Drivers of unhygienic desludging practices in unplanned settlements of Dar es Salaam, Tanzania

Desludging is considered a key step in urban sanitation improvement. Nevertheless, in most urban unplanned settlements, desludging is performed in an unhygienic manner, risking public health. This study used a combination of qualitative and quantitative methods to analyse the underlying factors for unhygienic desludging in three urban unplanned settlements of Dar es Salaam, Tanzania. Factors for unhygienic desludging were analysed using logistic regression analysis and the Integrated Behavioural Model for Water, Sanitation and Hygiene (IBM-WASH), respectively. It was found that 23.7% of the surveyed houses practised unhygienic desludging. Plot physical inaccessibility, outdated laws that stipulate low penalties and the lack of routine data were found to be key contextual factors. Furthermore, perceived inability to control unhygienic desludging among community members and the lack of feeling of shame among those performing unhygienic desludging are the key psychosocial factors. Technological factors contributing to continued practice of unhygienic desludging can be linked to the weakness in existing desludging technologies and general skepticism about the performance of manual desludging hand pumps. This paper concludes that unhygienic desludging in Dar es Salaam is likely to continue unless such factors are considered and integrated into the city’s sanitation improvement plans.


Study area
The study was conducted in Dar es Salaam

Study design
The study employed a cross-sectional study design employing a combination of both qualitative and quantitative methods for data collection and analysis. Qualitative methods involved key informants interview and focus group discussion (FGD) while quantitative method included research household survey and observation of the physical condition in study areas.

Sampling procedure
Quantitative method was employed in undertaking the household surveys. The sample size for the household survey was estimated using the formula by Israel () expressed as n ¼ N=[1 þ N(e) 2 ], where n ¼ sample size; N ¼ total population; e ¼ sampling error. The population used was a projection from the 2012 National Census Report obtained from municipal councils. The total population (N) for Kipawa, Manzese and Keko was 224,140 people. Therefore, with a sampling error of 5%, using the above formula, the sample size for the study was approximated to 400. The sample was distributed among the study areas proportional to the population size. The sample sizes for Kipawa, Manzese and Keko were 133, 124 and 143, respectively. Within the wards, the sample size was evenly distributed among sub-wards and systematic random sampling was applied to obtain the study houses.

Data collection procedures
The household survey was conducted in Kiswahili using the questionnaire customized in the Epicollect5 software and installed into smartphones. The questionnaire comprised mostly closed-ended questions. Pre-testing was carried out to test the usability of smartphones, relevance, and flow of questions in the questionnaire. Data were collected by trained field survey assistants from higher learning institutions who worked under the supervision of the principal investigator.
During data collection, plot accessibility was assessed by measuring the size of the access roads to the houses and was done by pacing. An access road of 4 m width and above was regarded as accessible while a road width of less than that was regarded as inaccessible (Mchome ). The topography assessed the natural landscape of the area, particularly location of the plot, whether located in flat or low land.
Whether the house was located in a flat or low lying area was assessed by observing the terrain in relation to the possibility of parking a vehicle in the area. Lined pit referred to the pit that is strengthened to prevent its wall from collapse.
The assessment of pit lining was based on owner self-reporting. Furthermore, toilets were further categorized as being wet or dry type depending on whether water is required for directing excreta to the pit (wet) or categorized as dry when water is not required for directing excreta to the pit and water is not the dominant anal cleansing material.

Key informant's interviews
Before the quantitative study, interview with 19 key informants from the government and non-government organizations (NGOs) was conducted. Six informants were drawn from government, three were vacuum tanker operators, and ten from NGOs that implement sanitation upgrading projects in Dar es Salaam. A snowball sampling was applied for selection of NGOs until the data collected became saturated.

Focus group discussions
Overall, the study conducted six FGD sessions, two sessions in each study area. Two sessions were planned because of availability of financial resources and time. Plans were made to reach a maximum of 12 participants per session (Cook & Crang ) which would make a total of 72 participants for the entire study. However, four participants missed the sessions. Therefore, in total, 68 participants par-

Data analysis
Data analysis was performed in stages. First, quantitative data were analysed to describe the current desludging situation and associated factors using bivariate analysis with x 2 -test. In order to predict factors associated with unhygienic desludging practices, a model-based approach was adopted where two models were run: the logistic regression model for quantitative data and the IBM-WASH for analysis of qualitative data. With the logistic regression analysis, a model of the form logit(π) ¼ α þ βX þ ε was run. In the formula, π is the probability of using unhygienic desludging methods given a set of explanatory variables denoted as X; β is a vector of regression coefficients, α is a reference group that represents individuals employing hygienic desludging methods in every variable and ε is an error term (Hosmer & Lemeshow ). The dependent variable was modelled against various categorical variables that can be grouped as structural factors (e.g., type of toilet and lining), plot physical accessibility, topography and House rent of up to TZS 30,000 ¼ (US$ 13.12) corresponded to low SES while house rent of TZS (31,000) and above (US$ 13.57) corresponded to higher SES. All quantitative data were analysed using SPSS version 20 and variables were considered significant at p < 0.05.
In the analysis of factors associated with unhygienic desludging using IBM-WASH framework, the study per- The IBM-WASH model is an appropriate analytical framework to use in this case study because it is both multidimensional and multi-scalar.

