Abstract
This Campbell systematic review assesses the effectiveness of economic and educational strategies for ownership and appropriate use of insecticide-treated bednets (‘ITN’) in developing countries. The review also examines whether changes in ITN ownership and use affect malaria-specific morbidity rates. The review summarises findings from 10 studies, nine of which were conducted in rural Africa and one in rural India.
Compared to providing ITNS at full market or a subsidized price, giving away ITNs for free increases the number of people owning an ITN. However, the provision of free ITNs increases their use only slightly or not at all.
Providing education in the appropriate use of ITNs increases the number of people sleeping under bednets compared to a control group which didn't receive the education.
Combining these strategies with unspecified incentives does not increase ITN ownership, leading to little or no differences in their appropriate use. Embedding the promotion of ITNs within specific health- or finance-focused marketing messages only leads to small or no differences in bednet ownership and use.
There is some evidence of improved malaria-specific morbidities among children and adults as a result of increased ITN ownership and use. However, the evidence supporting this finding is of low certainty and should be interpreted with caution.
Plain language summary
STRATEGIES TO INCREASE THE OWNERSHIP AND USE OF INSECTICIDE-TREATED BEDNETS TO PREVENT MALARIA
Researchers in the Cochrane and Campbell Collaborations conducted a review of the effect of different strategies to increase people's ownership and use of insecticide-treated bednets to prevent malaria. In February 2013 they identified 10 relevant studies. Their findings are summarized below.
Malaria is a life-threatening disease caused by a parasite. A person becomes infected from being bitten by a mosquito carrying the parasite. People with malaria may have symptoms such as chills, fever, vomiting, diarrhea, and jaundice. About 40 percent of the world's population is at risk of malaria, mostly in the poorest countries.
Insecticide-treated bednets fit over a bed and act as a barrier between insects and the person sleeping. The bednets are dipped in insecticide, a chemical that kills or repels mosquitoes, and effectively prevent malaria. Insecticide-treated bednets cost money and it is important to find ways of ensuring that people who need them own them. Even when people own bednets they may not always use them properly. To be effective, bednets need to be used every night. They must also hang properly and be treated with enough insecticide. It is important to measure whether different strategies not only increase people's ownership but also people's use of bednets.
The studies included in this review took place in Africa and India. In five of the studies, people were either given insecticide-treated bednets free, or could buy them at a subsidized price or full market price. In the other five studies, people were educated about how to use the bednets properly, for instance through visits at home or through information on the radio, on television and in the community. The included studies show the following:
Providing free insecticide-treated bednets: Probably increases the number of people who own bednets compared to providing subsidized bednets or bednets offered at full market price. Probably leads to little or no difference in the use of bednets compared to providing subsidized bednets or bednets offered at full market price.
Providing education for the appropriate use of insecticide- treated bednets: May increase the number of adults and children under five using bednets (sleeping under bednets).
Providing incentives to encourage use of insecticide-treated bednets: Probably leads to little or no difference in ownership or use of bednets compared to those who did not receive an incentive.
A possible side effect when providing free or subsidized insecticide-treated bednets may be that the governments and institutions who pay for the bednets take this money from other priority issues. However, none of the included studies measured whether these or any other side effects had occurred.
Abstract
BACKGROUND
Malaria is a life-threatening parasitic disease and 40 percent of the world's population lives in areas affected by malaria. Insecticide-treated bednets (ITNs) effectively prevent malaria, however, barriers to their use have been identified.
OBJECTIVES
To assess the evidence on the effectiveness of available strategies that focus on delivery and appropriate use of ITNs.
SEARCH METHODS
We searched the EPOC Register of Studies, CENTRAL, MEDLINE, EMBASE, HealthStar, CINAHL, PubMed, Science Citation Index, ProQuest Dissertations and Theses, African Index Medicus (AIM), World Health Organization Library and Information Networks for Knowledge (WHOLIS), LILACS, Virtual Health Library (VHL), and the World Health Organization Library Information System (WHOLIS). Initial searches were conducted in May 2011, updated in March 2012 and February 2013. Authors contacted organizations and individuals involved in ITN distribution programs or research to identify current initiatives, studies or unpublished data, and searched reference lists of relevant reviews and studies.
SELECTION CRITERIA
Randomized controlled trials (RCTs), non-randomized controlled trials, controlled before-after (CBA) studies, and interrupted time series evaluating interventions focused on increasing ITN ownership and use were considered. The populations of interest were individuals in malaria-endemic areas.
DATA COLLECTION AND ANALYSIS
Two authors independently screened studies to be included. They extracted data from the selected studies and assessed the risk of bias. For RCTs, we used the Cochrane Collaboration's ‘Risk of bias’ tool and we used the risk of bias criteria suggested by EPOC for other study designs. When consensus was not reached, any disagreements were discussed with a third author. The magnitude of effect and quality of evidence for each outcome was assessed.
MAIN RESULTS
Of the 3,032 possibly relevant records identified, 10 studies were included in this review; eight cluster RCTs, one RCT, and one CBA study. Overall, three studies were assessed as having moderate risk of bias and seven studies were assessed as high risk of bias.
Effect of ITN cost on ownership:
Four studies including 4,566 households and another study comprising 424 participants evaluated the effect of ITN price on ownership. These studies suggest that providing free ITNs probably increases ITN ownership when compared to subsidized ITNs or ITNs offered at full market price.
Pooled data for two studies suggested that receiving an ITN at no cost probably increases ITN ownership, compared to purchasing an ITN at the market price (SMD 0.69, 95% confidence interval (CI) 0.39 to 0.99, moderate-certainty evidence) or purchasing an ITN using a loan (SMD 0.37, 95% CI 0.27 to 0.47, moderate-certainty evidence). There is probably little to no difference in ITN ownership when comparing groups in which ITNs were purchased using a loan and those in which ITNs were purchased at the market price (SMD 0.29, 95% CI -0.06 to 0.63, moderate-certainty evidence).
Effect of ITN cost on appropriate use of ITNs:
Three studies including 9,968 households and another study comprising 259 individuals found that there is probably little or no difference in the use of ITNs when they are provided free, compared to providing subsidized ITNs or ITNs offered at full market price.
Effect of education interventions on use of ITNs:
Five studies, including 12,637 households, assessed educational interventions regarding ITN use and concluded that education may increase the number of adults and children using ITNs (sleeping under ITNs) compared to no education.
One study, including 519 households, assessed the effects of providing an incentive (an undisclosed prize) to promote ITN ownership and use, and found that incentives probably lead to little or no difference in ownership or use of ITNs, compared to not receiving an incentive.
None of the included studies reported on adverse effects.
IMPLICATIONS FOR POLICY AND RESEARCH
Five studies examined the effect of price on ITN ownership and found moderate-certainty evidence that ownership was highest among the groups who received the ITN free versus those who purchased the ITN at any cost. In other words, demand for ITNs is elastic with regard to price and hence subsidies (providing full or partial cost recovery) will probably increase ownership. However, once the ITN is supplied, the price paid for the ITN probably has little to no effect on its use; the four studies addressing this outcome failed to confirm the hypothesis that people who purchase nets will use them more than those who receive them at no cost. Educational interventions for promoting ITN use have an additional positive effect. In other words, the evidence suggests that ITN use is highly inelastic with respect to price, and therefore additional encouragement (such as through education) is probably needed to promote appropriate use, over and above any subsidies to increase ownership. However, the impact of different types or intensities of education is unknown.
Laienverständliche Zusammenfassung
Strategien zur Erhöhung des Besitzes und Gebrauchs von mit Insektiziden behandelten Bettnetzen zur Prävention von Malaria
Forscher der Cochrane Collaboration führten einen Review durch, der die Wirksamkeit verschiedener Malaria-Präventions-Strategien untersucht, die den Besitz und Gebrauch von Bettnetzen, die mit Insektiziden behandelt werden, erhöhen sollen. Im Februar 2013 identifizierten sie 10 relevante Studien. Ihre Ergebnisse sind unten zusammengefasst.
Malaria ist eine lebensgefährliche Erkrankung, die durch einen Parasiten verursacht wird. Eine Person infiziert sich indem sie von einem Moskito gestochen wird, der diesen Parasiten in sich trägt. Menschen mit Malaria können Symptome wie Schüttelfrost, Fieber, Erbrechen, Durchfall und Gelbsucht haben. Bei circa 40% der Weltbevölkerung, besonders in den Ärmsten Ländern, besteht die Gefahr, sich mit Malaria zu infizieren.
Die mit Insektiziden behandelten Bettnetze, die über dem Bett angebracht werden, dienen als Schranke zwischen Insekten und der schlafenden Person. Die Bettnetze werden in Insektizide, einem chemischen Wirkstoff, der Mosquitos tötet oder abschreckt, getaucht. Auf dieser Weise verhindern sie effektiv die Übertragung von Malaria. Diese mit Insektiziden behandelten Bettnetze sind teuer und es ist wichtig, Möglichkeiten zu finden, dass die Menschen, die die Netze brauchen, diese auch besitzen. Selbst wenn Menschen Bettnetze besitzen, kann es aber sein, dass sie diese nicht korrekt verwenden. Damit sie wirksam sind, müssen die Bettnetze jede Nacht verwendet werden. Zudem müssen sie richtig befestigt sein und mit ausreichend Insektiziden behandelt werden. Es ist somit wichtig zu untersuchen, ob unterschiedliche Strategien nicht nur den Besitz sondern auch den Gebrauch von Bettnetzen erhöhen.
Die Studien, die in diesem Review eingeschlossen wurden, wurden in Afrika und Indien durchgeführt. In fünf dieser Studien wurden den Menschen mit Insektiziden behandelte Bettnetze entweder kostenlos zur Verfügung gestellt, oder sie konnten diese zu einem vergünstigten oder vollen Preis erwerben. In den anderen fünf Studien wurden die Menschen darüber aufgeklärt, wie die Bettnetze korrekt verwendet werden, zum Beispiel durch Hausbesuche oder durch Informationen über das Radio, das Fernsehen oder in der Gemeinschaft. Die eingeschlossenen Studien zeigten folgendes:
Versorgung mit kostenlosen mit Insektiziden behandelten Bettnetzen: Erhöht wahrscheinlich die Zahl der Menschen, die ein Bettnetz besitzen im Vergleich zu jenen, die vergünstigte Bettnetze oder solche zum vollen Marktpreis angeboten wurden. Führt vermutlich nur zu einem geringen bis keinem Unterschied im Gebrauch von Bettnetzen im Vergleich zur Versorgung mit vergünstigten Bettnetzen oder solchen, die zum vollen Marktpreis angeboten werden.
