) 283 96 (33.9) 187 (66.1) Extremely serious 5 1 (20.0) 4 (80.0) 26 9 (34.6) 17 (65.4) Problems ever using a net 0.184 0.696 0.438 Yes 43 6 (14.0) 37 (86.0) 72 25 (34.7) 47 (65.3) No 491 40 (8.1) 451 (91.9) 466 151 (32.4) 315 (67.6) Likelihood a HH member will get malaria in next month 0.687 0.526 Not at all likely 26 1 (3.8) 25 (96.2) 4 2 (50.0) 2 (50.0) Somewhat likely 66 8 (12.1) 58 (87.9) 59 21 (35.6) 38 (64.4) Equally likely as unlikely 142 11 (7.7) 131 (92.3) 277 88 (31.8) 189 (68.2) Likely 396 31 (7.8) 365 (92.2) 164 50 (30.5) 114 (69.5) Extremely likely 1 0 1 (100) 50 21 (42.0) 29 (58.0) Wealth quartile 0.809 0.234 1st Quartile 160 12 (7.5) 148 (92.5) 118 34 (28.8) 84 (71.2) 2nd 18 (8.5) 193 (91.5) 26 (25.2) 77 (74.8) Quartile 211 103 3rd 8 (6.3) 120 (93.8) 55 (35.3) 101 (64.7) Quartile 128 156 4th Quartile 129 12 (9.3) 117 (90.7) 187 69 (36.9) 118 (63.1) Ability to afford a 0.888 net 0.753

152

Yes 442 36 (8.1) 403 (91.2) 494 157 (31.8) 322 (65.2) No 198 15 (7.6) 182 (91.9) 73 23 (31.5) 49 (67.1) Survey

Timing 0.101 0.071 Early (March- 3 (21.4) 11 (78.6) 3 (75.0) 1 (25.0) May) 14 4 Mid (June- 36 (8.7) 379 (91.3) 125 (35.8) 224 (64.2) July) 415 349 Late (Aug- 13 (6.2) 197 (93.8) 56 (27.3) 56 (27.3) Dec) 210 205 Sublocatio n 0.038* 0.284 Chepsonoi 221 22 (10.0) 199 (90.0) Kiborgok 204 12 (5.9) 192 (94.1) Tindinyo 217 18 (8.3) 199 (91.7) Kabar 106 (33.5) 210 (66.5) Central 316 Kabar West 258 78 (30.2) 180 (69.8) Owns 1 net:2 members 0.211 0.668 Yes 238 21 (8.8) 218 (91.6) 184 134 (72.8) 50 (27.2) No 400 31 (77.5) 369 (92.3) 374 272 (72.7) 102 (27.3) % HH slept under a net 0.115 0.610 0-25% 227 11 (4.8) 216 (95.2) 15 4 (26.7) 11 (73.3) 26-50% 67 8 (11.9) 59 (88.1) 12 2 (16.7) 10 (83.3) 51-75% 46 5 (10.9) 41 (89.1) 16 5 (31.2) 11 (68.8) 76-100% 293 28 (9.6) 265 (90.4) 515 173 (33.6) 342 (66.4) Malaria in the last year B 0.995 0.865 Yes 475 39 (8.2) 436 (91.8) 430 141 (32.8) 289 (67.2) No 158 13 (8.2) 145 (91.8) 128 43 (33.6) 85 (66.4) Fever in the last 2 weeks B 0.624 0.480 Yes 176 16 (9.1) 160 (90.9) 306 97 (31.7) 209 (68.3) No 456 36 (7.9) 420 (92.1) 252 87 (34.5) 165 (65.5) Own cattle 0.974 0.852 Yes 355 29 (8.2) 326 (91.8) 288 96 (33.3) 192 (66.7) No 284 23 (8.1) 261 (91.9) 270 88 (32.6) 182 (67.4) Own sheep 57 6 (8.1) 68 (91.9) 0.992 0.754 Yes 74 6 (8.1) 68 (91.9) 117 40 (34.2) 77 (65.8) No 565 46 (8.1) 519 (91.9) 441 144 (32.7) 297 (67.3)

153

Own goats 0.191 0.028* Yes 56 2 (3.6) 54 (96.4) 175 69 (39.4) 106 (60.6) No 583 50 (8.6) 533 (91.4) 383 115 (30.0) 268 (70.0) Own chickens 0.117 0.000* Yes 497 45 (9.1) 452 (90.9) 441 163 (37.0) 278 (63.0) No 141 7 (5.0) 134 (95.0) 117 21 (17.9) 96 (82.1) Own dogs 0.004* 0.212 Yes 155 4 (2.6) 151 (97.4) 325 114 (35.1) 211 (64.9) No 484 48 (9.9) 436 (90.1) 233 70 (30.0) 163 (70.0) HH reports seeing others repurpose bed nets 0.000* 0.011* Yes 450 48 (10.7) 402 (89.3) 525 180 (34.3) 345 (65.7) No 187 4 (2.1) 183 (97.9) 32 4 (12.5) 28 (87.5) APearson’s Chi-Square test; *Statistically significant p<0.05; BAt least one member in the household

References

1. Effiong U. Fighting malaria is going to take more than just nets. 2015. http://medicalexpress.com/news/2015-02-malaria-nets.html. Accessed 29 Jan 2019. 2. Shah S. In Africa anti-malaria nets go unused by recipients. Los Angeles Times. 2 May 2010; http://articles.latimes.com/2010/may/02/opinion/la-oe-shah-20100502. Accessed 29 Jan 2019. 3. Utuk NM, Abasiattai AM, Ugege WE. Misuse of long-lasting insecticidal nets in Akwa Ibom state, South-South Nigeria. J Case Rep Images Med. 2017; doi:10.5348/Z09-2017-37-LE-9 4. Eisele TP, Thwing J, Keating J. Claims about the misuse of insecticide-treated mosquito nets: are these evidence-based? PLoS Med, 2011; doi: 10.1371/journal.pmed.1001019 5. Bennett A, Smith SJ, Yambasu S, Jambai A, Alemu W, Kabano A, Eisele TP. Household possession and use of insecticide-treated mosquito nets in Sierra Leone 6 months after a national mass- distribution campaign. PLoS ONE, 2012; doi:10.1371/journal.pone.0037927 6. Honjo K, Chaves LF, Satake A, Kaneko A, Minakawa N. When they don’t bite, we smell money: understanding malaria bednet misuse. Parasit, 2013; doi:10.1017/S0031182012002077 7. Baume CA, Reithinger R, Woldehanna S. Factors associated with use and non-use of mosquito nets owned in Oromia and Amhara Regional States, Ethiopia. Malar J, 2009; doi:10.1186/1475-2875-8- 264 8. Roll Back Malaria Partnership (2018) Consensus statement on repurposing ITNs : applications for BCC messaging and actions at the country level 9. Krezanoski PJ, Campbell JI, Santorino D, Bangsberg DR. Objective monitoring of insecticide-treated bednet use to improve malaria prevention: SmartNet development and validation. PLoS ONE, 2017; doi: 10.1371/journal.pone.0168116 10. Koenker H, Kilian A, De Beyl CZ, Onyefunafoa EO, Selby RA, Abeku T, et al. What happens to lost nets: a multi-country analysis of reasons for LLIN attrition using 14 household surveys in four countries. Malar J, 2014; doi:10.1186/1475-2875-13-464 11. Coalson J, Santos EM, Little AC, Anderson E, Stroupe N, Agawo M, Munga S, Ernst KC (2019) Inadequate ratio of long-lasting insecticidal nets (LLIN) to household members challenges universal coverage in western Kenya: a cross-sectional study Under Review

