Study setting and design
A cross sectional study that that utilised quantitative methods to collect data from selected public and private not for profit HCFs in the GKMA from January to March 2019. The GKMA includes the districts of Kampala, Wakiso and Mukono whose HCFs serve over 14% of Uganda’s population (8). In this study, we considered HCFs at level III or above since these have a core mandate to deliver Maternal, New-born and Child Health (MNCH) services. The study was restricted to public and private not for profit HCFs because these offer affordable MNCH services to majority of the population in the GKMA. In Uganda, the health care system is organised into a four-tier system (i.e., hospitals, health centres of levels IV, III and II) (9). Level II health centres (HC) have a catchment population of about 5,000 people and only provide outpatient care and community outreach services. Level III HCs with a catchment population of about 20,000 people provide basic preventive, promotive, laboratory and curative services. They have limited inpatient capacity mainly maternity and general patient wards. Level IV HCs (catchment population 100,000) provide outpatient and inpatient services, maternity, children and adults’ wards, laboratory and blood transfusion services as well as an operating theatre. General hospitals (catchment population 500,000) provide preventive, promotive, curative, maternity, and inpatient health services and surgery, blood transfusion, laboratory, and medical imaging services. In this study, we considered HCFs at level of HCIII or above since these have a core mandate to deliver MNCH services.
Sample size and sampling procedure
We sampled 60 out of 105 HCFs in the GKMA. In the sampling, we included all public hospitals and HC IVs since these provide MNCH services to majority of the population. High volume private not for profit (PNFP) hospitals and HC IVs were also purposively selected. We selected all the 8 PNFP hospitals, and 2 out of the 4 PNFP HC IVs. PNFP HC IVs with the largest catchment population were considered for the survey. We purposively selected 28 out of 42 public, and 13 out of the 29 PNFP HC IIIs. HC IIIs with the largest catchment population were sampled.
Data collection and measurement of study variables
Data collection was conducted using the validated WASH Conditions (WASHCon) tool on the Commcare mobile data collection platform. The tool, developed by the Centre for Global Safe Water, Sanitation, and Hygiene (CGSW) at Emory University has been used to evaluate WASH conditions within HCFs in low- and middle-income countries including Uganda (10). The WASHCon tool relies on data collected through surveys, observational checklists and water quality testing. Data collection was done using mobile devices. The data was then uploaded into pre-programmed dashboards via a cellular or wireless internet network (not required during data collection). The WASHCon tool has been previously used in a number of studies (11, 12)
For this study, the outcome of the WASHCon tool was WASH service which was categorized as basic, limited or unimproved/no service similar to the JMP WASH service ladders (10). Based on WASHCon indicators, WASH service is a composite variable generated from five variables (water, sanitation, environmental cleanliness, hand hygiene and waste management services). In order to establish the water service, data was collected on source and accessibility, quantity and quality of water. Sanitation service was assessed by collecting data on accessibility to toilet facilities, quantity of toilets and existence of the infrastructure, while for hand hygiene services data was collected on availability of hand hygiene services and availability of associated supplies. Assessment of environmental cleanliness service was based on availability of cleaning supplies, cleaning practices and frequency, and facility hygiene. In order to establish the availability of waste management service, data were collected on segregation, treatment and disposal of healthcare waste.
Using the WASHCon dashboard, evaluation scores were calculated on a scale of 1 – 3 for each of the WASH domains, as well as an overall score that is an average of all the domains. The scores were determined based on the responses to the survey questions, observation checklists, and water quality testing results (appendix 1). These scores were further categorized into basic, limited or improved/ no service. HCFs that scored between 2.8 to 3.0 were classified as basic, and were considered to meet the minimum WASH in HCF requirements or were on track to meet them; HCFs that scored between 1.9 to 2.7 were classified as limited, and were considered to have made some progress towards meeting minimum WASH in HCFs but were not on track to meet them; while HCFs that scored between 1.0 to 1.8 were classified as having no service or unimproved (Supplementary file 1). Such facilities were considered to have made little or no progress towards achieving the minimum requirements for WASH in HCFs (10).
The independent variables included ownership (public vs. PNFP) and level of facility (HC III, HC IV and Hospital).
Prior to data collection, study enumerators received training on the use of the WASHCon tool, quality control and research ethics. The observations and interviews were conducted by trained enumerators who had a minimum of a Bachelor’s degree in Environmental Health Science; Nursing; or Social Sciences. All the study enumerators were supervised to ensure quality control.
Microbial water quality analysis
In order to determine the availability of water services in HCFs, microbial water quality tests were conducted. At each HCF, observations were done to establish the type of water source and availability of water. Observations were followed by collection of duplicate water samples from the maternity ward. Maternity wards were prioritised due to an elevated risk of transmission of HAIs compared to other patient care areas (13). Water samples were collected using Whirl-Pak bags of 100 mls (with sodium thiosulfate to halt chlorine action in chlorinated supplies) and stored on ice until laboratory analysis. All samples were analysed within four hours from the time of collection. Water was tested for faecal coliform, i.e. E. coli using the membrane filtration method (14). Chromocult agar was used for culturing E-Coli at 37°C for 24 hours. Colonies of E-coli (i.e. dark blue to violet in colour) were counted and results recorded per 100ml of sample.
Data management and analysis
The data obtained using the WASHCon Commcare app, preinstalled on a mobile device were uploaded onto a server managed by Makerere University School of Public Health and Emory University CGSW. Forms were synchronized daily by each enumerator. The investigators had access to preliminary results through a pre-programmed dashboard.
Analysis was performed using Stata version 14 (StataCorp, Texas) and R 3.5.2. Descriptive statistics such as frequencies and proportions were used to summarize quantitative categorical data. Continuous data were expressed as means and standard deviations. Classification of WASH service and its five domains into basic, limited and unimproved/no service was guided by the scoring tool shown in appendix 1.