This was a cross- sectional population based survey, individuals stayed in that household for a period of not less than 10 days prior to the day of data collection were enrolled in the study to rule out imported malaria individual cases. Children less that 3 months and individual who were unable to communicate were excluded from the study. The study was conducted in Ilala Municipality councils, one of the five Municipalities in Dar es Salaam Region; administratively the council is divided into 4 divisions, 36 wards and 109 streets. It covers an area of 210 km² with an estimated population of 1,648,861 living in 256,357 households (2017 population projections). Multistage sampling was used to select study areas, three wards out of 36 were selected conveniently based on malaria prevalence. Number of respondents in each study areas was selected based on the probability proportion to size of the total number of the households in each study areas. Individual member of the households who consented were tested using Malaria Rapid Diagnostic test (mRDT) to provide evidence on the presence of asymptomatic individual cases of malaria at the household level. Both positive and negative tested individuals were interviewed using structured questionnaires to provide information on social demographic characteristics, individual risk behaviour for malaria infection. Individuals tested positive to mRDT were treated using first line recommended ACT drug. Observations checklist was used to assess house characteristics for risk exposure to mosquito bite. Tools for data collection were pre tested for validity and reliability. Sample size calculation based on the prevalence of malaria in urban peripheral.

The estimated sample size was 628 individuals with 80% power (a = 0.05). The formula; N = g X Z2 p (1-p)/E2 for single proportion sample size cross- sectional study was used with 10% non response rate. STATA version 14.2 was used for analysis. Bivariate logistic regression was used to determine association between exposure and outcome variables using Prevalence Odds ratio (POR). Variables showed association in bivariate analysis were fitted into multivariate logistic regression model to control for confounders. The variable with p value ≤ 0.05 was regarded as causal factor for persistent malaria transmission in the study areas. Confidentiality was observed during data collection and written informed consent was obtained followed by data collection.

A total of 830 participants were recruited in the study, mean age was 24yrs ±20.4SD. Majority 489 (58.9%) were female, 459 (55.3%) were >18 yrs old, primary or no education were 687(82.8%), farmer or unemployed were 639 (77%). Msongola ward contributed 406 (48.9%). Table 1 below shows demographic characteristics of the study population.

The overall prevalence of malaria infection in the study areas was 4.5%. The prevalence was higher among residents in Msongola (7.4%) than the other wards. It was also high among those <18 years of age (5.9%), those with secondary education and above (6.3%), and among farmers or unemployed (5.5%). The difference in prevalence of malaria by age, sex and education level was not significant (p > 0.05). However for occupation and residence was significant (p< 0.05). Table 2 shows the prevalence of malaria by demographic characteristics of study participants.

Overall, 16.9% admitted to have nets, 83.0% did not have nets., 57.4% had ITNs among those with nets and 19.9% had non-ITNs, and 22.7% didn’t know if their nets were treated or not. Net use was 85.8% among those with nets and 14.9% were not using their nets. Majority of those who reported not to have nets reported their nets were torn beyond repair. The prevalence of malaria among those with nets was 2.8%, and among those without nets was 4.8%. The prevalence of malaria among these not using nets was 15.0%. The difference in prevalence of malaria by ownership and use of nets was not significant. Table 3 below shows the prevalence of malaria by ownership of nets.

The prevalence of malaria was also assessed based on travel history outside, duration of stay outside, outdoor activities after dusks and time of going to bed/sleep (Table 4). Only 6.2% of the study subjects had a history of travel outside Dar es Salaam, and prevalence of malaria among them, was 3.9%, while for those with no history of travel outside Dar es Salaam prevalence was 4.5%. The prevalence was high (6.5%) among individuals with outdoors activities after dusk compared to 3.7% for individuals with no outdoor activities after dusks. The prevalence was high (5.4%) among individuals who going to bed between 23.00 to mid night), compared to 2.9% individuals who go to bed between 19:00 to 21:00 hours. The difference in prevalence of malaria among participants with history of travel outside Dar es Salaam was not statistically significant (p=0.825) and also the difference in prevalence of malaria by duration of stay outside Dar es Salaam was not statistically significant (p=0.87). Table 4 below shows the relationship between malaria prevalence with individual travel history, duration of stay, outdoor activities and time of going to bed/sleep.