154

12. Kibe LW, Kamau AW, Gachigi JK, Habluetzel A, Mbogo CM. A formative study of disposal and re- use of old mosquito nets by communities in Malindi, Kenya. MWJ, 2015;6,9 13. Lover AA, Sutton BA, Asy AJ, Wilder-Smith A. An exploratory study of treated-bed nets in Timor- Leste : patterns of intended and alternative usage. Malar J, 2011:10,199 14. Minakawa N, Dida GO, Sonye GO, Futami K, Kaneko S. Unforeseen misuses of bed nets in fishing villages along Lake Victoria. Malar J, 2008; doi: 10.1186/1475-2875-7-165 15. Loll DK, Berthe S, Faye SL, Wone I, Koenker H, Arnold B, Weber R. User-determined end of net life in Senegal: A qualitative assessment of decision-making related to the retirement of expired nets. Malar J, 2013; doi:10.1186/1475-2875-12-337 16. Doda Z, Solomon T, Loha E, Gari T, Lindtjørn B. A qualitative study of use of long-lasting insecticidal nets (LLINs) for intended and unintended purposes in Adami Tullu, East Shewa Zone, Ethiopia. Malar J, 2018; doi:10.1186/s12936-018-2209-5 17. Hoibak S (2010) Presentation of preliminary findings: long lasting insecticide treated net coverage and durability surveys, Dadaab Refugee Camps; 2010 18. Mutuku FM, Khambira M, Bisanzio D, Mungai P, Mwanzo I, Muchiri EM, et al. Physical condition and maintenance of mosquito bed nets in Kwale County, coastal Kenya. Malar J, 2013; doi:10.1186/1475-2875-12-46 19. Tan, K. R., Coleman, J., Smith, B., Hamainza, B., Katebe-Sakala, C., Kean, C., … Craig, A. S. (2016). A longitudinal study of the durability of long-lasting insecticidal nets in Zambia. Malaria Journal , 15 (1), 1–12. http://doi.org/10.1186/s12936-016-1154-4 20. D.K., L., S., B., S.L., F., I., W., B., A., H., K., … R., W. (2014). “you need to take care of it like you take care of your soul”: Perceptions and behaviours related to mosquito net damage, care, and repair in Senegal. Malaria Journal , 13 (1), 1–12. http://doi.org/10.1186/1475-2875-13-322 21. Miller, J. E., Jones, C. O. H., Ndunguru, S., Curtis, V., & Lines, J. (1999). A new strategy for treating nets. Part 2: Users’ perceptions of efficacy and washing practices and their implications for insecticide dosage. Trop Med Int Health , 4(3), 167–174. 22. Honjo, K., & Satake, A. (2014). N-player mosquito net game: Individual and social rationality in the misuse of insecticide-treated nets. Journal of Theoretical Biology , 342 , 39–46. http://doi.org/10.1016/j.jtbi.2013.11.001 23. Mclean, K. A., Byanaku, A., Kubikonse, A., Tshowe, V., Katensi, S., & Lehman, A. G. (2014). Fishing with bed nets on Lake Tanganyika: a randomized survey - Kate_A_McLean.pdf, 1–5. Retrieved from http://ihi.eprints.org/2937/1/Kate_A_McLean.pdf 24. Habonimana, D. (2017). Improving Long-Lasting Insecticidal Nets use in Kayange community of North-Western Burundi : A Quality Improvement Study A research report submitted to the Faculty of Health Science in partial fulfillment of the requirements for the degree of Master of Science ( MSc ) in Epidemiology - Implementation Science. 25. Larsen, D. A., Welsh, R., Mulenga, A., & Reid, R. (2018). Widespread mosquito net fishing in the Barotse floodplain: Evidence from qualitative interviews. PLoS ONE , 13 (5), 1–16. http://doi.org/10.1371/journal.pone.0195808 26. Short, R., Gurung, R., Rowcliffe, M., & Hill, N. (2018). The use of mosquito nets in fisheries : A global perspective, 1–14. 27. Public Health Service (2003) Toxicological profile for pyrethrins and pyrethroids. U.S. Department of Health and Human Services. Atlanta, Georgia 28. Bush, E. R., Short, R. E., Milner-Gulland, E. J., Lennox, K., Samoilys, M., & Hill, N. (2017). Mosquito Net Use in an Artisanal East African Fishery. Conservation Letters , 10 (4), 450–458. http://doi.org/10.1111/conl.12286 29. Haya, K. (1989). Toxicity of pyrethroid insecticides to fish. Environmental Toxicology and Chemistry , 8(5), 381–391. http://doi.org/10.1002/etc.5620080504 30. Hill, I. R. (1989). Aquatic organisms and pyrethroids. Pesticide Science , 27 (4), 429–457. http://doi.org/10.1002/ps.2780270408

155

Declarations

Ethics approval and consent to participate:

The current analysis involved secondary analyses of the de-identified data from a primary cross-sectional study. The primary study was approved by the Kenyan Medical Research

Institute (KEMRI) (SSC #2810) and the University of Arizona ethical review boards with deferral of primary oversight from the University of Arizona to KEMRI. Informed consent was obtained from all household heads. Parental consent was obtained for children less than 7 years of age, and assent was obtained for those 8 to 17 years of age.

Consent for publication:

Not Applicable

Availability of data and material:

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

The present analyses were prepared with no funding sources. The primary study was funded via NIH/NIAID R15 Al100118-01.

Authors’ contributions

ES analyzed and interpreted the data, and drafted the manuscript. JE, MA, and KE assisted with interpreting the results. EJ, YK, and KE offered scientific guidance in preparation of the analyses, and assisted in the manuscript preparation. SM and MA reviewed the results and

156

conclusions and offered contextual support. MA and JE performed data cleaning. All authors read and approved the final manuscript.

Acknowledgements

We thank Lindsay Hansen for her writing guidance and editing contributions to this manuscript.

157

APPENDIX E: MANUSCRIPT 3

Malaria education interventions addressing bed net care and repair practices: a systematic review

Ellen M Santos* 1 [email protected] Corresponding Author

Jean McClelland 1 [email protected]

Colleen Shelly 2 [email protected]

Elizabeth T Jacobs 1 [email protected]

Yann C Klimentidis 1 [email protected]

Kacey C Ernst 1 [email protected]

1 Mel and Enid Zuckerman College of Public Health, University of Arizona, 1295 N Martin Ave, Tucson, AZ 85724

2 Massachusetts General Hospital Cancer Center, 55 Fruit St, Boston, MA 02114

158

Malaria education interventions addressing bed net care and repair practices: a systematic review

Introduction

The distribution of long lasting insecticide treated nets (LLINs) is a key strategy for malaria prevention. Within recent decades, bed net technology has evolved along with the recommendations for maintaining them. Early untreated nets offered physical protection from mosquito bites and showed the ability to reduce parasitemia [1] and required careful handling to avoid rips and tears. The introduction of insecticide treated bed nets (ITNs) led to even greater reductions in malaria morbidity and mortality [2, 3] as reduction in vector populations was coupled with physical protection, particularly when deployed at high levels within communities

[4, 5, 6]. Introducing ITNs required campaigns to treat existing nets and new distributions of nets that were already impregnated with insecticide, which met challenges in reaching high levels of net treatment and retreatment [7]. Recommendations for ITN care included (1) washing the net the day prior to treatment/retreatment (2) soaking the net in the insecticide mixture (3) gently wringing out the net, but not too vigorously or insecticide can be lost (4) Drying flat in the shade

(5) washing gently, as seldom as possible, and never in lakes, rivers, or streams and (6) retreating every 6 months or after every 3 washes [8].

The newest models of bed nets include LLINs, which eliminate the need to retreat nets with insecticide with an intended effective life of 3-5 years. LLIN care recommendations include

(1) never to seldomly washing, at most once every 3 months (2) washing gently with mild soap, and never in a lake, river, or stream (3) drying in the shade and avoiding exposure to direct sunlight (4) tying or hanging the net up during the day and (5) repairing all holes quickly [9].

However, recommendations for some OlysetTM nets include a regeneration process by exposing washed nets to sunlight for 5-6 hours in the original transparent vinyl packaging [10].

159

Bed net care and repair refers to the actions taken to adhere to maintenance instructions, prevent damage, and repair damage to nets [11]. Many studies describing net care and repair behaviors conclude that education interventions or behavior change communication (BCC) strategies should be implemented to improve care and repair adherence [12, 13, 14, 15, 16, 17,

18, 19]. We refer to education interventions as an activity that includes a form of messaging aimed at changing or improving knowledge and/or behaviors. These include interventions such as BCC, Information, Education, and Communication (IEC) campaigns, and social marketing campaigns.