Ethical considerations
A written research permit with Ref. No. AB3/12(B) was granted by the University of Dar es Salaam. Permission for collecting data in the field was sought from all levels of government administration from the national level (ministry responsible for health and the ministry responsible for regional administration and local government) to the subward level. At the residential house level, the purpose of the study was explained and verbal consent was sought after explaining the purpose of the study, method and expected results. Interviewees were free to terminate the interview at any time they wished. No personal identification was recorded but the GPS coordinates of the house location were taken. were from residential houses occupied by both landlords and tenants, 118 (29.9%) from residential houses owned and occupied by families and 73 (18.5%) were occupied by tenants. All but one reported their highest level of education attained. The majority of the respondents, 240 (60.8%), had primary school education followed by 94 (23.9%) who had attained secondary school education. Others, 29 (7.4%), 23 (5.8%) and 8 (2.0%), had not attended any formal education, attained higher-level education and attained informal education, respectively.

Desludging situation
A total of 190 (47.7%) reported having performed desludging in the past when the toilet was full. As seen in Figure 1, unhygienic desludging is common in all study areas and nearly a quarter of the residents reported using unhygienic desludging methods when their pits were full of faecal sludge. Table 1 shows the association between unhygienic desludging and various factors. Unhygienic desludging was more common in Keko and Manzese than in Kipawa. It is further seen from Table 1 that unhygienic desludging is common in low lying areas and is more common among residents who lack space for vehicle parking within 30 metres.
Overall, in all wards, vacuum tankers were the most common desludging methods (accounting for 75.4%), followed by manual pit emptying ( Figure 2). It is likely that there was under-reporting of flooding out practices. This could be attributable to the fact that many toilets in the communities were observed to be connected with flood outlet pipes. Residents claimed that they are using the pipes for discharging grey water, but it was clear that also FS was desludged this way ( Figure 3).
Determinants of unhygienic desludging using logistic regression analysis Table 2 presents the logistic regression analysis results of predictors of unhygienic desludging. The parameter estimates for most of the selected variables were positive, suggesting a contribution to the unhygienic desludging practice. As further seen from Table 2, lack of space for vehicle parking (including desludging vehicles) was a statistically significant factor, thus suggesting that  which is ten times higher than the penalties the municipalities currently charge.  Habitual level contextual factors: At the habitual contextual level, it was found that there is a reliance on toilets for management of culturally sensitive wastes such as used menstrual pads and underwear ( Figure 5). Solid items are thrown into the pits and eventually clog the desludging equipment, increase desludging time and lead to high energy consumption. Operators were reported to discontinue pumping the pit content when they find large quantities of solid waste. Improvement of desludging in these areas should, therefore, go together with education on the proper disposal and management of solid waste, especially taboo-ridden and culturally sensitive waste such as menstrual waste, which is often disposed of in the latrine pit (Gabrielsson ).

Psycho-social factors
Community level psycho-social factors: 'We need to be informed of an improved toilet that is The comments were given regarding the Gulper and the inability of vacuum trucks to pump sludge from some nonsewered sanitation facilities. Thus unhygienic desludging is resorted to due to the aspirational hygienic desludging methods that are currently lacking.  Habitual psycho-social factors: Timing of unhygienic desludging was also identified as one of the major psychosocial factors at the habitual level contributing to the prevalence of unhygienic desludging. Across study areas, unhygienic desludging is performed during the night and when it rains, taking advantage of the absence of inspections at these times. The practice has had negative consequences for those using MDHPs, as they have a low customer base, making the business unprofitable. The introduction of desludging hand pumps in Keko has ceased because of lack of customers and all Gulper groups that previously had been established were non-functional, thus leaving communities with only two desludging options to choose from: vacuum trucks and manual emptying.

Technological factors
Societal/structural technological factors: At the societal/structural level, it was found that skepticism among institutions on the performance of MDHPs, particularly the Gulper, acts as a significant barrier against their wider promotion.
'I have never seen a manual desludging pump that is effective.