Study design
The original study was a cross-sectional design conducted among 1200 sample of health facilities, 6866 health workers and 4007 pregnant women across Tanzania. However, for the purposes of this study, variables were constructed from the primary data based on the WHO’s service availability and service readiness assessment (SARA) tool [14] and using indicators of integration as offered in a previous study of Mallick et al. [13].
Procedures for variable construction
Secondary variables were constructed as a composite of the primary variables via adopting and adapting indicators of integration as defined by Mallick et al. [13]. This included the outcome variable under measure: pregnant woman’s receipt of integrated care;
‘Trained staff’ was constructed as composite of ANC health staff that had recently received training in an infectious disease area. For example, to be classified as technically trained in HIV, the health worker should be the first one that offers ANC service and secondly should be trained in HIV counselling and testing services.
Another indicator, ‘availability of disease service’ was constructed as a composite indicator of having both the ID service and ANC at least three working days within the week [13]. ID service on the same site with ANC was constructed having the disease service at the same place as the antenatal care.
‘Availability of diagnostics’ was classified as a composite indicator of all essential diagnostics of ID service available at the ANC facility [13]. Another indicator, ‘availability of medicines and commodities’ were also constructed as having all essential medicines for each ID service during all times of the ANC facility.
After constructing these individual integration components, a final composite indicator score of all these components was constructed to measure the readiness of facilities to integrate these disease services. This indicator was then disaggregated by disease type.
Outcome variable
The main outcome variable for the study was receipt of integrated care of infectious disease services (HIV, TB and Malaria). This was based on the hypothesis that if facilities had all five components of integration present, then pregnant women should receive an integrated care during antenatal sessions. The construction of integrated care (a composite indicator) was based on three parameters; first pregnant women’s receipt of insecticide treated nets (ITNs) during antenatal care, second, observation of pregnant women’s consumption of antimalarial (IPTp-SP) during all antenatal sessions after 16 weeks, third was pregnant women’s receipt of HIV counselling and testing during first antenatal care. As guided by the ANC policy guidelines [15], no outcome for tuberculosis was included in this composite outcome variable.
Data sources
The main source of data for this study was the service provision assessment (SPA) conducted at facilities in both the mainland Tanzania and Zanzibar (the island) in 2015/2016 [16]. The data were collected under the auspices of National Bureau of Statistics (NBS) and the office of the Chief government statistician in Zanzibar and funded by USAID under the demographic and health survey (DHS) program.
The Tanzanian SPA was a survey of availability of services in facilities which spanned from those owned by government, private, parastatal to faith-based organisation (also termed as mission facilities). Also, the survey originally collected information on facilities at all levels of care (i.e. from primary, secondary to tertiary facilities at district, regional and national level). The areas covered during the survey included maternal, neonatal and child health, family planning, sexually transmitted disease services, antenatal care, non-communicable diseases (i.e. cardiovascular diseases, diabetes 1 and chronic respiratory diseases). For each of these disease service areas, the SPA measured availability and functionality of the services which included a quality assessment of equipment, diagnostic tools and essential medicines.
Data analysis
SPSS (version 20) and Microsoft Excel (2013) were used for all analysis in this study. Five main steps were involved in the analysis. Each of these are explained below.
First, a background characteristics of all ANC facilities and availability of infectious diseases were computed. These were disaggregated by facility type and location.
Second, using the ‘countif’ excel formula, all composite indicators for each ID service area was calculated. The data was then processed in SPSS.
Third, associations and emerging relationships between integration and facility characteristics were explored using an appropriate test. A chi-squared test was used to determine the association between each of the five components of disease integration and background characteristics of facilities.
Fourth, all the five indicators of integration were put together as a single indicator. For pragmatic reasons, this indicator was known as overall integration score for each infectious disease area. This indicator was computed for each of the three disease areas (i.e. malaria, TB and HIV).
Finally, for the association between the components of integration and the outcome variable, a logistic regression model was created to report the odds of association. The model was adjusted for all sociodemographic characteristics and facility/provider characteristics that could have a potential cofounding effect on the results.
Significance level for all tests were pegged at 95% confidence level (i.e. p < 0.05).
Ethical consideration
The original service provision assessment was approved by the National institute for medical research, Tanzania and the macro institutional board of Inner City Fund (ICF) International. According to the primary report of the survey [16], all participants signed a consent form. All participants were fully aware of the implications of engaging in the study including a possibility to have a secondary analysis like this study. The datasets are confidential and do not in any way expose study participants in a manner that affects them. For this particular study, ethical approval was granted by the Institute of Global Health and Development at the Queen Margaret University. For the purposes of protecting identities of the study participants and/or specific facility, all analysis were conducted from district to national level only.