Changing health behaviors is a goal of many public health initiatives, and malaria education interventions are no different. Health behavior theories guide understanding of health behaviors, and grounding interventions with goals of behavior change within appropriate theory is essential for a comprehensive and effective campaign [20, 21]. A systematic review of health behavior articles from 2000-2005 revealed that 35.7% of the included studies mentioned theory

[22]. For interventions to translate into evidence-based public health practice, they must be described in appropriate detail for comparability and replication [23]. Furthermore, education interventions involve instruction at their core – methods to convey content to the target audience.

Beyond sound health behavior theory, interventions should describe the specific instructional strategies employed [24], as delivery of the educational events is highly important to the intervention’s success.

While multiple studies suggest the need for education interventions to improve adherence to bed net care and repair recommendations, the evidence of intervention effectiveness across differing contexts remains unclear. Understanding the methods that lead to increases in adhering to net care and repair recommendations will allow for more comprehensive and more effective

160

education interventions. We conducted a systematic review of education interventions that address elements of bed net care and repair adherence, and assessed health behavior and instructional theoretical frameworks involved.

Methods

We conducted a systematic literature review with the objective of assessing whether education interventions increase knowledge of or adherence to recommended bed net care and repair practices. We evaluated net durability or longevity as a potential secondary outcome.. We searched for interventions targeting either ITNs or LLINs. Though ITNs are no longer distributed, interventions focusing on ITNs were included because we were interested in the change in knowledge or practices, and because many of the recommendations for ITNs and

LLINs overlap. We hypothesized that the efficacy of education campaigns is modified by the specific care and repair practices targeted and by the degree to which the intervention applied educational and health behavior theories.

Search strategy

The protocol was registered through PROSPERO International Prospective Register of

Systematic Reviews on 30 April 2019 (registration number CRD42019123932). We identified studies through a literature search using controlled vocabulary terms and key word combinations for “education intervention” terms, “mosquito net” terms and malaria (Table 1, Appendix H).

The search was conducted on February 13, 2019 in MEDLINE PubMed (1946-2019, 13

February), Embase (1947-2019, 13 February), Cochrane Controlled Register of Trials

161

CENTRAL (1966-2019, 13 February), Bibliomap (indexing date not reported – 2019, February

13), ProQuest Dissertations & Theses (1982-2019, 13 February), Trials Register of Promoting

Health Interventions TRoPHI (2004-2019, 13 February) and Clarivate Web of Science Core

Collection (1900-2019, 13 February) databases (Table 2). Citations were imported into EndNote

X8 (Thomas Reuters) for deduplication and management.

Table 1. Search terms adapted for use in database searches for peer-reviewed studies Category Search Terms Bed nets “mosquito nets” [Mesh] “LLIN” “LLINs” “long lasting insecticide treated nets” “long lasting insecticide treated bednets” “long lasting insecticide treated bed nets” “long lasting insecticidal nets” “long-lasting insecticide-treated nets” “ITN” “ITNs” “Insecticide treated nets” “insecticide treated bednets” “insecticide treated bed nets” “Icon Life” Yorkool Olyset Netprotect Permanet Malaria “Malaria” [Mesh] Malaria Education interventions “health education” [Mesh] “health behavior” [Mesh] “intervention” “education intervention” “malaria education” “education campaign” “behavior change” “behaviour change”

The searches identified 6507 studies prior to removing duplicates. Deduplication in

EndNote X8 (Thomas Reuters) resulted in 5927 studies. Studies were reviewed following the

162

protocol set forth by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses

(PRISMA) guidelines. Initial screening of titles and abstracts was completed by 2 reviewers (ES,

CS) with a third reviewer (KE) assessing discrepancies, according to the following study inclusion criteria: malaria education intervention, communication campaign, or behavior change campaign pertaining to bed net care and repair practices for ITNs or LLINs. We managed data in

Google Sheets (Google, Mountain View, CA) with individual reviewer columns hidden to other reviewers to maintain a blinded screening process.

After initial screening, full text articles for 446 studies were retrieved and screened by 2 reviewers (ES, CS) with a third reviewer (KE) assessing discrepancies through the same blinded process as initial screening. Studies were assessed according to the same inclusion criteria as in the initial screening process with the application of the following exclusion criteria (Figure 1): malaria intervention that does not pertain to bed net care and repair, education intervention not relevant to malaria, review papers, conference abstracts, and commentary/opinion pieces.

Grey literature was identified through relevant organizations that support bed nets for malaria prevention (Table 2). Organizations were searched between 12-20 March, 2019. All records related to malaria education and/or bed net care and repair practices were retrieved and reviewed to be summarized.

Table 2. Databases and malaria organizations searched for malaria education intervention records Peer-Reviewed Literature Databases Grey Literature: Malaria Organizations BiblioMap African Leaders Malaria Alliance Cochrane Central Register of Controlled Against Malaria Foundation Trials (CENTRAL) Embase Bill & Melinda Gates Foundation ProQuest Dissertations and Theses Global The Global Fund PubMed Imagine No Malaria Trials Register of Promoting Health Kenya NGOs Alliance Against Malaria Interventions (TRoPHI) Web of Science Malaria Atlas Project JSTOR Malaria Consortium

163

Malaria No More Medicines for Malaria Venture National Institute of Malaria Research Nothing But Nets Roll Back Malaria South African Malaria Initiative United States Agency for International Development World Health Organization

Data abstraction

The resulting 43 peer-reviewed studies were included for data abstraction. We collected details of the study location, population, and years of intervention along with the reported malaria epidemiology for context. Additionally, we abstracted data on the intervention purpose and description, the care and repair practices targeted, the results from intervention and comparison groups, the types and descriptions of education activities, and the details of health behavior theories and/or instructional models described.

Quality Assessment

Due to the variety of study designs including quantitative and qualitative designs, a single existing quality screening tool was not appropriate to assess the quality of the included studies in relation to the study question. We developed a 6-item quality criteria tool (Table 3) and scored the studies that had education interventions as a main component. Studies received either 1

(contains the criteria item) or 0 (does not contain the criteria item) points for each quality criteria, resulting in a potential score range between 0-6. Articles containing 1-2 criteria received

1 star, those containing 3-4 received 2 stars, and those containing 5-6 received 3 stars.

Table 3 . Education intervention quality criteria tool Quality Criteria 1 point 0 points 1 Was there a well-defined comparison group? Yes (1) No (0)

2 Was the comparison group selected appropriately? Yes (1) No or answer

164

to question 1=0 (0) 3 Was there an explicit link to appropriate health behavior theory? (An Yes (1) No (0) appropriate theory is one that targets the appropriate level of audience of individual, community, children, or adults) 4 Was there an adequate description of educational activities? Yes (1) No (0)

5 Were the educational activities equitable for diverse learners? (Equitable Yes (1) No (0) activities include the implementation of multiple types of activities such as audio, visual, interpersonal, etc) 6 Was there hypothesis testing of the differences between groups pertaining Yes (1) No (0) to net care knowledge and/or practices?

Results

We identified 6,507 peer-reviewed records from the database search (Table 2) and removed duplicates in EndNote X7 (Clarivate Analytics, Philadelphia, PA), leaving 5,927 records for title and abstract screening (Figure 1). We excluded 5,481 records through title and abstract review where 1,292 records were duplicates not caught during the initial de-duplication process and 4,189 were deemed not relevant in initial screening. This resulted in 446 full text articles that were reviewed, and from this process we excluded 391 articles using the inclusion and exclusion criteria. Most articles were excluded because they were not malaria education interventions (n=206) or because they were education interventions, but were not related to bed net care and repair practices in any way (n=87) (Figure 1). Others were found to be conference abstracts (n=7), commentary pieces (n=46), reviews (n=29), or were excluded or other reasons

(n=16). Of the 43 remaining articles, we abstracted data from 29, excluding 14 in the process because they lacked care or repair results (n=6) or because the education was too general, meaning we found no evidence that education on net care and repair practices was provided

(n=8).