(Table 5) below shows malaria prevalence in relations to the presence of mosquito breeding sites near the house and the distance of breeding site from the house. The findings show that, majority (71.7%) of the study individuals reside in households surrounded by mosquito breeding sites and that 95.6% their households were located less than 5 kilometres from the mosquitoes breeding sites. The most commonly types of mosquito breeding sites present in the study areas were; terrace cultivation, small scale rice padding, mud brick holes, small streams, open bore holes and pools of stagnant water. Malaria prevalence among the individuals residing in households surrounded by mosquitoes breeding sites was 5.5% and was 4.4% for the individuals stayed in the households located less than 5 kilometres from the mosquito breeding sites. The Pearson chi square test (x²) was used to test for significance. The results showed statistically significant difference in malaria by presence of mosquito breeding sites around the house (p=0.001). There was no statistically significant difference by distance of the households from the mosquito breeding sites (p=0.833). The relationship between household and presence of mosquito breeding sites, the distance of the households and malaria is shown in Table 5 below.

(Table 6) below shows malaria prevalence by characteristics of the houses where households reside. These characteristics were presence, whether there were open eaves on top of the wall, the type of building/roofing material and window screening. The findings have revealed that 45% of the individuals resided in households with open eaves on top of the walls and 26.1% of the individuals reside in the households with unscreened windows. Malaria prevalence among individuals resided in households with open eaves on top of the walls was 8.3% and was 14.7% for individuals residing in households with unscreened windows. Only twenty people (2.4%) resided in houses whose walls were made up of earth or thatches/grass, and prevalence of malaria among them was 30.0%. The Pearson chi square test (x²) was used to test for significance. The results showed statistically significant difference in malaria prevalence by presence of open eaves on top of the walls (p= 0.000), presence of unscreened windows (p= 0.000), and type of building materials used for walls (earth or thatches/grass) (p=0.000). The relationship between household characteristics and malaria prevalence is shown in Table 6 below.

Generally our study found that, majority of the study participants 48.9% were from Msongola ward, 58.9% of the study participants were female, 55.3% were more than 18 years of age, 82.8% were primary or no education and 77.0% were farmer or unemployed.

Prevalence of malaria by socio - demographic characteristics of study population

Our findings have revealed overall malaria prevalence of 4.46%, higher than the average regional malaria prevalence of 1.1%. Msongola ward had the highest (7.4%), followed by Zingiziwa (3.21%) and Chanika the least (0.49%) prevalence than the other study sites. This finding is consistent with a previous study conducted in Mainland Tanzania which revealed variations in malaria prevalence in urban peripherals of Dar es Salaam Region (12,43). A previous study conducted in Dar es Salaam showed increased risk of malaria infection in the administrative units of peri urban Dar es Salaam Region as compared to the urban centre) 17. Low malaria prevalence observed in Chanika (0.49%)might be due to the fact that Chanika ward is characterized by more of urban setting than urban peripheral.

High malaria prevalence have been observed among the age <18 years (5.9%), which is consistent with the previous study conducted in Mainland Tanzania which showed high malaria prevalence in the school going age group 43. Similarly, a study conducted in Congo showed a relationship between malaria infection and age and high prevalence was also observed in the age above 5 years 53. Malaria Indicator Survey conducted in Mainland Tanzania in 2017 has also revealed age shift of malaria burden from under five to school aged children 69, similar a study conducted in Rwanda showed a relationship between older age group with Plasmodium infection 70. The findings of this study have also shown malaria prevalence of 6.3% for secondary education and colleges individuals. This finding is inconsistent with studies conducted in Burkina Faso and Rwanda, In Burkina Faso the findings showed households with mother who had higher education, had low risk of malaria as compared to households with mother with primary education or no education 71. In Rwanda, the findings showed reduced odds of malaria infections for the households with parents that had secondary or higher education level 70.

Prevalence of malaria by individual risk behaviour of the study population

Only a small number of individuals 141/830 (17%) admitted to own nets, and these included ITNs (54.7%), non-ITNs (19.9%) and 22.7% were not sure if their nets were treated or not.