Anbieten von Schulungen zum korrekten Gebrauch der mit Insektiziden behandelten Bettnetze: Erhöht vielleicht die Zahl der Erwachsenen und Kinder unter fünf Jahren, die Bettnetze verwenden (die unter Bettnetzen schlafen).
Anreize, um den Gebrauch von mit Insektiziden behandelten Bettnetzen zu erhöhen: Vermutlich führt dies zu einem geringen oder keinem Unterschied im Besitz und Gebrauch von Bettnetzen im Vergleich zu solchen, die keine Anreize bekommen haben.
Ein möglicher Nebeneffekt von kostenlosen oder vergünstigten, mit Insektiziden behandelten Bettnetzen mag sein, dass Regierungen und Institutionen, die für diese Bettnetze bezahlen, dieses Geld von anderen wichtigen Themen wegnehmen. Jedoch hat keine der Studien gemessen, ob diese oder andere Nebeneffekte tatsächlich auftraten.
Anmerkungen zur Übersetzung
Cochrane Schweiz
எளியம ொழிச் சுருக்கம்
மலேரியாவைத் தடுக்க பூச்சிக்ககால்லிகளால் பதப்படுத்தப்பட்ட ககாசு ைவேககவளச் க ாந்தமாக்கிப் பயன்படுத்துைவத அதிகரிக்க உதவும் உத்திகள்
காக்லேன் ஒத்துவைப்பு ஆோய்ச்சியாளர்கள் மலேரியாவைத் தடுப்பதற்கான பூச்சிக்ககால்லிகளால் பதப்படுத்தப்பட்ட ககாசுைவேகவள மக்கள் தமதுவடவம யாக்குதற்கும், பயன்பாட்வட அதிகரிப்பதற்கும் லமற்ககாண்ட பல்லைறு வியூகங்களின் விவளவுகவள மதிப்பீடு க ய்ய ஒரு திறனாய்வு நடத்தினர். பிப்ேைரி 2013 இல் அைர்கள் இதலனாடு கதாடர்புவடய 10 ஆய்வுகவள அைர்கள் கண்டறிந்தனர். இைர்களின் கண்டுபிடிப்புகளின் சுருக்கம் பின்ைருமாறு.
மலேரியா என்றால் என்ன? மற்றும் பூச்சிக்ககால்லிகளால் பதப்படுத்தப் பட்ட ககாசு ைவேகளால் எவ்ைாறு அவதத் தடுக்க முடியும்?
மலேரியா என்பது ஒரு குறிப்பிட்ட ஒட்டுண்ணியால் ஏற்படுத்தப்படும் ைாழ்க்வகவய அச்சுறுத்தும் ஒரு லநாய். இந்த ஒட்டுண்ணிவய எடுத்துச் க ல்லும் ஒரு ககாசு கடிப்பதினால் ஒருநபருக்கு லநாய்த்கதாற்று ஏற்படுகிறது. மலேரியாவினால் பாதிக்கப்பட்டைர்களுக்கு குளிர் காய்ச் ல், காய்ச் ல், ைாந்தி, ையிற்றுப்லபாக்கு, மஞ் ள் காமாவே லபான்ற அறிகுறிகள் இருக்கோம். உேக மக்கள் கதாவகயில் சுமார் 40% லபர் மலேரியா ஆபத்துக்கு உள்ளாகி உள்ளனர். இைர்கள் கபரும்பாலும் ைறிய நாடுகளில் ைசிக்கின்றனர்.
பபூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகள் படுக்வகயின் மீது கபாருத்தப்பட்டு, தூங்குலைாருக்கும் ககாசுவுக்கும் இவடலய ஒரு லைலியாக இருக்கும். திறன்பட மலேரியாவைத் தடுக்க, ககாசு ைவேகள் ககாசுக்கவளக் ககால்லும் அல்ேது துேத்தும் ே ாயனத்தில் முக்கப்படுகிறது. பூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகள் அதிக விவே யுவடயதால் லதவையானைர்கள் அைற்வற க ாந்தமாகப் கபறுைவத உறுதிப் படுத்துைதற்கான ைழிகவளக் கண்டறிைது முக்கியமாகும். க ாந்தமாக ககாசு ைவேகள் இருந்தலபாதும் மக்கள் அைற்வற ரியாக பயன்படுத்துைது இல்வே. ககாசு ைவேகள் பயனுள்ளதாக இருக்க லைண்டும் என்றால் ஓவ்கைாரு இேவும் பயன்படுத்தப்பட லைண்டும். அைற்வற முவறயாகத் கதாங்கவிட லைண்டும் மற்றும் லபாதிய அளவு பூச்சிக்ககால்லி ககாண்டு பதனிடப்பட லைண்டும். கைவ்லைறு அணுகுமுவறகளால் மக்கள் ககாசுைவேகவள க ாந்தமாகப் வைத்திருப்பது மட்டுமில்ோமல் அைற்வறச் ரியாகப் பயன்படுத்துதலும் அதிகரிக்கிறதா என்பவத அளவிடுைது மிக முக்கியமாகும்.
இந்த ஆோய்ச்சி என்ன கூறுகிறது:
இந்த திறனாய்வில் ல ர்க்கப்பட்ட ஆய்வுகள் ஆப்பிரிக்கா மற்றும் இந்தியாவில் க ய்யப்பட்டது. ஐந்து ஆய்வுகளில், பூச்சிக்ககால்லி சிகிச்வ யளிக்கப்பட்ட ககாசு ைவேகள் மக்களுக்கு இேை மாக ைைங்கப் பட்டது அல்ேது மலிைான விவேயில் ைாங்கலைா அல்ேது முழு ந்வத விவேயில் ைாங்கலைா ைவக க ய்யப்பட்டது. ஐந்து ஆய்வுகளில்அைற்வற எவ்ைாறு ரியாகப் பயன்படுத்துைது என்பதற்கான கல்வி அளிக்கப்பட்டது. உதாேணமாக வீடுகளுக்கும் லநரில் க ன்று பார்வை இடுைது அல்ேது ைாகனாலி, கதாவேக்காட்சி மூேமும் முதாயம் மூேமும் க ய்திகவளப் பேப்புைது. ல ர்க்கப்பட்டுள்ள ஆய்வுகள் பின்ைருைவதக் காட்டுகிறது:
பூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகவள இேை மாக ைைங்குதல்: மானியத்துடன் விற்கப்படும் ககாசு ைவேகள் அல்ேது முழு ந்வத விவேயில் விற்கப்படும் ககாசு ைவேகள் இைற்லறாடு ஒப்பிடும்லபாது இேை மாக ைைங்கும் ககாசு ைவேகவள க ாந்தமான க ாந்தமாக வைத்திருப்லபார் எண்ணிக்வக அதிகரிக்கோம். மானியதுடன் விற்கப்படும் ககாசு ைவேகள், முழு ந்வத விவேயில் விற்கப்படும் ககாசு ைவேகவள ஒப்பிடும்லபாது மக்கள் ககாசு ைவேகள் உலபாகித்தலில் சிறிய மாற்றம் அல்ேது ஒரு மாற்றமும் இோது.
பூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகள் உலபாகித்தல் பற்றிய கல்வி ைைங்குதல்: கபரியைர்கள் மற்றும் ஐந்து ையதுக்கு உட்பட்ட குைந்வதகள் ககாசு ைவேகள் பயன்படுத்துதல் அதிகரிக்கோம் (ககாசு ைவேக்குள் தூங்குைது)
பூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகள் பயன்படுத்துதவே ஊக்குவித்தல்: ககாசு ைவேகவள க ாந்தமாக்குைது மற்றும் பயன்படுத்துதல் ஊக்கத்கதாவகத் கபறாதைர்கவள ஒப்பிடும்லபாது அலனகமாக சிறிய மாற்றம் அல்ேது ஒரு மாற்றமும் இோது.
பூச்சிக்ககால்லியால் பதப்படுத்தப்பட்ட ககாசு ைவேகவள இேை மாகலைா மானியவிவேயிலோ ைைங்கும் லபாது பக்க விவளைாக அேசு மற்றும் இந்த ககாசுைவேகளுக்கு நிதி ைைங்கும் நிறுைனங்கள் இந்த நிதியிவன மற்ற முன்னுரிவம இனங்களுக்கு எடுத்துக்ககாள்ளும் சிக்கலும் ஏற்படோம். இருப்பினும் இந்த திறனாய்விற்கு உட்படுத்தப்பட்ட எந்த ஆய்வுகளும் இத்தவகய அல்ேது லைறுவிதமான பக்க விவளவுகள் எற்பட்டதா என அளவிடவில்வே.
கமாழிகபயர்ப்பு குறிப்புகள்
கமாழிகபயர்ப்பு: க. ஹரிஓம், தங்கசுைாமி மற்றும் சி.இ.பி.என்.அர். குழு
Summary of findings table
Footnotes
One study provided free ITNs to pregnant women, subsidized ITNs and social marketing to the community and compared this to control communities with access to subsidized ITNs and social marketing (Muller 2008). The second study provided women attending prenatal clinics with free ITNs, KES 10, KES 20, or KES 40 (Cohen 2010).
The studies included in this outcome are Cohen 2010, Dupas 2009, Fink 2012, Muller 2008, and Tarozzi 2011. Cohen 2010 included 424 participants and the remainder of the studies reported the number of households (4666 households).
Baseline imbalances for most variables in the Muller 2008 study and some outcomes mentioned in the protocol were not reported in the final study. The Cohen 2010 study has small sample sizes in each of the 4 groups. Unclear ROB. Baseline characteristics of study participants not described and some mismanagement (leakage) of ITNs. The Dupas 2009 study had small samples sizes in each price group and unclear ROB. The Tarozzi 2011 study had unclear method of randomization and the intervention and control clients may have different characteristics. Allocation concealment was not described in Fink 2012. Some baseline differences between groups (fewer sleeping spaces in the control group, fewer ITNs at baseline in the free group).