After searching multiple malaria organizations (Table 2) we identified 387 records pertaining to malaria education (Figure 1) in the grey literature search. We screened all of these

165

records for eligibility to be summarized, excluding 98 in the process. We excluded records if they did not mention anything about malaria health education or bed net care and repair (n=85) or if they were toolkits or guides, hence not an activity that included education on care and repair

(n=13). This resulted in 289 grey literature records to be summarized (Figure 1

Figure 1. PRISMA scheme of record selection and exclusion criteria

Peer Reviewed Literature

Malaria education as a main focus

An education intervention was the main activity of 16 (55.2%) of the 29 included peer- reviewed articles (Table 4), and a relatively minor activity of the remaining 13 articles (44.8%)

(Table 5). Of the 16 articles with education intervention as a main activity, 12 were studies conducted in sub-Saharan Africa and the remaining 4 conducted in Latin America, the Middle

East, and Southeast Asia (Table 4). Regarding the main outcomes measured in Table 4, 6 studies focused on increasing ITN or LLIN use [25, 26, 27, 28, 29, 30], 6 focused on generally improving malaria knowledge, attitudes, and/or practices [27, 29, 30 31, 32, 33], and 5 focused more specifically on outcomes related to net durability or care and repair [34, 35, 36, 37, 38].

Concerning the care and repair practices addressed in the articles, 9 provided education of ITN treatment or retreatment with insecticide [26, 27, 30, 31, 32, 33, 34, 39, 40], 4 educated about net storage during the day [25, 36, 37, 38], 2 about appropriate soap types to use in washing nets [37,

38], 1 about recommended drying practices [37], and 1 about the manner of washing nets gently

[38].

Key findings of education on net care and repair knowledge and/or practices are described for each article in table 3. We assessed whether studies were informed by or utilized

166

health behavior and/or instructional theories or models (Table 6). Of the 16 articles that had health education as a main activity, 10 were informed by, applied, or built health behavior theories [25, 26, 31, 33, 35, 36, 37, 38, 39, 40] while 6 mentioned no grounding in theory (Table

4) [27, 28, 29, 30, 32, 34]. Of those mentioning theory, 6 were informed by theory, 2 applied theory, and 2 built or built upon theory.Based on the key bed net care and repair findings, we summarized whether there was evidence that the education interventions led to improvements in care and repair knowledge, practices, and/or net durability. There were 8 studies that reported overall improvements [25, 28, 30, 32, 33, 38, 39, 40], 4 that had mixed results [27, 31, 35, 37], and 4 that were unclear in whether improvements occurred [26, 29, 34, 36]. There is no apparent pattern in care and repair conclusions in regard to whether theory was mentioned or how theory was used (Table 6). Of those articles not grounded in theory (n=6), 50% reported positive care and repair outcomes, while 5 (50%) of the articles grounded in theory (n=10) reported positive care and repair outcomes.

Table 6. Care and repair conclusions in relation to theory use of education interventions Care and repair conclusions How theory was Postiive Negative Mixed Unclear used No theory [28], [30], [32] [27] [29], [34] Informed by theory [33], [39], [40] [35], [37] [26] Applied theory [31] [36] Built or built upon [25], [38] theory

Table 7 describes the care and repair conclusions in relation to the types of education strategies implemented. Mass education includes passive interventions where no personal contact is made between those implementing and the recipients of the education or active interventions where there is personal contact between education implementers and recipients. There were 6 interventions that used passive mass education, which included activities such as posters,

167

slogans, print and audiovisual media, pamphlets, t-shirts, radio programs, newspapers, maintenance labels, and social marketing [26, 30, 33, 35, 37, 38]. Active mass education activities were conducted by 12 interventions and included dramas, music, poetry, art, demonstrations, community events and workshops, lectures, small group flip-chart presentations, and advocacy by community health workers [25, 26, 30, 31, 32, 33, 34, 35, 37, 38, 39, 40].

Interpersonal education involves activities such as education through house-to-house visits, establishing village bed net committees that talk with community members, dialogues and conversations, debates, training local health leaders, and individual flip-chart education. There were 8 interventions that used interpersonal methods [25, 27, 28, 32, 33, 36, 37, 38]. Discerning between mass and interpersonal education methods, there were 6 interventions that used mass strategies only, and 3 interventions that used interpersonal strategies only (Table 7). Care and repair conclusions of those 6 that used only mass education were 2 positive, 1 negative, 2 unclear, and 2 mixed results (Table 7). There was a positive, unclear, and mixed result each from those 3 articles using only interpersonal strategies. Stronger results were observed from interventions that combined at least one mass education strategy with interpersonal methods, where 4 interventions had positive care and repair results and 1 had mixed results (Table 7).

Table 7. Care and repair conclusions in relation to education methods [citation]  Care and repair Mass education Interpersonal conclusions education

Passive mass Active mass education education

[25]  Positive X X

[34]  Unclear X

[26]  Unclear X X

[27]  Mixed X

168

[31]  Negative X

[35]  Mixed X X

[32]  Positive X X

[39, 40]  Positive X

[33]  Positive X X X

[28]  Positive X

[29]  Unclear

[30]  Positive X X

[36]  Unclear X

[37]  Mixed X X X

[38]  Positive X X X

Figure 2 depicts the care and repair outcomes through the interaction between education type (mass vs. interpersonal vs. both) and theory use. The only clear pattern in outcomes exists among interventions that used both mass and interpersonal education methods, regardless of theory use, where 4 articles reported positive results and 1 reported mixed results. The article

[32] that used both education methods but did not use theory reported positive net care and repair results. A variety of results was found among articles that used a single method of education and did or did not use theory (Figure 2).

Figure 2. Care and repair conclusions in relation to education methods and theory use

Quality Assessment

Education intervention quality was assessed for the 16 articles that had health education as a main activity (Table 4) using the rubric detailed above (Table 3, Appendix F). We classified

169

the education intervention component of 3 articles as single star [27, 29, 34], of which 2 had unclear results and one had mixed results of bed net care and repair measurements. Four articles had two stars for their education intervention components [26, 36, 39, 40]. Two of these articles had unclear care and repair conclusions, while improved knowledge was measured in the other two. However, the two articles reporting improved knowledge were describing the same study

[39, 40]. There were 9 articles that received 3 stars, 6 of which reported improvements in care and repair knowledge and/or practices [25, 28, 30, 32, 33, 38], and 3 of which reported mixed results of care and repair knowledge or practices [31, 35, 37]. None of the one star quality-rated education interventions, all of the two star interventions, and 6 of the 3 star interventions mentioned theoretical framework.

Malaria education as a minor element

Of the 13 articles which contained an element of malaria education, but education was not a main focus, 9 were studies conducted in sub-Saharan Africa, and the remaining 4 in South and Southeast Asia and South America (Table 5). The main outcomes studied ranged from malaria indices [41, 42, 43], aspects of bed net care and repair [10, 44, 45, 46, 47], program evaluation [43, 48, 49], and bed net coverage and/or use [44, 46, 47, 50, 51]. The bed net care and repair practices addressed by the articles in table 4 include washing frequency [10, 41, 42,

46, 47, 50, 51, 52], ITN treatment and/or retreatment with insecticide [42, 44, 45, 46, 48, 49, 52], net storage during the day [42], bed net repair [10, 42, 44, 52], appropriate soap type [47], proper drying practices [47, 51], and general care messages [43, 45]. Only one of the 13 articles in table

4 mentioned theoretical framework. Snow et al 1999 used social marketing strategies to increase net ITN retreatment and found that house to house retreatment schemes can be highly effective in increasing retreatment, but that the delivery mechanism is more important than social

170

marketing [45]. Overall, 5 articles reported improvements in bed net care and repair knowledge and/or practices [42, 48, 47, 49, 52], 3 articles had mixed results [44, 45, 51], 4 had unclear results [41, 43, 46, 50], and 1 reported negative results following education efforts [10] (Table

5).

The number of bed net care and repair messages varied between the 29 peer-reviewed studies. More than half of interventions provided education on only one care and repair recommendation (15 studies), 4 articles educated on 2 recommendations, 6 articles on three recommendations, and only 3 studies educated on 4 or more recommendations (Tables 4 and 5).

