The TSPA surveys use standardized data collection instruments, which are detailed in the 2014–2015 TSPA report [30]. TSPA surveys include four main questionnaires: facility inventory, client observation protocols, exit interviews and health provider interviews. For the current analysis, the facility inventory and the health provider questionnaires data were used to identify the services available and assess the general and specific service readiness at each facility. Interviewer observation protocols and client exit interviews for ANC were used to assess the clients’ perception of the services received. Selected supply-side service availability and facility readiness components (see “Details of measurement” below) were analysed to explore their effects on uptake of IPTp by pregnant women.
Sampling
The TSPA survey used a master facility list representative of the country’s formal-sector facilities. TSPA had a list of 7102 health facilities, 6838 on the mainland and 264 in Zanzibar. A multi-stage sampling technique was used to first select the facilities randomly and thereafter select the health providers and clients for observation and exit interviews: 1200 health facilities (1090 from Mainland and 110 from Zanzibar) were selected. An average of eight health providers at each facility was interviewed; for facilities with fewer than eight providers available, all providers present were interviewed. ANC clients were identified systematically with a maximum of five clients per provider and 15 per facility were observed. Further details are explained in the 2014–2015 TSPA report [30].
Details of measurement
The SARA manual [31] was used to identify variables for availability and readiness. Service availability is defined as the physical presence and sufficient supply of services at the facility and competent workforce for provision of services. Service readiness is defined as the capacity and ability of health facilities and providers to deliver the services [31]. The following variables from three data files were considered to assess the main outcome of interest; number of pregnant women who reported receipt IPTp doses.
Facility data
Facility type (national referral, regional, district, other hospital, health centre, clinic and dispensary), managing authority (public or private), region, services available, whether the facility offered any malaria or ANC services (including the diagnosis and treatment of malaria, the number of malaria and ANC providers available at the facility), and those who had received training, technical support and work supervision, availability of IPTp and national ANC guidelines, and whether SP was in stock at the facility.
Health provider data
Permission to interview, technical qualification of the provider, whether providers diagnosed and/or treated malaria, whether providers provided ANC care, whether providers conducted any malaria lab services, whether providers had received in-service training for diagnosing malaria and for providing IPTp to pregnant women.
Interviewer observations and exit interviews for ANC client data
Whether malaria rapid diagnostic test (RDT) is routinely provided, whether provider discussed the importance of at least four ANC visits, whether DOT for uptake of IPTp was available, whether explanations were given to the client regarding how to take an anti-malarial and the possible side effects of an anti-malarial, the importance of further doses of IPTp and the use of ITNs, the gestational age in weeks as indicated on the ANC card, and client characteristics such as age and education level.
Data management
Since there is very low prevalence of malaria in Zanzibar and IPTp is no longer implemented, only facilities from mainland Tanzania were included in this analysis. Out of 1090 facilities, 12 (1.1%) did not respond, and other 129 (12%) did not offer any ANC services, leaving 949 (88%) for this analysis. From the sample of health providers who were present on the day of the visit, only the 6418 (98%) who consented to be interviewed and reported to provide any client services were included. Client–provider consultations were observed by the interviewers to assess the quality of services provided. In particular, interviewers observed whether protocols and national guidelines were followed and assessed whether providers were prepared to provide the required services. Personal characteristics of the clients were reported in the exit interview questionnaires. Only 3463 (99.6%) pregnant women who had gestational age in weeks indicated on their ANC cards were included in these analyses. Doses of IPTp received were indicated on women’s ANC cards, of which only those doses received IPTp at the facility where they were interviewed were included (95.5%).
Managing authority was defined as public versus private facilities, with parastatal categorized as public and mission/faith-based categorized as private facilities. The 25 mainland Tanzania regions were categorized into six zones as per ministry of health classifications; Central, Coastal, Lake, Northern, Southern Highlands and Western [32]. Gestational age was used to categorize women into trimesters, with women in their 1st to 13th weeks in the first trimester, 14th to 26th weeks in the second trimester and 27th to 44th weeks in the third trimester. The final analyses only included women in their second or third trimester, as Tanzania guidelines do not suggest IPTp uptake before the second trimester. The number of IPTp doses recorded on the ANC card was used to categorize women as not having received IPTp dose or having received at least one IPTp dose. Analysis was performed for both; including and excluding the IPTp dose received on the day of interview. Availability of SP at the facility was categorized into three; never available, available (observed non-expired SP and reported availability) and not available today (not available or invalid).
An availability score was created using the following variables: whether the facility offers any malaria tests, whether the facility offers diagnosis or treatment of malaria, whether providers diagnose and/or treat malaria, whether providers provide any ANC/PNC care, whether ANC providers conduct any malaria lab services, distance of the facility from the clients’ home, and whether the provider offered an ITN to client.
A readiness score was created using the following variables: observation of training manual/poster/job aid for RDT, whether the national guidelines for treatment of malaria observed, whether the provider had received training in diagnosing malaria in adults, providing IPT for pregnant women, and how to perform malaria microscopy, whether the provider had received any in-service training for ANC/PNC or any lab in-service training or malaria microscopy.
Data analysis
Stata 15 software was used to analyse data for this study. Frequencies and proportions of selected availability and readiness indicators were described by managing authority. The main outcome of interest was whether an IPTp dose was received by the women for current pregnancy at the index facility, as indicated on their ANC cards. Generalized estimating equation (GEE) models with logit link and the binomial distribution were used to study the association between selected availability and readiness variables and doses of IPTp. Principal component analysis (PCA) was used to create scores for the availability of malaria services and readiness for the provision of malaria services, for which three tertiles were created (high, medium and low). An alpha value of p = 0.05 was used to determine the statistical significance for all analyses. Factors significant at p ≤ 0.05 in univariable model, confounders that were significantly associated in literature, along with variable of interest were added to form multivariable models. The odds ratio (OR) was used as a measure of association between exposure and outcome variables and is reported with its 95% confidence interval (CI). To take into account disproportionate sampling of facilities by regions a multiplier (weight) was used in all analyses in order to restore representativeness for over/under sampling. Health facility level (national referral, regional, district, another hospital, health centre, clinic and dispensary) was controlled for, in the multivariable analysis, to give a better estimate of the true difference between IPTp uptake at public and private facilities.