Dupas 2009 randomly assigned households to a subsidy level for ITNs (40% to 100%). ITN cost ranged from free to KES 300 (USD 4.62). Households received a voucher to obtain their net at the randomly assigned price. They were also assigned a “marketing” group (health vs. financial framing vs. none), a “commitment” group, and a “targeting” group (female head of household, male head, or both) (Dupas 2009).
The included studies are Cohen 2010, Dupas 2009, Fink 2012, and Tarozzi 2011. Cohen included 259 participants and the remaining studies included 9968 households.
Downgraded because of inconsistency.
Included studies are Amoran 2012 and Krezanoski 2010. The latter compared education with incentives vs no education or incentives. The number of participants reflects the number of households since these studies measured use at the household level.
Baseline imbalances in the Amoran 2012 study (knowledge about appropriate ITN mounting was higher in the intervention group at baseline).
Included studies are De La Cruz 2009, Deribew 2012, and Keating 2012. The number of participants reflects the number of households since these studies measured use at the household level.
Baseline imbalances in all three studies in which ITN use was higher in the intervention groups. Inconsistency between studies (De La Cruz 2009; Deribew 2012; Keating 2012). Deribew 2012 is a CBA.
The included study is Krezanoski 2010.
Single study with small sample size (Krezanoski 2010).
1 Background
Malaria is a life-threatening parasitic disease transmitted by female Anopheles mosquitoes. Malaria has the highest impact on young children who have not yet developed acquired immunity. In pregnant women, malaria is associated with increased risk of severe anemia, low birth weight, as well as with an increase in miscarriages and maternal deaths (Desai 2007; WHO 2010b). According to the World Health Organization (WHO) World Malaria Report 2014, there were 198 million cases of malaria and 584,000 deaths in 2013, of which 78 percent were among children under five years old (WHO 2014). Approximately 40 percent of the world's population is at risk of malaria; those at risk are primarily in the world's poorest countries.
Malaria causes a significant economic burden at both the individual and the regional level; it has an enormous and long-lasting effect on economic growth and development. Malaria contributes to localized differences in gross domestic product (GDP) between countries with and without malaria, especially in Africa. The WHO states that in some countries personal and public spending on costs related to malaria accounts for up to “40% of public health expenditures, 30% to 50% of inpatient hospital admissions, [and] up to 60% of outpatient health clinic visits.” The costs of malaria are borne disproportionately by those who can afford them the least: the poor and those who have limited access to health services (WHO 2010b).
The increase in international funding for fighting malaria has resulted in better access to malaria-preventive measures (WHO 2010c), including insecticide-treated bednets (ITNs) (Lengeler 2004); and indoor residual spraying and screens (Morel 2005; Pluess 2010). ITNs can reduce malaria mortality by 55 percent for children under 5 in sub-Saharan Africa (WHO 2014). However, the proportion of children under five years and pregnant women sleeping under ITNs is much lower than the coverage rates (Alaii 2003; Eisele 2009a). According to surveys conducted in Africa between 1999 and 2004, the median proportion of children under five sleeping under ITNs was only 3 percent. In countries with subsidized or free-of-charge ITN distribution, use has been scaled up successfully (WHO 2014). By 2013, ITN ownership increased in Africa: 49 percent of households owned at least one ITN and 44 percent of children slept under one (WHO 2014). Increased funding contributed to the rapid rise in the number of ITNs procured and distributed within countries. For example, between 2004 and 2006 the Global Fund increased ITN funding from USD 1.35 million to USD 18 million and between 2004 and 2007 the United Nations Children's Fund (UNICEF) increased its procurement from 7 million ITNs to nearly 20 million (UNICEF 2007). The number of ITNs delivered by manufacturers in sub-Saharan Africa increased from 11 million in 2005 to 392 million in 2013. This suggests that ITN use was scaled up effectively in countries with subsidized or free ITNs (WHO 2014). Globally, 83 countries with ongoing malaria transmission provide free ITNs to all age groups (WHO 2014).
One of the barriers to the effective ownership and use of ITNs is the associated cost. For example, one Kenyan study found that although rural residents wanted to own and use ITNs, they were prohibitively expensive (Guyatt 2002). It was estimated that ITNs for an entire household would cost about the same as paying for three children to attend one year of primary school (Guyatt 2002). Another barrier to ITN use is that people are often unfamiliar with ITNs, or do not have a habit of using them, so they need to be convinced of their usefulness and persuaded to use them on a regular basis (WHO 2010a). The culture of ITN use is more developed among some ethnic groups and this has a significant impact on ITN coverage. Wealth, living in an urban rather than rural area and higher levels of education are other important factors positively associated with ITN ownership and use, and their impacts on coverage need to be assessed (Belay 2008; Eisele 2009b; Monasch 2004).
1.1 DESCRIPTION OF THE INTERVENTION
WHO defines an insecticide-treated bednet (ITN) as a “mosquito net that repels, disables and/or kills mosquitoes coming into contact with insecticide on the netting material” (WHO 2007). Appropriate use of ITNs requires that they are hanging correctly and that they are used consistently (Eisele 2009b).
Strategies to increase ownership and use of ITNs include social marketing, health education campaigns by multidisciplinary teams, developing a ‘net culture’ through promotion and publicity, increased availability (e.g. local production of high-quality ITNs), free ITN distribution campaigns and cost reduction (e.g. reduced taxes imposed on ITNs) (WHO 2010a).
In the preceding decade, there has been much controversy over how ITNs should be distributed. This controversy can be roughly summarized as social marketing of subsidized ITNs versus free distribution of ITNs (Curtis 2003; Lines 2003). Social marketing of ITNs applies methods of commercial marketing, such as advertising, and includes the provision of ITNs through voucher schemes or direct sales in which the purchaser pays a portion of the product's cost, with public sources financing the difference (Armstrong Schellenberg 1999). Free distribution entails the provision of free ITNs at selected points of care, such as antenatal clinics, and mass campaigns of regional or national provision, usually in tandem with other health interventions (Lengeler 2007). Many, probably most, of these ITNs were freely distributed and many were piggybacked on other public health campaigns, such as immunization campaigns (Grabowsky 2005; Grabowsky 2007).
1.2 WHY IT IS IMPORTANT TO DO THIS REVIEW
While mass distribution of ITNs is necessary, it is not sufficient to ensure mass use of ITNs. Therefore, it is important to identify strategies that will both increase the ownership of ITNs and encourage appropriate use. This review assesses current evidence on strategies to increase ownership and use of ITNs and addresses a global health issue that is a core part of the United Nations' Millennium Development Goal (MDG) number six, which aims to ‘Combat HIV/AIDS, Malaria and other diseases’ (UN 2008). This review can help policy-makers and practitioners make appropriate and evidence-based decisions regarding ITN programs to help roll back malaria by identifying effective strategies to increase the ownership and appropriate use of ITNs, and the contexts and populations in which these strategies are effective.
2 Objectives
To assess the evidence on the effectiveness of available strategies (subsidies and education) that focus on delivery and appropriate use of ITNs.
3 Methods
The study methodology was published as a protocol (Augustincic Polec 2011).
3.1 CRITERIA FOR CONSIDERING STUDIES FOR THIS REVIEW
3.1.1 Types of studies
In this review, we included studies that reported on strategies that may increase ownership and the appropriate use of ITNs. Evaluating these interventions using randomized controlled trials (RCTs) may be challenging since scale-up strategies are delivered at a population level and are the standard of care in at-risk populations (WHO 2010c). Further, achieving unbiased randomization is difficult for a vector-borne disease where the vector's ecology and disease transmission can be different between comparators even when there is geographic displacement of short distances (under a kilometre). Thus, we also considered non-randomized trials in our review.
We considered any study design meeting the study design criteria from the Cochrane Effective Practice and Organisation of Care Group (EPOC 2014a). Studies considered were: RCTs; cluster RCTs; non-randomized controlled trials; controlled before-after studies (CBAs) with contemporaneous data collection and with two or more control and intervention sites; and interrupted time series studies (ITSs) with a clearly defined point in time when the intervention occurred and at least three data points before and after implementation of the intervention.
3.1.2 Types of participants
We included children and adults with permanent residence in malarious areas. Participants could include individuals or groups of individuals (e.g. families, households). We excluded military populations, travelers, students, those living in transient refugee camps for less than one malaria season, and others not permanently residing in the study area. These groups were excluded because their characteristics differ from the characteristics of populations that permanently reside in these areas since they may have greater access to ITNs and other malaria-preventive measures. For example, in addition to ITNs, military populations may have access to mosquito repellents and antimalarial drugs that are not available to ordinary residents (Ockenhouse 2005).
3.1.3 Types of interventions
Our systematic review included both unifaceted and multifaceted interventions that aimed to increase the ownership and appropriate use of ITNs. Unifaceted interventions are interventions with only one component (e.g. ITN distribution only), whereas multifaceted interventions consist of two or more components (e.g. free ITNs bundled with education about appropriate ITN use). To clarify the relationship between the interventions and the outcomes (Tugwell 2010), we developed a logic model (Figure 1). Interventions are grouped into three main categories depending on their focus: increasing ITN ownership (through delivery strategies), appropriate use of ITNs, and combinations of these.
Logic model: strategies to increase the use of insecticide-treated nets in households and vulnerable populations to reduce morbidity and mortality from malaria in endemic settings
ITN Ownership
The framework by Kilian 2010 was adapted and incorporated into our logic model (Figure 1 and Figure 2). Kilian's framework is a descriptive categorization of the various components of ITN delivery strategies intended to increase ITN ownership. Categories that are defined by this framework include ITN delivery channel, duration of distribution, cost to user, choice of ITN in terms of net preference and accessibility, and sectors involved in the distribution (Kilian 2010).