Grey Literature

Of the 289 grey literature records included for summarization (Figure 1), 286 (99%) were from the Against Malaria Foundation [53]. These records encompass 136 net distributions completed by various organizations between 2006 – 2017 across 26 countries in sub-Saharan

Africa, 5 countries in South and Southeast Asia, and 3 countries in Latin America. Each distribution included a pre-distribution report and at least one distribution report detailing the plans and completed actions of the distributions including education components. We summarized the specific bed net care and repair education messages reported through these distribution reports (Figure 3). More than half (80, 58.8%) of the distributions did not mention any aspects of bed net care and repair, while 38 (27.9%) of the distributions included messaging around general net care defined as stating only “maintenance” or “general care instructions”.

There were 26 (19.1%) of distributions addressed specific care and repair practices including washing (15.4%), drying (6.6%), avoiding damage (2.2%), appropriate storage during the day

(2.2%), repair (1.5%), and appropriate soap types (0.7%) (Figure 2).These do not add to 100% because some reports included both general care and at least one specific care practice.

171

Figure 2. Bed net care and repair messages reported among Against Malaria Foundation-funded bed net distributions

The remaining 3 grey literature records included were from Imagine No Malaria (1 record) and Malaria Consortium (2 records). Imagine No Malaria did not detail a particular education intervention, but rather reported that the organization supports bed net distributions by training community health workers to educate households on proper net use and care of nets during bed net delivery and installation activities [54].

Malaria Consortium included 2 particular projects that contained components of bed net care and repair education. The Pioneer Project conducted from 2009-2014 in Uganda promoted the message to “Dry your net under the shade after washing it” though net care was not a main focus of BCC activities [55]. Interviews and focus groups discussions from The Malaria

Prevention and Control Project in Mozambique from 2011-2017 discussed exposure to net care and repair messaging during a LLIN distribution 6 months previously [56]. Participants had high awareness that net care and repair is important and good recall of messages to repair torn or damaged nets, but the accurate recall of other care recommendations was inconsistent [56].

Specifically, respondents did not know appropriate net washing locations (in the river vs. a washtub or bucket) or whether nets should be dried in the shade or sun [56].

Discussion

Education interventions resulted in improvements of bed net care and repair knowledge and/or practices in less than half of all included peer-reviewed articles (n=13, 44.8%) and in 8

(50%) of the articles with an education intervention as a main study component. Interventions that received 3 stars in the quality assessment were more likely to include an appropriately selected comparison group, be grounded in theory, and use equitable instructional activities

172

(Appendix F). Evaluation of education intervention effectiveness is discussed assessing theoretical grounding, education strategies and methods, and the implications on net durability and longevity.

Theoretical grounding

It is thought that theory based health education interventions are more effective than non- theory based interventions [21] because they outline the rationale and process for how the intervention will result in change [57]. However, we found no differences in success between interventions grounded in theory and those that did not mention theory. This was further confirmed by the interaction between education method type and theory use (Figure 2) which showed that theory does not appear to modify care and repair outcomes in relationship to the method of education. Rather, it seems that using a combination of mass education and interpersonal education methods shows more evidence for intervention effectiveness than grounding in theory, though this requires further exploration for confirmation.

Interventions that were informed by or applied theory achieved varied net care and repair results. Though there were only 2 interventions that built or built upon theory, both resulted in positive care and repair outcomes. Building or building upon theory may be the most successful way to implement education interventions because it requires extremely context-specific knowledge of the target populations, and care and repair practices are known to vary by community [12]. However, a larger evidence base is needed to determine how utilizing health behavior theory influences the success of education interventions aimed at improving bed net care and repair practices. Potentially clouding the effect of theory on intervention effectiveness is the wide variety of theories utilized. No two articles used the same theory. This is not surprising given the large number of health behavior theories available. As of the 1990s, 66 health behavior

173

theories and models were found among 536 articles to inform education intervention designs

[58], and more theories and models are continually developed for differing health contexts such as the net care and repair conceptual framework [38] utilized by Helinski et al [37].

Education

Comparison groups

Studies that do not adequately describe the selection of appropriate comparison groups makes it difficult to synthesize the overall effectiveness of education interventions on changing knowledge or behaviors. Without knowing whether comparison groups could be contaminated with intervention messaging, or without directly measuring this contamination, it is difficult to know if exposure misclassification affects intervention outcomes. Some studies purposefully selected control or comparison groups to limit the possibility of contamination. Amoran et al. randomly selected intervention and control sites in geographically separated areas [30]. Helinksi et al. chose culturally similar, but geographically separated sites where the populations spoke different languages [37], and Koenker et al. chose a control site that was culturally similar to the intervention site, but that was outside of the range of broadcasting services [38]. Furthermore, studies that use passive mass education methods survey respondents on exposure to specific intervention slogans to measure the level of exposure misclassification [37, 38]. Furthermore, interventions that can provide evidence of temporality, dose-response relationships of behavior adoption and exposure to education activities, explanations for how change occurred, and methods to minimize effects of non-randomization will help build evidence of intervention effectiveness [59].

Descriptions of instructional strategies

174

Interventions receiving 3 stars in the quality assessment detailed replicable descriptions of the instructional strategies that were implemented, however still lacked a framework for the actual provision of the instructional events. Health education interventions are inherently multidisciplinary, often requiring expertise in public and global health, health promotion, and anthropology, yet they appear to lack involvement of experts in education – those that understand and apply learning theories and instructional frameworks. Of the articles with malaria education as a main focus (Table 4), only 3 indicate author affiliations with individuals outside of public health, medical, and anthropology realms including one author from a humanities department [Alvarado 2006] and authors from a center for communication programs [37, 38]. In

2005, Kinzie linked health behavior theories and instructional frameworks specifically to provide instructional design strategies for health educators [24], but this resource was not cited by any of the articles in Table 3. Though not related to net care and repair, one intervention aimed at improving malaria knowledge among travelers explicitly utilized learning theory to develop an educational game, resulting in significant knowledge improvements [60]. More emphasis should be placed on the learning process for differing target audiences in the design and implementation of malaria education interventions, particularly interventions that educate through interpersonal communication.

Types of instructional strategies

Though not explicitly defined as such, the interventions receiving 3 stars employed equitable educational activities, meaning activities accessible to diverse learners. This follows a principle of universal design for instruction which is often discussed for post-secondary education [61], but is applicable to all instructional settings. Offering various ways of delivering

175

educational content allows information to be accessible to multiple types of learners [61]. In the context of BCC campaigns, this means using various messaging tactics including visual, auditory, and interpersonal communication strategies to send a unified message. For example, one intervention employed only one mode of education (auditory messages during sermons) [34] while other interventions combined multiple instructional activities such as flipbooks, radio messaging, posters, and drama performances [25, 26, 28, 30, 31, 32, 33, 35, 36, 37, 38]. A review of malaria social and behavior change communication campaigns concluded that interventions using multiple activities, coupling mass media messaging with community-tailored strategies increased the likelihood of obtaining behavior change [62]. Providing multiple types of educational activities is particularly important in areas with disparities in literacy rates. Written materials along with other visual and auditory methods can ensure that populations with varying literacy levels are exposed to the same messaging in a way they can best learn.

Overall educational methods

Interventions reached target populations using mass education, interpersonal education, or a combination of the two. It is likely using the combination of mass and interpersonal strategies may be most effective, as all 5 interventions that used a combination reported positive

(4 studies) or mixed (1 study) care and repair results [25, 32, 33, 37, 38]. There was no discernable pattern of care and repair conclusions among interventions that used only mass or interpersonal education methods, however, these conclusions are tenuous due to the small sample size of interventions.

Comparing care and repair results across interventions is further complicated because implementation of education exposure differs between passive mass, active mass, and interpersonal education methods. Passive mass education is the easiest to implement

176

consistently, while more variation dependent on individuals providing education is expected with active mass and interpersonal education. Rhee et al implemented interpersonal education to increase the proportion of retreated ITNs, but obtained mixed results where more individuals who received education retreated ITNs, but there was no difference between intervention and control households in the number that retreated nets [27]. A potential explanation for weak results was found due to interaction in retreatment outcomes depending on the field guide who implemented the education even though all guides received the same training [27]. Education methods using individuals to provide the instruction should have some training in education theory to obtain the greatest possible effectiveness and highest possible degree of consistency between educators. For example, a school based education intervention that improved ITN retreatment practices in Ghana [33] specifically trained instructors in Participatory Learning and

Action (PLA), an evidence-based facilitated approach that actively engages participants in the targeted interventions [63].