The prevalence of malaria was low among those with nets than among those without nets (2.8% vs. 4.8%), although the difference was not significant, but higher malaria prevalence among non user of nets (15.0%) and the difference was significant (p = 0.01). This finding is consistent with study conducted in Morogoro Municipality in Tanzania which showed a low prevalence of asymptomatic malaria (5.4%) among 90.6% school children users of ITNs 72.

Study conducted in Central India showed reduction in overall malaria prevalence to 1% for users of ITNs 73. In Tanzania, study conducted in Ifakara showed positive relationship between increased coverage of ITNs and malaria burden and that an increases of 10% ownership of mosquito nets at village level had an average of 5.4% and 10% decrease of malaria deaths in children under five years 65.

Our study showed no relationship between malaria infection with history of travel and duration of stay outside Dar es Salaam (p- value = 0.825 and 0.87). This finding is not consistent with, other previous studies conducted in Mainland Tanzania 37 and also with World Health Organization report 55, which associated mobility, occupation and travel with malaria infection. Study conducted by Martens and Hall also associated population movement (trading, fishing and labouring) with malaria infections 58. Other study associated travels to rural areas as a risk for malaria in urban areas 32. The difference in our findings from other studies could be attributed by a small proportion of individuals (6.3%) who had a history of travel outside Dar es Salaam. Hence implies that individuals who tested positive for this study, got malaria infection in their areas of domicile, not outside Dar es Salaam. Our findings also indicated higher malaria prevalence (6.5%) among individuals with outdoor activities after dusk compared with those with no outdoor activities after dusk, although bivariate analysis revealed no association in malaria risk with outdoor activities after dusk (p- value = 0.93), a study conducted in Dar es Salaam showed high risk of malaria infections among adults participated in outdoor activities after dusk and that, the risk of malaria infection was 1.8 times than individuals who had not participated in outdoor activities 74.

Our findings also observed higher malaria prevalence (5.4%) among individuals who go to bed/ sleep late, than among early sleepers. However, bivariate analysis indicated no association between malaria prevalence and time of going to bed (p- value= 0.84). A previous study conducted in Dar es Salaam, found high proportion of malaria infection for individuals who rested outdoor after dusk (9.7%) 74. The difference in the findings of the two studies could be due to the fact that our study included all age groups, while the other study was done among adults only.

Prevalence of malaria in relationship to the presence of mosquito breeding sites in the study areas

Findings from our study also revealed an association between malaria prevalence and residing in the households surrounded by mosquito breeding sites (p- value = 0.005). This finding is consistent with the study conducted in Gambia 67 and Ifakara in Mainland Tanzania which revealed that distribution of malaria vectors, transmission rates and malaria incidence varies widely depending on the distance of the households from the mosquito breeding sites 75. Also, study conducted in Northern Botswana showed that households close to mosquito breeding sites were more exposed to mosquitoes bites 33. Also in Uganda, study showed households close to mosquito breeding sites were the risk factor for clinical malaria episodes 27.The findings of our study has also revealed that households surrounded by mosquito breeding sites had 5.08 odds of malaria infection compared to households not surrounded by mosquito breeding sites.

Prevalence of Malaria in Relationship with House Characteristics in the Study Areas

The findings of this study revealed higher malaria prevalence among individuals residing in the households with open eaves on top of the walls, and among individuals residing in the households with unscreened windows. While these two variables were shown to be significantly associated with malaria in bivariate analysis, only unscreened windows were still a significant factor in multivariate logistic regression models (p- value 0.001). Our findings is consistency with study conducted in Gambia which also showed that house screening substantially reduced the number of mosquitoes inside houses by 59% hence clinical episode of malaria infection 67, also in Uganda, study conducted showed reduced incidence of malaria episodes in children living in modern homes than in traditional homes 66. Also, the findings have indicated no association between malaria infections and the type of building materials used for walls and roofs (p- value = 0.537). This findings is inconsistent with the study conducted in Northern Batswana, which showed less chance of malaria cases in households residing in houses made up of bricks walls as compared to individuals member households residing in the houses, walls made up of dug and earth or thatch/ grass 33. The difference in the findings with our finding might be attributed by small proportion (2.4%) of houses in the study areas which their walls were made up of earth or thatch/grasses.