Framework of components of interventions to increase the ownership and use of ITNs
Appropriate ITN Use
Interventions focusing on appropriate ITN-use strategies were categorized using a logic model that builds on Kilian's framework (Figure 1 and Figure 2) (Kilian 2012). Examples of interventions to increase use are campaigns to increase ITN hang-up and volunteer home visits to educate populations about the appropriate use of ITNs. Interventions to encourage appropriate use can be delivered through different channels (community, outreach, routine services and at retail points). These interventions can be provided during limited periods of time (single or repeated) or they can be ongoing (continuous). Appropriate-use interventions can be delivered by the public sector (e.g. government) or private sector (civil society; non-governmental organizations (NGOs); faith- and community-based organizations; and the commercial sector). Interventions are further categorized as those focusing on education, peer monitoring (e.g. volunteer home visits), or publicity (e.g. media). We reported relevant context information that may affect the appropriate use of ITNs (e.g. national/regional culture, ITN stock-out periods, nomadic lifestyle).
Comparison groups included no intervention and other strategies aiming to increase ITN use (e.g. comparing two different interventions that are aiming to increase ITN use). We excluded studies focusing solely on the effectiveness of ITNs. Excluded studies informed the background and discussion sections of the review.
3.1.4 Types of outcome measures
Primary outcomes
ITN Ownership
Proportion of households with at least one ITN. Proportion of households with people/ITN ratio ≤ 2.0 among households with any bednets.
Appropriate ITN Use
Proportion of existing ITNs used. Proportion of population sleeping under ITNs. Proportion of pregnant women sleeping under ITNs. Proportion of children under five years sleeping under ITNs. Proportion of households with all children under five years sleeping under ITNs.
We included measures of ITN use during the previous night and measures of ITN use for which the timeframe of use was not reported.
Secondary outcomes
Equity ratio of household ownership with ITNs calculated across household income. Child all-cause mortality. Child malaria-specific mortality. Child malaria morbidity. All-cause mortality. Malaria-specific mortality. Malaria-specific morbidity. Anemia in pregnant women. Low birth weight.
We included these secondary outcomes since ownership and appropriate use of ITNs will result in measurable differences in these clinical outcomes, as demonstrated in a previous systematic review (Lengeler 2004). Our outcome measures are congruent with the Roll Back Malaria Monitoring & Evaluation Reference Group (MERG) minimum standards recommended for assessing malaria impact in countries in sub-Saharan Africa (Kilian 2011; Roll Back Malaria Partnership 2008).
We planned to record any adverse or any unintended effects reported (e.g. unable to afford other necessities due to money used by households to purchase ITNs).
3.2 SEARCH METHODS FOR IDENTIFICATION OF STUDIES
The initial search was conducted in May 2011. Updates of the initial search were conducted in March 2012 and February 2013. Search results are reported in a PRISMA flow diagram (Figure 3) (Moher 2009).
The EPOC Trials Search Coordinator developed search strategies (Appendix 1) in consultation with the authors. They comprise keywords and controlled vocabulary terms for malaria or parasites causing malaria in humans (Plasmodium falciparum/malariae/ovale/vivax and insecticide-treated bednets or mosquito nets). Language limits were not applied. Databases were typically searched from database start date forward; dates of database coverage are provided in the list of databases below. Two methodological search filters were used to limit retrieval to appropriate study design and interventions of interest: the Cochrane RCT Sensitivity/Precision Maximizing Filter (cf. the Cochrane Handbook for Systematic Reviews of Interventions 6.4d); and an EPOC Filter to identify study designs including CBA, ITS, and other quasi-experimental designs acceptable in EPOC reviews.
Databases
The Cochrane Central Register of Controlled Trials (CENTRAL), part of The Cochrane Library. www.thecochranelibrary.com [2013, Issue 1] MEDLINE, OVID [1948 to February 2013] EMBASE, OVID [1947 to February 2013] HealthStar, OVID [1999 to February 2013] Cumulative Index to Nursing and Allied Health Literature CINAHL, EbscoHost [1980 to February 2013] The EPOC Specialised Register, Reference Manager [2013, Issue 1] PubMed [up to February 2013] Science Citation Index, Social Sciences Citation Index [1975 to February 2013 (Web of Science) Dissertations and Theses, ProQuest [1861 to February 2013] African Index Medicus (AIM), World Health Organization (WHO) [up to February 2013] Latin American and Caribbean Health Sciences database (LILACS), Virtual Health Library (VHL) [up to February 2013] World Health Organization Library Information System (WHOLIS) [WHO publications, 1948 to February 2013; WHO journals & technical documents, 1986 to February 2013]
Study flow diagram
3.2.1 Searching other resources
Other systematic reviews
Related systematic reviews were identified by searching the Database of Abstracts of Reviews of Effectiveness (DARE) and the databases listed above.
Gray literature
An extensive gray literature search was conducted. Gray literature sites were identified using a list of databases relevant to low- and middle-income countries and websites compiled by the EPOC Group's Norwegian satellite office (http://epocoslo.cochrane.org/lmic-databases) and Grey matters: a practical search tool for evidence-based medicine (http://cadth.ca/resources/grey-matters). Search strategies for these databases and web sites were simpler than those used in mainstream databases such as MEDLINE and EMBASE. Gray literature sites, search terms, and number of results are provided in Appendix 2.
Additionally, we:
a) Reviewed reference lists of relevant systematic reviews and studies selected for inclusion in this review.
b) Contacted authors of relevant studies or reviews to clarify reported published information or to seek unpublished results/data.
c) Contacted agencies that either conduct studies or provide funding for malaria interventions and requested data from unpublished and ongoing studies. These agencies included the International Initiative for Impact Evaluation (3ie), UNICEF, United Nations Foundation, International Federation of Red Cross and Red Crescent Societies, World Health Organization, United States Agency for International Development (USAID), the Johns Hopkins Bloomberg School of Public Health, Population Services International (PSI), World Bank, World Vision, and Alliance for Malaria Prevention.
d) Conducted reference searches on studies selected for inclusion in this review, studies cited in related reviews, and other relevant citations in ISI Web of Science/Web of Knowledge.
3.3 DATA COLLECTION AND ANALYSIS
The data collection and analysis were conducted following the guidelines available in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
3.3.1 Selection of studies
Two individuals independently screened all titles and abstracts for potentially relevant studies. Full-text copies of all papers that were deemed potentially eligible by consensus of two authors were retrieved and screened against the inclusion criteria. Any disagreement about the eligibility was resolved by discussion between the two review authors and by consulting a third review author as necessary.
3.3.2 Data extraction and management
Two review authors independently undertook data extraction from the full text of each eligible study based on the EPOC Data Collection Checklist (EPOC 2010). Any disagreement was resolved by discussion between the two review authors and consultation with a third author if required. If any data were missing, we contacted the relevant corresponding authors. We collected data for specific populations (e.g. pregnant women, low socioeconomic status) for use in subgroup analyses, as described below.
3.3.3 Assessment of risk of bias in included studies
Two review authors independently assessed risk of bias for included studies. For RCTs, we assessed the risk of bias using the Cochrane Collaboration's ‘Risk of bias' tool (Higgins 2011), which assesses the following domains: generation of randomization sequence, allocation concealment, blinding of participants and providers, blinding of outcome assessors, selective outcome reporting, incomplete outcome data, and other sources of bias (e.g. major baseline differences, etc.). We assessed each potential source of bias and graded as high, low or unclear. A justification for our judgment is reported in the ‘Risk of bias' tables. Any disagreement about the risk of bias was resolved by discussion between two authors, and with a third review author when necessary. Due to the characteristics of interventions that we explored in our review, blinding of study participants and providers was not possible. However, lack of blinding was still considered a high risk of bias.
For other study designs, we used the risk of bias criteria suggested by EPOC: generation of allocation sequence; concealment of allocation; baseline outcome measurements; baseline characteristics; incomplete outcome data; blinding of outcome assessor; protection against contamination; selective outcome reporting; and other risks of bias (EPOC 2014b).
3.3.4 Measures of treatment effect
We reported pre-intervention and post-intervention means or proportions for both study and control groups. For any baseline imbalances, we calculated unadjusted and adjusted absolute change from baseline with 95 percent confidence intervals (CIs), where baseline results were available from RCTs, non-randomized controlled trials, and CBAs. We accounted for existing ITNs when reported by authors.
Additionally, we reported findings separately for RCTs, non-randomized controlled trials, and CBAs.
3.3.5 Unit of analysis issues
Where possible, we attempted to re-analyse data with potential unit of analysis errors. In cluster-randomized trials, a cluster is the unit of allocation. The cluster refers to a group of individuals enrolled in the study such as medical clinic patients, villages or families. The individuals within a cluster may be more similar in their characteristics than they are to individuals in other clusters. Therefore, it would be incorrect to analyse the data as if the individual participants were the unit of allocation. For cluster-randomized trials with unit-of-analysis errors, we used statistical methods to perform the analysis at the individual level while accounting for inter-cluster correlation. We used the Cochrane Handbook for Systematic Reviews of Interventions' methods to calculate the variance inflation factor (Higgins 2011). Eight of the ten included studies were cluster RCTs (cRCT). Muller 2008 required re-analysis because the sampling unit within the clusters was households and data in the paper were presented for individuals. Muller 2008 reported using a design effect of two to account for potential intra-cluster correlation (ICC). To calculate CIs, we adjusted the observed proportions using the design effect of two as reported in the study (Muller 2008). The other cRCTs were deemed to be appropriately analysed as they considered the unit of randomization to be at the cluster level (Amoran 2012; Cohen 2010; Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Tarozzi 2011).
3.3.6 Dealing with missing data
We attempted to obtain missing data from the authors of the included studies. When this was not possible, we performed the analysis using only the available data. We explored the impact of missing data on the review findings in the Discussion section of our review.
3.3.7 Assessment of heterogeneity
We described the variability of interventions identified in our review using a logic model and framework (Figure 1 and Figure 2). We planned to explore heterogeneity using forest plots and the I2 statistic, if possible. We planned to perform Chi2 tests and evaluate P values using 0.10 as a cut-off point. The importance of I2 would depend on the strength of evidence of heterogeneity (P value from Chi2 test or a CI for I2) and on the magnitude and direction of effects.