Comprehensive versus single message education

There is a lack of comprehensive care and repair messaging among both education interventions and bed net distributions. Ten (63%) of the articles with malaria education as a main component addressed only a single care and repair practice, 3 (18.8%) educated on 2-3 practices, and only 3 (18.8%) interventions educated on 4 or more practices. It is important to note that very few of the included interventions were focused on improving care and repair knowledge and/or practices. Most interventions were concerned with other outcomes such as improving net use, thus care and repair education was a relatively minor priority. There was also a serious lack of specificity in care education of net distributions funded by the Against Malaria

Foundation where 86.7% of distribution reports described no care and repair messaging or

177

general messaging defined as only stating “net care” or “net maintenance” in the education description (Figure 3). To achieve maximal improvements in net longevity and durability, comprehensive education interventions dedicated to improving care and repair practices are needed, and net distribution campaigns should ensure that care and repair messaging is consistent and comprehensive. One such intervention that focused on net care and repair knowledge and practices incorporated comprehensive messaging and resulted in improved care and repair knowledge, practices, and net longevity [38], suggesting that education interventions can lead to promising results, though more evidence is needed.

Impacts on bed net durability and longevity

While education interventions led to marginal improvements in knowledge and practices, there is little evidence that these improvements improved net durability and longevity, largely because only three studies measured bed net durability as an outcome. One found that education increased care and repair knowledge and attitudes, but did not improve net durability [37] and one found that repair practices significantly increased, but did not improve net condition [35], while the other found that nets tied up while not in use were more likely to be in serviceable condition at follow up [38]. More education interventions of bed net care and repair that measure net durability as an outcome are needed. Future interventions should be conducted through LLIN distributions before nets are used to demonstrate whether bed net care and repair education improves net durability and longevity. Furthermore, more emphasis on net care education should be placed on net distributions funded by non governmental organizations (NGOs) such as the

Against Malaria Foundation (AMF) to ensure all net recipients are receiving complete and accurate net care recommendations. When an organization wishes to implement a net distribution funded by AMF, they are required to provide plans for educational components of the

178

distribution [53]. However, this requirement is vague and does not specify particular content that should be conveyed, resulting in more than half of the 136 completed distributions that did not mention any education plans or education conducted regarding net care or repair (Figure 3).

Limitations

The search terms “ Plasmodium ” and “ Anopheles ” were not included in the search strategies, and thus could have caused us to miss some records. However, this is likely a minor issue considering we assessed a large number of records from a wide variety of databases. Due to the variety of study designs of the included interventions, the quality assessment used was not a standardized or validated assessment tool. Rather, it is a way to highlight important elements of education interventions and should be evaluated as a description rather than a judgement. We transparently reported the scoring of each included study so that readers can evaluate our decision and make their own judgements (Appendix F).

Conclusions and recommendations

It is difficult to make strong conclusions regarding the efficacy of education interventions on improving net care and repair knowledge, practices, or durability as most of the included interventions educated on care and repair as a secondary or minor objective. We did not find evidence that interventions grounded in health behavior theory achieved more positive results than those not grounded in theory. Interventions that employed both mass and interpersonal education showed promising results that warrant further investigation.

For increased rigor and comparability of future education interventions, we recommend the following:

179

1. Because education interventions on care and repair practices must be highly context-specific, a

focus on training instructors in pertinent education theory may help improve education quality

and consistency to yield stronger results.

2. Interventions should be planned, monitored, evaluated, and reported using guidance put forth by the RBM working group. There is a wealth of material freely available including standardized

indicators, survey development guidance, and checklists to ensure complete reporting [57, 64,

65, 66].

References

1. Burkot TR, Garner P, Paru R et al. (1990) Effects of untreated bed nets on the transmission of Plasmodium falciparum, P. vivax and Wuchereria bancrofti in Papua New Guinea. Transactions of the Royal Society of Tropical Medicine and Hygiene 84 , 773 – 779. 2. Alonso PL, Lindsay SW, Armstrong JRM, Conth M, Hill AG, et al (1991) The effect of insecticide-treated nets on mortality of Gambian children. Lancet 337 , 1499-1502 3. Lengeler C (2004) Insecticide-treated nets for malaria control: real gains. Bulletin of the WHO 82 (2), 84 4. Binka FN, Indome F, Smith T. Impact of spatial distribution of permethrin-impregnated bed nets on child mortality in rural northern Ghana. Am J Trop Med Hyg. 1998;59:80–5. 5. Howard SC, Omumbo J, Nevill C, Some ES, Donnelly CA, Snow RW. Evidence for a mass community effect of insecticide-treated bednets on the incidence of malaria on the Kenyan coast. Trans R Soc Trop Med Hyg. 2000;94:357–60. 6. Gosoniu L, Vounatsou P, Tami A, Nathan R, Grundmann H, Lengeler C. Spatial effects of mosquito bednets on child mortality. BMC Public Health. 2008;8:356 7. Fraser-Hurt N, Lyimo EO. (1998). Insecticide-treated nets and treatment service: a trial using public and private sector channels in rural United Republic of Tanzania. Bull World Health Organ, 76 (6), 607-615. 8. World Health Organization (2002) Instructions for treatment and use of insecticide-treated mosquito nets. WHO/CDS/RBM/2002.41 9. Vestergaard (2014) PermaNet 2.0 Accessed 11 May 2019 10. Shirayama Y, Phompida S, Kuroiwa C, Miyoshi M, Okumura J, Kobayashi J (2007) Maintenance behaviour and long-lasting insecticide-treated nets (LLITNs) previously introduced into Bourapar district, Khammouane province, Lao PDR. J Public Health 121 , 122-129. 11. NetWorks Summary Series (2015) ITN care & repair: improving net lifespan through behavior change communication. United States Agency for International Development. 12. Gnanguenon V, Azondekon R, Oke-Agbo F, Beach R, Akogbeto M. Durability assessment results suggest a serviceable life of two, rather than three, years for the current long-lasting insecticidal (mosquito) net (LLIN) intervention in Benin. BMC Infect Dis. 2014; doi:10.1186/1471-2334-14-69 13. Azondekon R‚ Gnanguenon V‚ Oke-Agbo F‚ Houevoessa S‚ Green M‚ Akogbeto M. A tracking tool for long-lasting insecticidal (mosquito) net intervention following a 2011 national

