Factors associated with access to Insecticide-Treated Nets and use of house modification in Bagamoyo and Ulanga districts, Tanzania

Factors associated with access to Insecticide-Treated Nets and use of house modification 1 in Bagamoyo and Ulanga districts, Tanzania 2 Odufuwa OG, Ross A, Mlacha YP, Juma O , Mmbaga S Msellemu DF and Moore 3 SJ 4 1 Ifakara Health Institute, Tanzania 5 Swiss Tropical and Public Health Institute, Basel, Switzerland 6 University of Basel, Basel, Switzerland 7 *Corresponding author: Olukayode G. Odufuwa*: oodufuwa@ihi.or.tz; 8 olukayode.odufuwa2014@gmail.com 9 Amanda Ross: amanda.ross@swisstph.ch 10 Yeromin Mlacha: ymlacha@ihi.or.tz 11 Omary Juma: ojuma@ihi.or.tz 12 Selemani Mmbaga: smmbaga@ihi.or.tz 13 Daniel Msellemu: dmsellemu@ihi.or.tz 14 Sarah Moore: smoore@ihi.or.tz 15


Study area and demographic characteristics
The study was conducted in Bagamoyo and Ulanga districts ( Figure 2). Bagamoyo district is 2 located on the east coast of Tanzania, approximately 60 km north of Dar es Salaam, the 3 economic hub of the country [19] and Ulanga district is located in rural south east Tanzania, 4 300km from the regional city of Morogoro and 500km from Dar es Salaam [20]. In 5 Bagamoyo and Ulanga, the average household size was 4.4 and 4.9 persons per household, 6 respectively in the 2012 Tanzania National Census [21]. Adult literacy in Bagamoyo was 7 58% and 66% in Ulanga [22]. The average rainfall and temperature are 1200 to 2100 mm per 8 year and approximately 28°C in both districts, with slightly higher average rainfall per year 9 and temperature in Bagamoyo due to its coastal location. Above 70% of the residents in both The study is a secondary analysis of two baseline cross-sectional surveys from a trial on 4 spatial repellents in Bagamoyo [25], and an insecticide residual spray (IRS) evaluation in 5 heads of households was categorized into three levels: young adults "18-24", adult "25-49" 1 and old adults "50-99".   The proportion of houses with more mosquito entries (windows and doors) were greater in -91] of the houses in the population were located within a mean distance of 609.5 metres to a 1 8 mosquito breeding site (Table 1).  Both lower than the 80% coverage target of the Tanzania National Malaria Strategic Plan 1 0 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Univariate analysis revealed that on access to ITN was on average lower among households 5 with fewer people and which head of house was younger, male, had little formal education, 6 and poor, as well as in houses that had fewer number of windows and doors, and far from a 7 breeding site in Bagamoyo and only lower among households with fewer people in Ulanga.

8
While in multivariable analysis, all factors estimated were not significantly associated with 9 access to ITNs except household size ( Table 2). The consistent determinant of access to ITN 1 0 was larger household size.

1
Similar to access to ITNs, on average enough ITNs per sleeping spaces was less common in 1 2 Bagamoyo among poor households whose head had no formal education, as well as houses Bagamoyo and Ulanga in the multivariable analysis (Table 3)

1
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2019. ;https://doi.org/10.1101https://doi.org/10. /2019    proportion of houses in both districts with the combination of screened windows and closed 7 eaves that reported use of an ITN the previous night of the surveys was less than 10%.

8
Factors that are associated with the use of house modification in the population 9 In the univariate analysis, use of window screens was lower in poor houses in which head of Ulanga, except that houses with fewer members were less likely to have a window screens.

3
Of these findings, only household size, age and wealth of head of household, mosquito entry 1 4 (higher numbers of doors and windows) and mean distance to breeding sites were strongly 1 5 associated with the use window screens in both Bagamoyo and Ulanga in the multivariable 1 6 analysis ( Table 4).

7
Larger households and those headed by older individuals, female and in the poor quintile 1 8 were less likely to use closed eaves in Bagamoyo and similarly in Ulanga, except that 1 9 households in the rich quintile had low usage of closed eaves in the univariate analysis. mud the multivariable analysis (Table 5). Other co-variates explored were not statistically 2 4 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 2 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   Population access to ITNs after 3 years of mass distribution was lower than targets of the 3 Tanzania National Malaria Strategic Plan. The difference between the access to ITNs in 4 Bagamoyo and Ulanga (64% vs 84%) estimates that access declines at an average of 10% 5 after the first year, this suggests that a minimum interval of two years ITN distribution 6 campaigns is necessary to maintain high coverage of ITN, this is a concern as the average 7 interval between campaigns in Tanzania is at least 4 years. Population access to ITNs is a 8 more robust indicator of measuring ITN coverage [28] and was lower among large 9 households even in Ulanga where the survey was conducted shortly after residents received 1 0 ITNs from the national campaign. This finding has also been observed in the lake region of 1 1 Tanzania where household numbers are on average larger, (Ikupa Akim pers. comm).