3.3.8 Assessment of reporting biases
We planned to use funnel plots to explore the possibility of publication bias, if 10 or more studies were included in any meta-analysis and if studies were of different sizes. We would have visually inspected funnel plots for asymmetry and explored reasons for asymmetry. However, we did not include 10 or more studies in the meta-analysis.
3.3.9 Data synthesis
We performed data synthesis using Review Manager 5.1 (RevMan) (RevMan 5.1). We reported the overall number of studies included in the review and the main research questions addressed. We synthesized information about study methods (e.g. study design, duration of intervention, follow-up); participants (e.g. sex/gender, age, country, setting); intervention (e.g. intervention description and its components, means of delivery, methods of communication); setting (for both control and intervention group if available); outcomes (list of outcomes and time points reported); and notes (other details of the study that do not fall under mentioned categories). We commented on study designs, analytical methods, risk of bias, generalizability and relevance of study results as well as other important study characteristics.
We reported the outcomes of interest as published in the original papers. Due to lack of standardization of outcomes, we were only able to pool the results from two studies for the outcome “nets owned per capita (measured as nets owned)” (Fink 2012; Tarozzi 2011). We used a random-effects model for this analysis and a generic inverse-variance model of analysis. When other outcomes were similar, we reported the direction of the effect (e.g. we considered “households with at least one ITN” and “households with appropriately hanging ITNs” as similar outcomes). Some outcome measurements are considered more reliable than others (e.g. inspection of ITNs in the house versus self-reporting of the use of the ITN), so we also reported the outcomes grouped by the way in which the outcome was assessed, if appropriate.
We included the magnitude of the effects of the interventions and the quality of evidence, by summarizing available data on each of the outcomes of interest. We included both relative and absolute measures of effect when possible.
We reported effect sizes for each included study using risk ratios (RRs) and odds ratios (ORs) (random-effects models) with 95 percent confidence intervals (CIs). Where possible, we have used RRs. For four studies, we have reported ORs for some outcomes, as reported by the study authors. These results were analyzed using logistic regression as the baseline data required to convert these to RRs were not available. Where studies reported more than one measure for each endpoint, we abstracted the primary measure (as defined by the authors of the study).
We compared included studies with an emphasis on explanation for potential differences between study results, taking into account the context where the intervention was implemented. We included a table that summarizes characteristics and findings of included studies in a consistent and systematic manner (Table 1), following our logic model and framework (Figure 1 and Figure 2).
We assessed the certainty of evidence for each primary outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach and the GRADEpro software (Guyatt 2008a). The certainty of the evidence may be downgraded for factors such as study limitations, inconsistency of results, imprecision, indirectness of evidence, or publication bias. Factors that increase the certainty of the evidence include a large magnitude of effect, plausible confounding, and dose-response gradient (Guyatt 2008b). Finally, we presented the main findings in ‘Summary of findings’ tables.
3.3.10 Subgroup analysis and investigation of heterogeneity
We planned to conduct subgroup analyses to explore heterogeneity, according to the following study characteristics: the type of intervention as per the logic model (Figure 1 and Figure 2) (e.g. cost to end user); specific population characteristics (e.g. gender/sex, ethnicity, geographic areas, rural/urban, socioeconomic status, education, age/subgroup (children under five years and pregnant women)); and the number and type of interventions in each strategy. However, we had insufficient data to complete these planned subgroup analyses.
We also extracted details on the above subgroup analyses conducted within the included studies.
4 Results
4.1 DESCRIPTION OF STUDIES
4.1.1 Results of the search
Our search strategy yielded 3,032 references. During the abstract screening process, we excluded 2,899 references for failing to meet one or more inclusion criteria. The remaining 133 references were reviewed as full-text, plus an additional three studies identified through reference-list checking and two identified through personal communication with experts. We excluded 117 studies that did not meet our eligibility criteria and excluded eight others after obtaining information from the study authors. Thirteen studies met our eligibility criteria and ten were included in our review. Of the remaining three studies, two are ongoing and one requires additional information from the authors. For details please see Figure 3 and the Characteristics of excluded studies table. Full results from the gray literature searches are available on request.
4.1.2 Included studies
Ten studies were included in this review: eight cluster RCTs (Amoran 2012; Cohen 2010; Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008, Tarozzi 2011); one RCT (De La Cruz 2009); and one CBA study (Deribew 2012). Nine of the included studies included 17,203 households (Amoran 2012; De La Cruz 2009; Deribew 2012; Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008; Tarozzi 2011). The remaining study reported by individual and included 424 participants (Cohen 2010).
Additionally, three other studies were eligible but not included: one study is awaiting classification (Soleimani Ahmadi 2012); and the two remaining studies are ongoing (Kilian 2012; Siekmans 2012).
Study participants
Study participants in our review were recruited in rural Africa (Amoran 2012; Cohen 2010; De La Cruz 2009; Deribew 2012, Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008); and rural India (Tarozzi 2011). They belong to several categories: adults who were enrolled during household visits (Krezanoski 2010; Dupas 2009); pregnant women enrolled through prenatal clinics (Cohen 2010; Muller 2008); mothers of children under five years old (Amoran 2012); adults and women of reproductive age enrolled using client lists of NGOs that offer financial services in the community (De La Cruz 2009; Tarozzi 2011); rural cotton farmers (Fink 2012); heads of households, pregnant women, and children under five (Deribew 2012), and all households in selected villages (Keating 2012).
Characteristics of intervention
Interventions in studies included in our review aimed to increase the ownership or use of ITNs. We categorized them according to the logic model developed for this review in which interventions are divided into those focusing on different ITN delivery strategies (to increase ITN ownership) and those interventions focusing on appropriate use (Figure 1 and framework in Figure 2). A detailed classification of interventions used in each included study can be found in Table 1 in which the included studies are mapped according to the logic model.
Five included studies used a combination of strategies focusing on ITN delivery to increase ITN ownership and appropriate use (Cohen 2010, Dupas 2009, Krezanoski 2010, Muller 2008, Tarozzi 2011). One study focused on delivery strategies to increase ownership only (Fink 2012) and four studies examined appropriate use strategies (Amoran 2012, De La Cruz 2009, Deribew 2012, Keating 2012).
Delivery strategies to increase ITN ownership included administering the intervention through routine services (e.g. prenatal clinics) (Cohen 2010; Muller 2008); or administering the intervention in the community (e.g. house-to-house visits) (Dupas 2009; Fink 2012; Krezanoski 2010; Tarozzi 2011).
Interventions to promote the appropriate use of ITNs were provided mainly through the community (Amoran 2012; Dupas 2009; De La Cruz 2009; Deribew 2012; Keating 2012; Muller 2008; Tarozzi 2011). Different educational methods were used to increase the use of ITNs. These included: training for household heads (Deribew 2012); house-to-house visits to teach the population how to appropriately hang and use an ITN, educational sessions plus printed materials (Amoran 2012; Keating 2012); and train-the-trainer education sessions (De La Cruz 2009). Appropriate ITN-use interventions were available as one-time activities (e.g. during house-to-house visits (Dupas 2009); or during the whole duration of the campaign (e.g. Muller 2008 used long-term ITN social marketing); and were coordinated by civil society or the public sector, mainly by NGOs with local community services involved (Deribew 2012; Dupas 2009; Muller 2008; Keating 2012). Some studies identified pregnant women and children under the age of five as populations of special interest (Amoran 2012; Cohen 2010; De La Cruz 2009; Deribew 2012; Keating 2012; Tarozzi 2011).
The interventions in all studies were part of campaign evaluation and were of a relatively brief duration. Furthermore, study participants could choose the type of ITN in only one study - in this study participants could choose the size of the ITN they purchased (Tarozzi 2011). When evaluating the cost to end users, ITNs were provided to study participants free (Amoran 2012; Cohen 2010; Deribew 2012; Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008); partially subsidized (Cohen 2010; Dupas 2009; Muller 2008; Fink 2012); and at a full market price (Fink 2012; Tarozzi 2011). The public sector (e.g. prenatal clinics) (Cohen 2010; Fink 2012; Keating 2012; Muller 2008), the commercial sector (e.g. local retailers) (Dupas 2009; Muller 2008), and civil society (e.g. NGOs) (Deribew 2012; Dupas 2009; Krezanoski 2010; Tarozzi 2011) were involved in the distribution of ITNs.
Outcomes measured
The outcomes reported in the identified studies were ITN ownership, ITN use, and malaria morbidity. Outcomes measured in pregnant women or children under the age of five were reported separately as these populations are identified as vulnerable groups.
4.1.3 Excluded studies
Eighteen articles were identified as potentially eligible during the full text screening, of which eight articles referring to seven studies were excluded from the review (Characteristics of excluded studies). Studies were excluded for the following reasons: not an eligible study design, or they assessed the efficacy of ITNs rather than a strategy to increase ownership, use, or both.
4.2 RISK OF BIAS IN INCLUDED STUDIES
The risk of bias for each included study is described in the Characteristics of included studies and shown in Figure 4 and Figure 5.
Risk of bias: review authors' judgements about each risk of bias item presented as percentages across all included studies.
4.2.1 Allocation (selection bias)
Selection bias was assessed as low in most studies (Cohen 2010; Dupas 2009; Fink 2012; Krezanoski 2010; Tarozzi 2011). The method of randomization was unclear for four studies (Amoran 2012; De La Cruz 2009; Keating 2012; Muller 2008). One study was a non-randomized study and assessed as high risk for selection bias (Deribew 2012).
4.2.2 Blinding (performance bias and detection bias)
Assessment of outcomes was both objective (e.g. visual observation of ITN being used, blood tests for malaria parasite and hemoglobin levels) (Cohen 2010; Deribew 2012; Fink 2012; Krezanoski 2010; Muller 2008); and subjective (e.g. self-reported ITN use) (Amoran 2012; Cohen 2010; Dupas 2009; De La Cruz 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008; Tarozzi 2011). Cohen 2010 used blinding for both study participants and the individuals involved in the intervention administration by not disclosing that ITN use would be evaluated. However, due to the nature of the interventions (e.g. distributing free ITNs), blinding of study participants or assessors was not always possible. In Deribew 2012, the same individuals who administered the intervention also collected outcome data.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
4.2.3 Incomplete outcome data (attrition bias)
The risk of bias for incomplete outcome data reporting was assessed as unclear in nine of the studies, because data on attrition and loss to follow-up were not fully reported (Amoran 2012; Cohen 2010; Dupas 2009; De La Cruz 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008; Tarozzi 2011). The Deribew 2012 study had a high number of dropouts which was not explained by the authors and therefore was assessed as being at high risk of bias.