180

distribution in Benin. Parasit Vector. 2014;7:6 14. Loll DK, Berthe S, Faye SL, Wone I, Arnold B, Koenker H, et al. (2014) “You need to take care of it like you take care of your soul”: perceptions and behaviors related to mosquito net damage, care, and repair in Senegal. Malar J 13, 322. 15. Hunter GC, Scandurra L, Acosta A, Koenker H, Obi E, Weber R. We are supposed to take care of it: A qualitative examination of care and repair behaviour of long-lasting, insecticide-treated nets in Nasarawa State, Nigeria. Mal J. 2014; doi:10.1186/1475-2875-13-320 16. Mutuku FM, Khambira M, Bisanzio D, Mungai P, Mwanzo I, Muchiri EM, et al. (2013) Physical condition and maintenance of mosquito bed nets in Kwale County, coastal Kenya. Malar J 12 , 46. 17. Leonard L, Diop S, Doumbia S, Sadou A, Mihigo J, Koenker H, et al. Net use, care and repair practices following a universal distribution campaign in Mali. Mal J. 2014; doi:10.1186/1475- 2875-13-435 18. Scandurra L, Acosta A, Koenker H, Kibuuka D, Harvey S. “It is about how the net looks”: a qualitative study of perceptions and practices related to mosquito net care and repair in two districts of eastern Uganda. Mal J. 2014;13:504. 19. Wills AB, Smith SC, Anshebo GY, Graves PM, Endeshaw T, Shargie EB, et al. Physical durability of permanet 2.0 long lasting insecticidal nets over three to 32 months of use in Ethiopia. Mal J. 2013; doi:10.1186/1475-2875-12-242 20. Michie S, Abraham C, Eccles MP, Francis JJ, Hardeman W, Johnston M (2011) Strengthening evaluation and implementation by specifying components of behaviour change interventions: a study protocol. Implement Sci 6,10. 21. Glanz K, Bishop DB (2010) The role of behavioral science theory in development and implementation of public health interventions. Ann Rev Public Health 31 , 399-418. 22. Painter JE, Borba CP, Hynes M, Mays D, Glanz K (2008) The use of theory in health behavior research from 2000-2005: a systematic review. Ann Behav Med 35 , 358-62. 23. Michie S, van Stralen MM, West R (2011) The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci 6, 42. 24. Kinzie MB (2005) Instructional design strategies for health behaviour change. Patient Ed Counsel 56 , 3-15. 25. Marsh VM, Mutemi W, Some ES, Haaland A, Snow RW (1996) Evaluating the community education programme of an insecticide-treated bed net trial on the Kenyan coast. Health Policy Plan, 11 (3), 280-291 26. Minja H, Schellenberg JA, Mukasa O, Nathan R, Abdulla S, Mponda H, Obrist B (2001) Introducing insecticide-treated nets in the Kilombero Valley, Tanzania: the relevance of local knowledge and practice for an information, education and communication (IEC) campaign. Trop Med Int Health, 6 (8), 614-623 27. Rhee M, Sissoko M, Perry S, McFarland W, Parsonnet J, Doumbo O. (2005) Use of insecticide- treated nets (ITNs) following a malaria education intervention in Piron, Mali: a control trial with systematic allocation of households. Malar J, 4 , 35. doi:10.1186/1475-2875-4-35 28. Rickard DG, Dudovitz RN, Wong MD, Jen HC, Osborn RD, Fernandez HE, Donkor CI (2011) Closing the gap between insecticide treated net ownership and use for the prevention of malaria. Prog Community Health Partnersh, 5 (2), 123-131. doi:10.1353/cpr.2011.0018 29. Ahmadi M, Vatandoost H, Shaeghi M, Raeisi A, Abedi F, Eshraghian M, et al . (2012) Effects of educational intervention on long-lasting insecticidal nets use in a malarious area, southeast Iran. Acta Med Iran, 50 (4), 279 ‐287 30. Amoran OE, Fatugase KO, Fatugase OM, Alausa KO (2012) Impact of health education intervention on insecticide treated nets uptake among nursing mothers in rural communities in Nigeria. BMC Research Notes 5:444 31. Alvarado BE, Gomez E, Serra M, Carvajal R, Carrasquilla G. (2006). Evaluación de una estrategia educativa en malaria aplicada en localidades rurales del Pacífico colombiano.

181

Biomedica, 26 (3), 342-352. 32. Nonaka D, Kobayashi J, Jimba M, Vilaysouk B, Tsukamoto K, Kano S, et al (2008) Malaria education from school to community in Oudomxay province, Lao PDR. Parasitol International 57 , 76-82 33. Ayi I, Nonaka D, Adjovu JK, Hanafusa S, Jimba M, Bosompem KM, et al. (2010) School-based participatory health education for malaria control in Ghana: engaging children as health messengers. Malar J 9,98. 34. Mfaume MS, Winch PJ, Makemba AM, Premji Z (1997). Mosques against malaria. World Health Forum, 18 (1), 35-38 35. Panter-Brick C, Clarke SE, Lomas H, Pinder M, Lindsay SW (2006) Culturally compelling strategies for behaviour change: a social ecology model and case study in malaria prevention. Soc Sci Med, 62 (11), 2810-2825. doi:10.1016/j.socscimed.2005.10.009 36. Harvey SA, Olórtegui MP, Leontsini E, Asayag CR, Scott K, Winch PJ. Trials of improved practices (TIPs): A strategy for making long-lasting nets last longer? Am J Trop Med Hyg. 2013; doi:10.4269/ajtmh.12-0641 37. Helinski MH, Namara G, Koenker H, Kilian A, Hunter G, Acosta A., et al. Impact of a behaviour change communication programme on net durability in eastern Uganda. Mal J. 2015; doi:10.1186/s12936-015-0899-5 38. Koenker H, Kilian A, Hunter G, Acosta A, Scandurra L, Fagbemi B, Lynch M (2015) Impact of a behaviour change intervention on long-lasting insecticidal net care and repair behaviour and net condition in Nasarawa State, Nigeria. Malar J, 14 , 18. doi:10.1186/s12936-014-0538-6 39. Widmar M, Nagel CJ, Ho DY, Benziger PW, Hennig N (2009) Determining and addressing obstacles to the effective use of long-lasting insecticide-impregnated nets in rural Tanzania. Malar J, 8 , 315. doi:10.1186/1475-2875-8-315 40. Ho D. (2012) An approach to achieve sustainable bed-net use in rural Tanzania. (1509713), Mount Sinai School of Medicine, Ann Arbor 41. Alonso PL, Lindsay SW, Armstrong Schellenberg JR, Gomez P, Hill AG, David PH, Greenwood BM (1993) A malaria control trial using insecticide-treated bed nets and targeted chemoprophylaxis in a rural area of The Gambia, west Africa. 2. Mortality and morbidity from malaria in the study area. Trans R Soc Trop Med Hyg, 87 Suppl 2 , 13-17. 42. Brieger W, Onyido A, Sexton J, Ezike V, Breman J, Ekanem O (1996) Monitoring community response to malaria control using insecticide-impregnated bed nets, curtains and residual spray at Nsukka, Nigeria. Health education research, 11 (2), 133 ‐145. Retrieved from https://www.cochranelibrary.com/central/doi/10.1002/central/CN-00183839/full 43. Khadka A, Perales NA, Wei DJ, Gage AD, Haber N, Verguet S, et al . (2018) Malaria control across borders: quasi-experimental evidence from the Trans-Kunene malaria initiative (TKMI). Malar J, 17 (1), 224. doi:10.1186/s12936-018-2368-4 44. Kachur SP, Phillips-Howard PA, Odhacha AM, Ruebush TK, Oloo AJ, Nahlen BL (1999) Maintenance and sustained use of insecticide-treated bednets and curtains three years after a controlled trial in western Kenya. Trop Med Int Health, 4 (11), 728-735. 45. Snow RW, McCabe E, Mbogo CN, Molyneux CS, Some ES, Mung'ala, VO, Nevill CG (1999) The effect of delivery mechanisms on the uptake of bed net re-impregnation in Kilifi District, Kenya. Health Policy Plan, 14 (1), 18-25. 46. Wacira DG, Hill J, McCall PJ, Kroeger A (2007) Delivery of insecticide-treated net services through employer and community-based approaches in Kenya. Tropical medicine & international health, 12 (1), 140-149. doi:10.1111/j.1365-3156.2006.01759.x 47. Fernando SD, Abeyasinghe RR, Galappaththy GN, Gunawardena N, Rajapakse LC (2008) Community factors affecting long-lasting impregnated mosquito net use for malaria control in Sri Lanka. Trans R Soc Trop Med Hyg, 102 (11), 1081-1088. doi:10.1016/j.trstmh.2008.06.007 48. Cham MK, D'Alessandro U, Todd J, Bennett S, Fegan G, Cham BA, Greenwood BM (1996) Implementing a nationwide insecticide-impregnated bednet programme in The Gambia. Health