2
Therefore our study also demonstrates that limiting the number of ITNs distribution by not measured by ITNs per sleeping spaces (93%) rather than access to ITNs (84%) in Ulanga, 1 7 this demonstrates that more than two people are using a single sleeping space. This is 1 8 consistent as it was found that larger households had more children (< 5 years old), who were 1 9 more likely to share sleeping space with their parent or a sleeping space occupied by more 2 0 than two children. This has been found to decrease the durability of ITNs due to stretching of access in order to maintain effective malaria control seen in previous years due to universal 2 3 ITN coverage campaigns.

4
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2019. ; https://doi. org/10.1101org/10. /2019 Family size is an important factor that needs to be considered when planning ITNs 1 distribution campaigns. The last campaign in Tanzania achieved high coverage in Ulanga, 2 considering the indicator of population access to ITNs, but it is necessary to ensure that the as against discarding while waiting for campaigns, as it has been shown that damaged nets 1 5 remain insecticidal [48]. It is clear, however that more effort is needed in improving access to 1 6 ITNs. Ulanga, a remote rural area with low income and lack of access to window screening 2 2 suppliers. Building materials are likely to be more expensive far from economic centres due 2 3 to transport cost. The majority of the houses in Bagamoyo had opened eaves possibly due to 2 4 low awareness of closed eaves as a tool to reduce indoor mosquito density [12] and high 2 5 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. use of open eaves is an efficient means of cooling houses [11] which is also a plausible 2 explanation for the observed pattern, especially since Bagamoyo has an average temperature 3 2 o C higher and greater humidity than Ulanga due to its proximity to the ocean. Wealthy 4 households in Bagamoyo were more likely to have closed eaves where window screens were 5 also in common use, while in Ulanga, the poorest households were more likely to close their 6 eaves but were unable to afford window screens that gives protection from mosquitoes [50].

7
As window screening is clearly a function of wealth, it would be prudent for screens to 8 receive a subsidy and for insecticide treated screens to be made more widely available as a 9 long-term vector control intervention to protect individual users as well as the community -1 0 through killing vectors that contact the screens. The acceptance of closed eaves in the rural 1 1 areas do not seem worrisome as it is already a common housing structure among modern  Households headed by young individuals had more window screens installed. This group had 1 9 more years of formal education and were wealthier than the older age group, as access to made them more knowledgeable about the importance of the tool in preventing indoor 2 2 mosquito entry and their wealth afforded them the opportunity to install them in their houses.

3
Window screens are useful because they 1) require no compliance, 2) tend to be long-lasting 2 4 and 3) protect all members of a household. This also serves as an indicator that younger 2 5 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2019. ; people are adaptive to the uptake of new tools for controlling malaria. However, the use of 1 window screens alone is not sufficient to control malaria [54]. The incorporation of 2 insecticides into window screens would improve the efficacy of this tool against mosquitoes 3 as it has improved the protective efficacy of conventional bednets and other vector control The study area in Tanzania provides conducive breeding sites based on ecological conditions 9 and human behaviour, hence high mosquito density notable for nuisance biting [58]. Our 1 0 study found that window screens were less common among households at an average of 1 1 609.5 metres to a breeding site. This is because the poorest households were located at closer 1 2 proximity to breeding sites and couldn't afford the installation of window screens, which 1 3 means that poorer households are exposed to higher risk of mosquito bites. It is reported in Our study was affected by non-response bias as shown in Figure 3 and desirability bias, that of ITNs after the study. Therefore, it was necessary to use number of ITNs reported to be 2 4 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2019. ;https://doi.org/10.1101https://doi.org/10. /2019 used the previous night as a proxy for ownership, it was more adequately measured as 1 interviewers asked the use of each ITN in respect to where it was used in the household. This study demonstrates the need to improve access to ITNs especially among larger 1 3 households in the population, due to the fact that access already dropped within 3 years of 1 4 distribution and among larger family even just after recent government distribution. Our The study was funded by the University of NOTRE DAME (UND Fund #261655). The authors declare no competing interest 1 1 [7] Guyatt HL, Gotink MH, Ochola SA, and Snow RW. Free bednets to pregnant women 2 7 through antenatal clinics in Kenya: a cheap, simple and equitable approach to delivery.