4.2.4 Selective reporting (reporting bias)
Muller 2008 was the only study that provided a study protocol. This study did not report all the outcomes described in the protocol (e.g. costs per malaria case, insecticide content on ITN, acceptance of health staff and population) and thus the risk of bias for selective reporting was assessed as high. Deribew 2012 described the study methods and baseline results in a separate publication and all outcomes assessed at baseline were assessed post-intervention. Therefore, this study was assessed as low risk of bias. The risk of bias for selective reporting was assessed as unclear for the remaining studies (Amoran 2012; Cohen 2010; De La Cruz 2009; Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Tarozzi 2011).
4.2.5 Other potential sources of bias
One study was reported to be at high risk of other sources of bias as it did not include any objective measures of ITN use following the educational intervention and relied on self-reported ITN use (Amoran 2012).
Five of the studies were evaluated to have unclear risk for other potential sources of bias for reasons described below (Cohen 2010; De La Cruz 2009; Dupas 2009; Keating 2012; Muller 2008).
Cohen 2010 observed some leakages and mismanagement of ITNs in 4 of 11 clinics asked to sell ITNs at a positive price. In De La Cruz 2009 there was a difference in the characteristics of study participants at baseline between the intervention and the control groups.
Keating 2012 found that some lay or community health workers (CHW) delivered the intervention to control group communities and the results indicate that some control houses received information about malaria or assistance with hanging a net. Additionally, the intervention group had higher ITN use at baseline. The authors also reported contamination across communities, with 47 percent of children in the intervention communities living in houses reporting exposure to the CHW intervention and 31 percent of children in control houses reporting the same exposure (Keating 2012).
Two studies observed some differences in participants' characteristics at baseline (Deribew 2012, Muller 2008). Dupas 2009 did not report on similarity in baseline characteristics between the intervention and the control groups enrolled in their study so the risk of other potential sources of bias was unclear.
Other potential sources of bias were not observed or suspected in the other three included studies (Fink 2012; Krezanoski 2010; Tarozzi 2011).
4.3 EFFECTS OF INTERVENTIONS
See Summary of findings table. The results are reported by intervention, as described in the logic model.
4.3.1 Impacts of distributing free ITNs compared to making ITNs available for purchase through different mechanisms
The five studies that broadly addressed this comparison indicate overall that providing free ITNs probably increases the number of people who own ITNs compared to providing subsidized ITNs or ITNs offered at the full market price (moderate-certainty evidence) (Dupas 2009; Cohen 2010; Muller 2008; Fink 2012; Tarozzi 2011). The studies also indicate that providing free ITNs probably leads to little or no difference in usage of ITNs, compared to providing subsidized ITNs or ITNs offered at the full market price (moderate-certainty evidence).
The following section reports the results of the specific comparisons separately. Where an outcome is not listed, it was not assessed by the studies in that comparison.
Comparison 1: Free ITN compared to paying for an ITN (range of costs)
Three studies addressed this comparison (Dupas 2009; Cohen 2010; Muller 2008). Overall, these studies found that providing free ITNs probably increases ITN ownership compared to providing subsidized ITNs or ITNs at market price (moderate-certainty evidence) (Summary of findings Table 1).
The first study was conducted in rural Kenyan villages and evaluated four different interventions and their impact on ITN ownership and use among children and adults (Dupas 2009). In this study, households were randomly assigned to: a subsidy level (from 40% to 100% subsidy); one of three “marketing” groups (health framing, financial framing or no marketing); one of two “commitment” groups (verbal commitment or no verbal commitment); and one of three “targeting” groups (male head, female head, or both female and male head of household) (Dupas 2009). The effect of the “marketing”, “commitment”, and “targeting” interventions on use are presented later.
The second study included pregnant women from rural Kenya and compared the impact of different levels of subsidy for ITNs on ITN ownership and use (Cohen 2010). ITNs in this study were provided either free; or for KES 10 (approximately $0.15 USD; 97.5% subsidy); or KES 20 (approximately $0.30 USD; 95% subsidy); or KES 40 (approximately $0.60 USD; 90% subsidy).
Muller 2008 was conducted in rural Burkina Faso and evaluated the effect of social marketing with and without free distribution of ITNs to pregnant women through prenatal clinics. Highly subsidized ITNs were available for the whole study area.
Ownership
Three studies assessed this outcome (Cohen 2010; Dupas 2009; Muller 2008).
The Dupas 2009 study found that receiving a free ITN probably increases ITN ownership. ITN ownership was the highest among the free group compared to those who paid any price (RR 3.02, 95% CI 2.78 to 3.30, moderate certainty evidence) (Analysis 6.1). Dupas 2009 observed that an increase in price from no cost to USD 1 reduces ITN ownership among households by 35 percent and an increase in price from USD 1 to USD 2 reduces ITN take-up by an additional 25 percent (Dupas 2009) (Analysis 1.1; Analysis 2.1; Analysis 3.1; Analysis 4.1; Analysis 5.1).
Cohen 2010 found that providing free ITNs probably increases ITN ownership compared to those providing ITNs at any price point (RR 1.29. 95% CI 1.25 to 1.33, moderate-certainty evidence) (Analysis 10.1).
Free distribution of ITNs to pregnant women plus social marketing probably increases ITN ownership compared to social marketing alone (RR 1.49, 95% CI 1.14 to 1.95, moderate-certainty evidence) (Analysis 11.1) (Muller 2008).
Appropriate use
Overall, the two studies that assessed this outcome found that free ITN distribution probably leads to little or no difference in appropriate ITN use, compared to providing subsidized ITNs or ITNs offered at the full market price (moderate-certainty evidence).
Dupas 2009 reported that there is probably little to no difference in ITN usage between those who received the ITN at no cost and those who paid for it (moderate-certainty evidence) (Analysis 1.2; Analysis 2.2; Analysis 3.2; Analysis 4.2; Analysis 5.2).
Cohen 2010 also found that there is probably no pattern of increasing use with increasing ITN price. Specifically, the study noted that there was probably little to no difference in ITN usage between those who received the ITN free and those who paid KES 10, 20, or 40 (moderate-certainty evidence) (Analysis 7.2; Analysis 7.3; Analysis 8.2; Analysis 8.3; Analysis 9.2; Analysis 9.3).
Secondary outcomes
Equity ratio of household ownership
The included studies did not contain enough information to calculate the equity ratio of household ownership of ITNs according to household income. Dupas 2009 reported observational data that baseline wealth and financial access predicted ITN ownership but, once purchased, they did not predict ITN use. In the same study, baseline educational levels of household members (male or female) did not predict ITN ownership (Dupas 2009). Cohen 2010 reported that women who paid KES 40 for their ITN were slightly more likely to pay for transportation to the clinic, to be literate and to be wearing shoes when attending the clinic.
Comparison 2: Free ITN compared to ITN available for purchase using a microfinance loan
Two studies assessed this comparison (Fink 2012; Tarozzi 2011).
Fink 2012 had three arms: a control group who did not receive the intervention; a group that received free ITNs (up to one ITN for each uncovered sleeping space); and a loan group who could obtain one ITN per sleeping space at 50 percent subsidized cost. The loan group was required to repay the loan by the end of the harvest period.
Tarozzi 2011 conducted a large-scale study in rural India. The study had three arms: a control group that did not receive the intervention (ITNs were available at market price); a group that received free ITNs based on household composition (free group); and a group that was offered microfinance loans (due to be repaid in one year) (MF group). These loans could be used to purchase ITNs or re-treatments. Outcomes related to ITN ownership, ITN use, anemia and malaria infection status were reported. ITN use was measured both with self report and with direct observation (Tarozzi 2011).
Ownership
The pooled data for Fink 2012 and Tarozzi 2011 suggest that receiving an ITN at no cost probably increases ITN ownership, compared to purchasing an ITN at the market price (SMD 0.69, 95% CI 0.39 to 0.99, moderate-certainty evidence) (Analysis 12.1) or purchasing an ITN using a loan (SMD 0.37, 95% CI 0.27 to 0.47, moderate-certainty evidence) (Analysis 14.1). There is probably little to no difference in ITN ownership when comparing groups in which ITNs were purchased using a loan and those in which ITNs were purchased at the market price (SMD 0.29, 95% CI -0.06 to 0.63, moderate-certainty evidence) (Analysis 13.1).
Appropriate use
Fink 2012 found that there was probably little to no difference in ITN use between those who received a free ITN and those who bought ITNs using a loan or at the market price. Approximately 90 percent of households in all three arms were using ITNs when assessed at follow-up. The mean difference in ITN use (measured as ITNs seen hanging at follow-up) was: 1.42 higher in the free group compared to control; 0.79 higher for the microfinance group compared to control; and 0.63 higher for the free group compared to the microfinance group (moderate-certainty evidence).
Tarozzi 2011 noted that self-reported ITN use was probably lower in the microfinance group than in the free group for the previous night (OR 0.74, 95% CI 0.68 to 0.79, moderate-certainty evidence) (Analysis 14.4) and during the peak season (OR 0.66, 95% CI 0.61 to 0.72, moderate-certainty evidence) (Analysis 14.3). Field staff were able to verify the usage of ITNs in 93 percent of free households and 89 percent of microfinance households.
Secondary outcomes
Malaria-specific morbidity in children
Fink 2012 found that children under 15 years in the groups receiving free ITNs or through loans were probably less likely to report confirmed malaria than controls (Free group versus control: OR 0.547, 95% CI 0.231 to 0.993; Loan group versus control: OR 0.444, 95% CI 0.213 to 0.924, moderate-certainty evidence) (Fink 2012). Among children under 5 years old in the Tarozzi 2011 study, malaria prevalence was 11 percent in the control group but 18.4 percent in the free group and 19.8 percent in the microfinance group.