182

Policy Plan, 11 (3), 292-298. 49. Hiwat H, Hardjopawiro LS, Takken W, Villegas L (2012) Novel strategies lead to pre-elimination of malaria in previously high-risk areas in Suriname, South America. Malar J, 11 , 10. doi:10.1186/1475-2875-11-10 50. Gerstl S, Dunkley S, Mukhtar A, Maes P, De Smet M, Baker S, Maikere J (2010) Long-lasting insecticide-treated net usage in eastern Sierra Leone - the success of free distribution. Trop Med Int Health, 15 (4), 480-488. doi:10.1111/j.1365-3156.2010.02478.x 51. Ahmed SM, Hossain S, Kabir MM, Roy S (2011) Free distribution of insecticidal bed nets improves possession and preferential use by households and is equitable: findings from two cross-sectional surveys in thirteen malaria endemic districts of Bangladesh. Malar J, 10 , 357. doi:10.1186/1475-2875-10-357 52. Alaii J, Van DBH, Kachur S, Mwenesi H, Vulule J, Hawley W, et al . (2003) Perceptions of bed nets and malaria prevention before and after a randomized controlled trial of permethrin-treated bed nets in western Kenya. American journal of tropical medicine and hygiene, 68 (4 Suppl), 142 ‐ 148 53. Against Malaria Foundation (2019). Distributions. < https://www.againstmalaria.com/Distributions.aspx?MapID=1&StatusID=7> Accessed 11 May 2019 54. Caviness C, Gish A. Five ways The United Methodist Church has impacted malaria. Imagine No Malaria. http://umc.org/how-we-serve/five-ways-the-united-methodist-church-has-impacted- malaria Accessed 14 March 2019 55. Malaria Consortium. Pioneer Project 2009-2014: a holistic systems strengthening approach towards malaria control in mid-western Uganda. Successes and lessons. 56. Malaria Consortium (2018) Nets bring good health: a qualitative inquiry into mosquito net use and care practices in two northern provinces of Mozambique. London, United Kingdom Accessed 14 March 2019 57. The RBM Partnership to End Malaria (2018) The strategic framework for malaria social and behavior change communication 2018-2030. Venier, Switzerland: RBM 58. Glanz K, Lewis FM, Rimer BK (1997) Linking theory, research, and practices. In: Glanz K, Lewis FM, Rimer BK, editors. Health behavior and health education. 2 nd ed. San Francisco: Jossey-Bass; 3-18 59. RBM Partnership to End Malaria (2019) The art of asking questions: priority research areas and approaches for malaria social and behavior change programs. Venier, Switzerland: RBM. 60. Hartjes LB (2010) The development and evaluation of a malaria risk-reduction game for travelers. (3448913), The University of Wisconsin - Madison, Ann Arbor. 61. Burgstahler S (2009) Universal design for instruction (UDI): definition, principles, guidelines, and examples. University of Washington 62. The Health Communication Capacity Collaborative (HC3). (2017) Malaria SBCC evidence literature review. Baltimore, Maryland: Johns Hopkins Center for Communication Programs 63. Thomas S (2004) What is participatory learning and action (PLA): an introduction. Centre for International Development and Training (CIDT); London 64. RBM Partnership to End Malaria (2017) Malaria social and behavior change communication indicator reference guide: second edition. Venier, Switzerland: RBM 65. RBM Partnership to End Malaria (2019) Developing M&E plans for malaria social and behavior change programs: a step-by-step guide. Venier, Switzerland: RBM

183

APPENDIX F . QUALITY SCORING

Quality scoring of articles with education intervention as a main activity Quality Criteria Good Quality (1 point) Poor Quality (0 points) 1 Was there a well defined [6,18,55,57,58,59,60,61,62,63, [56, 65] comparison group? 64,66,67,68] ,

2 Was the comparison group [6, [56,57,58,59,63,65] selected appropriately? 18,55,58,60,61,62,64,66,67,68] 3 Was there an explicit link to [6,18,55,56,61,64,65,66,67,68] [56,57,59,60,62,63] appropriate health behavior theory? (An appropriate theory is one that targets the appropriate level of audience of individual, community, children, or adults) 4 Was there an adequate [6,18,55,56,58,63,60,61,62,64, [57,59,67,68] description of educational 65,66] activities?

5 Were the educational activities [6,18,55,56,58,60,61,62,64,65, [57,59,63,67,68] equitable for diverse learners? 66] (Equitable activities include the implementation of multiple types of activities such as audio, visual, interpersonal, etc) 6 Was there hypothesis testing of [6,18,57,58,59,61,62,64,66,67, [55,56,63,65] the differences between groups 68] pertaining to net care knowledge and/or practices?

184

APPENDIX G: PROSPERO REGISTRATION

185

186

187

188

189

190

191

192

193

194

195

APPENDIX H: SEARCH STRATEGIES

Database search strategies completed 13 February 2019

Database Search Strategy Number of Records Bibliomap Individually searched terms and manually combined searches 4 CENTRAL (("mosquito nets" OR "LLIN" OR "LLINs" OR "Long lasting insecticide treated nets" 230 OR "Long lasting insecticide treated bednets" OR "Long lasting insecticide treated bed nets" OR "long lasting insecticidal nets" OR "long-lasting insecticide-treated nets" OR “ITN” OR "ITNs" OR “Insecticide treated nets” OR “Insecticide treated bednets” OR “Insecticide treated bed nets” OR bednets OR "bed nets" OR "Icon Life" OR Yorkool OR Olyset OR Netprotect OR PermaNet)) in All Text AND (malaria OR "malaria") in All Text AND (intervention) OR ("education intervention") OR ("malaria education") OR ("malaria intervention") OR ("education campaign") Embase ('bed net'/exp OR 'bed net' OR 'mosquito nets'/exp OR 'mosquito nets' OR 'llin' OR 1655 'llins' OR 'long lasting insecticide treated nets' OR 'long lasting insecticide treated bednets' OR 'long lasting insecticide treated bed nets' OR 'long lasting insecticidal nets' OR 'long-lasting insecticide-treated nets' OR 'itn' OR 'itns' OR 'insecticide treated nets'/exp OR 'insecticide treated nets' OR 'insecticide treated bednets'/exp OR 'insecticide treated bednets' OR 'insecticide treated bed nets' OR 'bednets'/exp OR bednets OR 'bed nets'/exp OR 'bed nets' OR 'icon life'/exp OR 'icon life' OR 'yorkool'/exp OR yorkool OR 'netprotect'/exp OR netprotect OR 'olyset'/exp OR olyset OR 'permanet'/exp OR permanent) AND malaria AND ('health education'/exp OR 'health behavior'/exp OR intervention OR 'education intervention' OR 'malaria education' OR 'malaria intervention' OR 'education campaign' OR 'behavior change' OR 'behaviour change') ProQuest ("mosquito nets" OR "LLIN" OR "LLINs" OR "Long lasting insecticide treated nets" 1946 Dissertations & OR "Long lasting insecticide treated bednets" OR "Long lasting insecticide treated bed Theses nets" OR "long lasting insecticidal nets" OR "long-lasting insecticide-treated nets" OR "ITN" OR "ITNs" OR "Insecticide treated nets" OR "Insecticide treated bednets" OR "Insecticide treated bed nets" OR bednets OR "bed nets" OR "Icon Life" OR Yorkool OR Olyset OR Netprotect OR PermaNet) AND malaria AND (“intervention” OR “education intervention” OR “malaria education” OR “malaria intervention” OR “education campaign” OR “behavior change” OR "behaviour change”) Pubmed (((((“mosquito nets”[Mesh] OR "mosquito nets" OR "LLIN" OR "LLINs" OR "Long 873 lasting insecticide treated nets" OR "Long lasting insecticide treated bednets" OR "Long lasting insecticide treated bed nets" OR "long lasting insecticidal nets" OR "long-lasting insecticide-treated nets" OR “ITN” OR "ITNs" OR “Insecticide treated nets” OR “Insecticide treated bednets” OR “Insecticide treated bed nets” OR bednets OR "bed nets" OR "Icon Life" OR Yorkool OR Olyset OR Netprotect OR PermaNet)))) AND ((malaria) OR “malaria”[Mesh])) AND ((“health education”[Mesh] OR “Health Behavior”[Mesh] OR “intervention” OR “education intervention” OR “malaria education” OR “malaria intervention” OR “education campaign” OR “behavior change” OR "behaviour change”)) TRoPHI Individually searched terms and manually combined searches 24 Web of (“mosquito nets” OR "LLIN" OR "LLINs" OR "Long lasting insecticide treated nets" 1775 Science OR "Long lasting insecticide treated bednets" OR "Long lasting insecticide treated bed nets" OR "long lasting insecticidal nets" OR "long-lasting insecticide-treated nets" OR “ITN” OR "ITNs" OR “Insecticide treated nets” OR “Insecticide treated bednets” OR “Insecticide treated bed nets” OR bednets OR "bed nets" OR "Icon Life" OR Yorkool OR Olyset OR Netprotect OR PermaNet) AND malaria AND intervention OR “education intervention” OR “malaria education” OR “malaria intervention” OR “education campaign” OR “behavior change” OR "behaviour change”

196