Malaria-specific morbidity in adults
Fink 2012 noted that the number of confirmed cases of malaria in adults over 15 years was probably fewer in the free and loan groups compared to controls (Free group versus control: OR 0.361, 95% CI 0.122 to 1.072, moderate certainty evidence; Loan group versus control: OR 0.307, 95% CI 0.106 to 0.887, moderate-certainty evidence) (Fink 2012).
4.3.2 Impacts of interventions to promote the appropriate use of ITNs
Five studies assessed the impacts of educational interventions on the appropriate use of ITNs (De La Cruz 2009, Deribew 2012, Keating 2012, Amoran 2012, Krezanoski 2010). These studies found that providing an educational intervention may increase the number of adults and children using ITNs (sleeping under ITNs), compared to controls (Summary of findings Table 1).
Comparison 1: Malaria education versus an attention placebo of diarrhea education
De La Cruz 2009, conducted through a microfinance association that uses a ‘Credit with Education’ program in rural Ghana, compared three groups. One group of clients of the microfinance association received education about malaria; another group of clients received education about diarrhea prevention; and a third group from the same communities who were not clients of the microfinance association did not receive any educational intervention. Participants in the third group were not randomly selected and therefore have not been included in this review.
The malaria education curriculum consisted of 10 interactive sessions between 40 minutes and one hour in duration. The sessions included topics such as malaria transmission, where ITNs are sold, developing a plan to increase ITN use, identifying signs of serious malaria, the importance of chemoprophylaxis during pregnancy, and actions to improve prevention and treatment of malaria in the community (De La Cruz 2009). Participants who received diarrhea education learned about preventing and treating diarrhea through seven sessions. The follow-up time was approximately two years.
Ownership
Participation in malaria education sessions may increase ownership of ITNs (RR 2.01, 95% CI 1.08 to 3.74, low-certainty evidence), compared to participating in diarrhea education sessions (Analysis 15.1) (De La Cruz 2009).
Appropriate use
Providing education about malaria to parents may increase use of ITNs, measured by the number of children sleeping under an ITN (RR 2.36, 95% CI 1.02 to 5.46, low-certainty evidence), compared to providing diarrhea education (De La Cruz 2009) (Analysis 15.2).
Secondary outcomes
This study did not report on any of the secondary outcomes for the review.
Comparison 2: Malaria education for heads of households versus no education
The Deribew 2012 study was conducted in a malaria-endemic area in Ethiopia. The intervention included free ITNs and training for the heads of households on the appropriate use of ITNs. The comparison group received free ITNs but no training. Nine intervention village residents received a five-day training course for trainers on the appropriate use of ITNs. These village residents then offered training to all household heads in the intervention villages. ITN use was measured by direct observation by trained village residents (Deribew 2012).
Ownership
Ownership was not measured in this study.
Appropriate use
The study suggests that malaria education may increase ITN use among children under five years of age at both six-months follow-up (RR 1.20, 95% CI 1.16 to 1.24, low-certainty evidence) (Analysis 16.1); and twelve-months follow-up (RR 1.60, 95% CI 1.52 to 1.68, low-certainty evidence) (Deribew 2012) (Analysis 16.2), compared to diarrhea education.
Secondary outcomes
Malaria-specific morbidity in children
Children under 5 years old in the intervention group were less likely to be anemic than those in the control group at both 6-months and 12-months follow-up (Deribew 2012).
Malaria-specific morbidity in adults
In Deribew 2012, fewer participants in the intervention group reported having had a fever within the last two weeks at six-months follow-up (RR 0.29, 95% CI 0.26 to 0.31, low-certainty evidence) (Analysis 16.3); and twelve-months follow-up (RR 0.19, 95% CI 0.16 to 0.22, low-certainty evidence), compared to those in the control group (Analysis 16.4).
This study did not report on any of the other secondary outcomes.
Comparison 3: Malaria education provided by community health workers versus no education
In the cluster randomized trial conducted by Keating 2012, lay or community health workers (CHWs) delivered an interpersonal communication intervention to mothers and heads of households that focused on providing information about malaria transmission and prevention through the use of ITNs. CHWs gave information about ITN repair and re-treatment, who should use ITNs, and their importance year round. The CHWs also provided information to motivate individuals to use ITNs, such as information on the effectiveness of ITNs for malaria prevention. This information was provided through house-to-house visits using visual aids and information pamphlets. Community plays and demonstrations were also held. CHWs were trained to provide on-site assistance with hanging ITNs including provision of any required materials (e.g. string, nails, or hammers) and repairing ITNs.
Ownership
This study did not report on ITN ownership.
Appropriate use
The study found that malaria education by CHWs may increase ITN use among children under five years (OR 2.24, 95% CI 1.46 to 3.44, low-certainty evidence) (Analysis 17.2) and any ITN use in the house (OR 2.06, 95% CI 1.35 to 3.14, low-certainty evidence) (Analysis 17.1), compared to no education.
Secondary outcomes
This study did not report on any of the secondary outcomes.
Comparison 4: Malaria education provided to nursing mothers versus no education
Amoran 2012 provided a structured educational program based on the national malaria control program's course content (Amoran 2012). Nursing mothers from rural communities of Ogun State, Nigeria, were randomized to receive the intervention. The intervention consisted of a one-day training session of three units: malaria transmission, malaria prevention and treatment, and malaria practices. Participants were surveyed three months after the intervention to assess utilization of ITN and knowledge about malaria.
Ownership
This study did not report on ITN ownership.
Appropriate use
This study found that malaria education probably increases ITN use (self-reported), compared to no education (RR 2.29, 95% CI: 1.82 to 2.89, moderate-certainty evidence) (Analysis 18.1) (Amoran 2012).
Secondary outcomes
This study did not report on any secondary outcomes.
Comparison 5: Education with incentives versus no incentives on appropriate use of ITNs
In the cluster randomized trial by Krezanoski 2010, all study participants received a free ITN. However, the intervention communities were promised an undisclosed prize at one-month follow-up if they complied with the appropriate use of ITNs. A final follow-up home visit to assess ITN use was conducted by surveyors at six months to check for mounted ITNs.
Ownership
Participants in both groups were provided with free ITNs and there was no difference in ITN ownership between intervention and control groups (Analysis 19.1).
Appropriate use
This study found that providing incentives plus education probably leads to little to no difference in ITN use (measured by visual inspection for mounted nets) at one-month follow-up (RR 1. 10, 95% CI 0.94 to 1.28, moderate-certainty evidence) and six-months follow-up (RR 1.00, 95% CI 0.95 to 1.07, moderate-certainty evidence), compared to not providing incentives (Analysis 19.2) (Krezanoski 2010).
Secondary outcomes
This study did not report on any of the secondary outcomes.
4.3.3 Impacts of a combination of ITN delivery mechanisms and interventions to promote the appropriate use of ITNs
One study assessed the effects of implementation factors (Dupas 2009). The interventions evaluated are described above.
Ownership
The study found there is probably little to no difference in ITN ownership between a group that received a health-framing marketing message versus no framing and a group that received a financial-framing marketing message versus no framing (moderate-certainty evidence) (Analysis 20.1; Analysis 21.1). Additionally, there is probably little to no difference in ITN ownership when vouchers are given to either female heads versus both heads jointly or male heads versus both heads jointly or when there is a verbal commitment to purchase an ITN versus no verbal commitment (Analysis 22.1, Analysis 23.1, Analysis 24.1, moderate-certainty evidence) (Dupas 2009).
Appropriate use
The study found there is probably little to no difference in appropriate ITN use between a group that received a health-framing marketing message versus no framing and a group that received a financial-framing marketing message versus no framing (Analysis 20.2; Analysis 21.2, moderate-certainty evidence). Additionally, there is probably little to no difference in appropriate ITN use when vouchers are given to either female household heads versus both heads jointly or male heads versus both heads jointly or when there is a verbal commitment to purchase an ITN versus no verbal commitment (Analysis 22.2, Analysis 23.2, Analysis 24.2, moderate-certainty evidence) (Dupas 2009).
Secondary outcomes
Secondary outcomes were not assessed for this comparison.
4.3.4 Adverse Effects
None of the included studies reported on adverse effects of the intervention.
5 Discussion
5.1 SUMMARY OF MAIN RESULTS
The five included studies that examined the impact of ITN price on ownership found that ownership was highest when ITNs were available free, which indicates that ITNs may be acquired by those in lower socioeconomic groups when the cost is decreased (Cohen 2010; Dupas 2009; Fink 2012; Muller 2008; Tarozzi 2011). These studies varied the cost of ITNs: two provided ITNs at varying subsidized prices, one compared free ITNs distributed to pregnant women in prenatal clinics to subsidized ITNs for the general population, and two studies compared loans to free distribution (Cohen 2010; Dupas 2009; Fink 2012; Muller 2008; Tarozzi 2011).
Secondly, education about the appropriate use of ITNs may increase appropriate use, compared to no education intervention or an attention placebo (diarrhea education). This effect was found regardless of whether ITNs were available free or for a cost within the community (De La Cruz 2009), and when education was tailored for the household heads (Deribew 2012), or given to mothers (Amoran 2012). Thirdly, ITN use was not affected by other factors such as whether an incentive (such as prizes) was offered (Krezanoski 2010).
Four studies failed to confirm the hypothesis that people who purchase nets will use them more than those who receive them free (Cohen 2010; Dupas 2009; Fink 2012; Tarozzi 2011). These studies show that there is probably little or no difference in rates of ITN use among those who receive a free ITN compared to those who pay for one.
5.2 OVERALL COMPLETENESS AND APPLICABILITY OF EVIDENCE
The studies included in this review are very heterogeneous. The studies were conducted in diverse populations in low- and middle-income countries, with nine of the ten studies undertaken in rural Africa (Amoran 2012; Cohen 2010; De La Cruz 2009; Deribew 2012, Dupas 2009; Fink 2012; Keating 2012; Krezanoski 2010; Muller 2008) and one in rural India (Tarozzi 2011). Geographic location may impact the effectiveness of ownership and use strategies, depending on the burden of malaria in that setting.
A recent survey, conducted in Ethiopia, indicates that it may not be enough to educate people about appropriate use of ITNs unless other potential barriers are identified and addressed. Respondents reported not using an ITN the previous night for the following reasons: the ITN was considered to be too old, too badly damaged, too dirty, or unavailable for use because it was being cleaned (Batisso 2012). Our review did not identify barriers preventing ITN use as we focused on interventions to promote use. However, the findings of Batisso 2012 are important for identifying the reasons ITNs are not used.
There were not enough data available in the included studies to assess the impact of different strategies on the equity ratio of household ownership and appropriate use of ITNs. Two studies reported on observational analyses of household wealth and ITN ownership and use: Dupas 2009 found that baseline wealth and financial access predict ITN ownership but, once purchased, they do not predict ITNs' use. Keating 2012 found little to no difference in ITN use by children according to household wealth. However, this information was not presented separately for intervention and control groups.
None of the included studies reported on whether the authors looked for or found any evidence relating to adverse effects. However, interventions that provide subsidized or free ITNs to the community may create resource allocation issues by diverting resources from other priority issues to ITN delivery strategies.
5.3 QUALITY OF THE EVIDENCE
Most of the included studies were well-designed cluster randomized studies that provided very detailed descriptions of the methods used and were assessed to be at low risk of bias (Figure 3). The risk of bias was high for blinding of participants for all studies (Figure 4) since all study participants were aware of their assignment to delivery strategies such as free ITNs or education about ITNs. One study was assessed as high risk of bias because of contamination between study groups (Keating 2012).
5.4 POTENTIAL BIASES IN THE REVIEW PROCESS
A limitation of this systematic review is that publication bias was not explored formally as we had too few studies to develop a meaningful funnel plot. Studies with negative findings may be reported in the gray literature rather than peer-reviewed journals. We tried to minimize this risk by using a comprehensive search strategy including searching numerous gray literature sources and websites, and contacting experts.
Six of the included studies had missing data and required additional information from the authors (Cohen 2010; De La Cruz 2009; Dupas 2009; Fink 2012; Muller 2008; Tarozzi 2011). This information related to adjusting for clustering, relative risk calculations, random sequence generation, and allocation concealment. The authors of six of these studies provided the additional data needed to complete this review.
We attempted to reduce additional sources of bias through the development of a logic model. This model helped us to clearly define our search terms and develop our inclusion/exclusion criteria, especially with regard to outcomes (Anderson 2011). We also involved a number of stakeholders in the development of the logic model which led to a more robust model (Kellogg 2004).
5.5 AGREEMENTS AND DISAGREEMENTS WITH OTHER STUDIES OR REVIEWS
Previous systematic reviews by Willey and colleagues reviewed the scale-up of different malaria prevention strategies, including ITNs, intermittent preventive treatment in pregnancy, and artemisinin combination treatment (Willey 2010; Willey 2012). These reviews included both qualitative and quantitative study designs. The Willey 2012 review included nine studies that reported an equity ratio for ITN ownership according to socioeconomic status and found that ownership was highest among households in the poorest quintile in four studies and higher in the least poor quintile in five studies. Both reviews by Willey and colleagues found that strategies which delivered free ITNs achieved high ownership and use, and this is in agreement with the results of this review. Although we were unable to determine the equity ratio of ITN ownership, an equity effect is consistent with our results that ITN ownership is higher when ITNs are available free or at a reduced cost.
Sexton 2011 examined best practices for ITN distribution to ensure satisfactory ITN coverage and malaria control. They found that best practices included a combination of delivering free ITNs and a “catch up” component. For example, a “catch-up” program may include free or subsidized ITN distribution during routine immunization days within areas with low coverage (Sexton 2011). This approach is in agreement with our findings that ITN ownership increases as cost decreases.
A recent survey, conducted in Nigeria, assessed the effect of a free mass distribution campaign on the equity of ITN ownership (Ye 2012). This study did not have a control group and was therefore excluded from our review. However, the study found similar results. This study concluded that free mass distribution campaigns appear to be more effective than other strategies intended to increase ITN ownership, such as social marketing campaigns and subsidized ITNs (Ye 2012).
6 Authors' conclusions
6.1 IMPLICATIONS FOR PRACTICE
The results of our review suggest that providing ITNs free probably increases the number of people who own ITNs, compared to providing subsidized ITNs or ITNs offered at full market price. Removing the barrier of cost improves access to ITNs for the most disadvantaged groups in a population, who are also the most at risk for malaria.
Education strategies to increase appropriate use of ITNs may increase the number of adults and children under five using ITNs. While it was not possible to identify optimal types, duration, or frequency of educational interventions, such interventions may be helpful given that 10 percent of those owning an ITN in sub-Saharan Africa do not report sleeping under it (WHO 2014).
Other factors to consider are the potential adverse effects of redistributing resources to provide free or subsidized ITNs. None of the studies included in our review examined the adverse microeconomic effects of interventions providing free or subsidized ITNs to the community: for example, whether funds for other health programs were redirected to cover the costs of free or highly subsidized ITNs. However, if adverse microeconomic effects exist, they almost certainly would be outweighed by the large, beneficial macroeconomic effects of reduced malaria mortality and morbidity (Gallup 2001). To maximise the benefits of ITNs, strategies to deal with unwanted or damaged ITNs and other bednets may be necessary.
6.2 IMPLICATIONS FOR RESEARCH
Implications for systematic reviews
In future updates of this review, and if data are available, we plan to conduct subgroup analyses to explore heterogeneity, according to the type and number of interventions and specific population characteristics (e.g. gender/sex, ethnicity, geographic areas, rural/urban, socioeconomic status, education, age/subgroup (children under five years and pregnant women)).
Implications for primary research
Apart from the positive effect of lower consumer cost on ownership, there was inadequate evidence to compare the relative effectiveness of competing delivery strategies such as mass free distribution versus social marketing on increasing ownership rates. Additional high quality evidence from trials is needed to identify the most effective delivery strategies.
Education about appropriate use of ITNs was shown in two studies to boost ITN use. However, this evidence is of low certainty because of the limited number of studies. Further high-quality randomized controlled trials are needed of educational interventions for appropriate ITN use. These trials should clearly describe the type and duration of the educational intervention and include larger numbers of participants in order to determine the optimal strategies for increasing ITN use. More trials are also needed on related social interventions to determine whether strategies that use social marketing, incentives, or other interventions to address behavior change are effective in increasing ownership and appropriate use of ITNs.
A limitation of this review is that we were unable to report whether there are differences in outcomes for ownership and use based on factors such as gender and household wealth. While two of the included studies reported on observational analyses, none of the studies reported these results by intervention group. Future trials should assess whether ownership and use of ITNs is affected by social and demographic factors.
Future randomized controlled trials should assess the longer-term sustainability of interventions to increase ownership and use of ITNs.
7 Acknowledgements
We would like to thank UNICEF and the African Leaders Malaria Alliance (ALMA), the organizers of the ‘Enhancing Mosquito Net Utilization Meeting‘, Geneva 2011 and the meeting participants for their support and useful comments on the draft of the protocol. We would also like to thank Don de Savigny and Kara Hanson who encouraged us when we applied for the funding and provided letters of collaboration. We would like to express our gratitude to Michelle Fiander, Trials Search Co-ordinator, EPOC Group, for her help with search strategy development; Carolynne Gabrielle for her help with gray literature searching; Simon Lewin, Elizabeth Paulsen, Susan Munabi-Babigumira, Jenny Hill and Jan Odgaard-Jensen for their support and guidance; and Sabrina Lee for her assistance with screening.
8 Contributions of authors
PT and VW conceived the idea for the systematic review. LAP and JP drafted and revised the systematic review with suggestions from EU, VW, ETG, JPP, PT, AA and MG who extensively reviewed the paper and provided feedback on the draft. GW was consulted for statistical advice. Data extraction and risk of bias assessment was conducted by LAP, EU, JP, ETG.
9 Declarations of interest
AA has published papers on malaria drug quality in Africa and India. MG has been engaged in studies potentially eligible for our Cochrane Review. LAP, EU, VW, JP, ETG, JPP, GW and PT have no known conflicts of interest.
10 Differences between protocol and review
In the protocol, ‘equity ratio of household ownership with ITNs calculated across household income’ was listed as a primary outcome. Upon further consideration, the author team decided that since this outcome relies on the primary outcome of ownership it is more appropriate as a secondary outcome.
We were unable to conduct planned subgroup analyses to explore heterogeneity according to the following study characteristics: the type of intervention as per the logic model and framework (Figure 1 and Figure 2) (e.g. cost to end user); specific population characteristics (e.g. gender/sex, ethnicity, geographic areas, rural/urban, socioeconomic status, education, age/subgroup (children under five years and pregnant women)); and the number and type of interventions in each strategy.
We planned to perform sensitivity analysis considering the relevant issues identified during the review process. For example, we planned to perform a sensitivity analysis to assess whether a difference in parameters used for reporting the use of ITNs affected our findings (e.g. ITN usage reported with the time frame ‘the previous night’ compared to not specifying the time frame). We planned to investigate the following study characteristics: fixed-effect versus random-effects; odds ratios versus risk ratios; studies with imputation versus without imputation for standard deviations; and RCTs versus non-RCTs.
If ITS studies were included, we planned to use a time series regression to reanalyse each comparison when accounting for unit of analysis errors, in consultation with a statistician. However, we did not include any ITSs.
11 Characteristics of included studies
11.1 CHARACTERISTICS OF EXCLUDED STUDIES
11.2 CHARACTERISTICS OF STUDIES AWAITING CLASSIFICATION
11.3 CHARACTERISTICS OF ONGOING STUDIES
12 Additional tables
12.1 INCLUDED STUDIES MAPPED ACCORDING TO THE LOGIC MODEL
Footnotes
20 prenatal clinics were assigned to different subsidy levels: (5 clinics - ITNs free of charge; 5 clinics - KES 10 (97.5% subsidized); 5 clinics - KES 20 (95.0% subsidy); 3 clinics - KES 40 (90% subsidy); 5 clinics = control + lottery on announced random day for additional discount