Assessing health system responsiveness in primary health care facilities in Tanzania

Background: Health system performance is one of the important components of the health care delivery; its achievement depends on the quality of services rendered and the health system responsiveness of its beneficiaries. Health system responsiveness is a multi-dimensional concept and is usually measured through several domains. Health system responsiveness (HSR) remains to be a key indicator for evaluation of health system performance in any settings. This study aimed at assessing the situation of health system responsiveness in primary health facilities in Tanzania prior to introduction of the Direct Health Facility Financing (DHFF) program. Methods: This was a cross sectional study conducted between January and February in 2018. We collected data from 42 primary health facilities (14 health centers and 28 dispensaries) where a questionnaire was administered to a total of 422 participants. The questionnaire collected information on attention, respect to dignity, clear communication, autonomy, access to care, respect to confidentiality and basic amenities. Descriptive analysis was done to determine the distribution of the variables whereas ANOVA and linear regression analysis was employed to discern the association between variables. Results: More than 67% of participants had visited the same health facility more than 5 times. Sixty seven percent of the patients were residing within 5kms from the public primary health care facilities. The geographical access to health care scored the lowest (43.5% for Dispensaries and 36% for Health center) mean as compared to other domains of health system responsiveness. The highest score was in respect to confidentiality (86.7%) followed by respect to dignity (81.4%). Linear regression analysis revealed no statistical association between any of the social demographic features with the overall HSR performances. However, in post hoc analysis, Pwani and Shinyanga regions didn’t differ significantly in terms of their performances whereas those two regions differ from all other regions. Conclusion: Based on the study findings health system responsiveness domains has performed relatively poor in many regions except for respect of dignity and confidentiality scored high of all the domains. Shinyanga and Pwani regions scored relatively well in all domains this could have been due to the effect of Results Based financing (RBF) in the respective regions. All in all the Government and other stakeholders in the health sector they should deliberately invest on the access to care domain as seem to be a challenge as compared to others. In order to understand the Health system responsiveness in primary health facilities scattered in different geographical zones of the country, this study was conducted in seven regions namely Mbeya, Dodoma, Pwani, Shinyanga, Katavi, Manyara and Mtwara. The regions represent the seven zones of the country and comprise of 27% of the Tanzanian population. These regions were selected from respective zones so that to seek for generalizability of the study findings as there is geographical and cultural variation across the country. Approximately 70% of the Tanzanian population reside in rural areas and mainly depend on primary health facilities to address their health needs. The study included 42 health facilities (14 health centers and 28 dispensaries) located in 14 local government councils. The primary health facilities in Tanzania are divided into dispensaries and health centers. Dispensaries provide a basic range of preventive, health promotion, curative and maternal and child health (MCH) care, and health centres, offer inpatient and a higher level of delivery care and staffed by a wider range of more qualified health workers between 39 up to 52 than in the dispensaries which have between 15 up to 20 health service providers [19]. District hospitals serve as a referral at the primary health care level. There are some of the programs being implemented in some selected regions such studies are: – Results Based Financing (RBF). RBF is defined as “a cash payment or non-monetary transfer made to a national or sub-national government, manager, provider, payer or consumer of health services after pre-defined results have been attained and verified. Payment is conditional on measurable actions being undertaken [20]. These payments are usually made to health providers after performance of the pre-defined results from selected quantitative and qualitative indicators. Before commence of the RBF implementation, each facility has to develop a business plan, which act as guide during the course of implementation. A Business Plan is a quarterly work plan of the facility, which shows the targets to be reached and strategies required to reach the targets through identified qualitative and quantitative indicators. It is a tool to help health facility staffs and stakeholders to develop their ideas and innovations to improve their efficiency. This study employed a cross sectional study design. The study was done between January and early February 2018. Sampling was done using a four-stage sampling approach. The first stage included random selection of seven regions from the seven regions of Tanzania, located in seven geographical zones (each zone constituted between 3 and 4 regions). In the second stage, selection of district councils was done and two district councils basing on stratification were selected into one urban and one rural, in their respective regions. The third stage comprised of selection of health facilities to be included into the study, they were selected at random from each strata of each district council in the 7 regions. A total of 3 primary health facilities were selected randomly from each district’s list of each type of Public Primary Health Care Facilities (PPHCF) (i.e. 1 Health Centers and 2 Dispensaries) (http://hfrportal.ehealth.go.tz) [21] i.e. Making a total of 42 health facilities (14 health centers and 28 dispensaries). The fourth stage included selection of the participants to participate in the study; the exit interview patients were conducted after they have received the services on their way to their homes. Respondents eligible for interview included all exiting patients or relatives of patients in case the patient is a minor (aged below 18 years), and stratified gender sampling was conducted to ensure that; there are an equal number of men and women in the study in order to ensure gender representation. The sample size for patients/clients to participate in the study was determined by using the Cochran formula (1977) [22] with a 50.0% probability of the responsiveness of patients to primary health facilities, a α – error of 5%, an 80% power and a 95% confidence interval [23]. The calculation indicated a sample size of 384 with a 10% non-response contingency being added, making the total sample size required 422 patients. Each patient or relative of patients were systematically selected for an exit interviews after receiving medical consultations basing on their gender stratification. Ten patients were interviewed per each primary health care facility. Respondents eligible for interview included all exiting patients or relatives of patients (aged above18 years), and were sampled to ensure an equal number of men and women. A closed-ended structured questionnaire was adapted from the health systems responsiveness questionnaires used in the WHO multi-country studies [4, 24, 25]. This standardized d questionnaire had 37 health system responsiveness closed-ended Likert scale questions that were grouped under 7 domains that have ordinal response categories [24]. The 37 questions were divided among the seven domains of responsiveness namely: prompt attention (7 questions), respect for dignity (3 questions), and communication (7 questions), quality of basic amenities (10 questions), respect to confidentiality (3 questions), access to care (4 questions) and autonomy (3 questions) (Additional file 1). The questionnaire was administered to systematically selected patients (10 patients per health facility) exiting the health facility, to measure their experiences with health care services. To ensure accuracy of the collected information, research assistants underwent 4 days training on data collection using mobile phone devices (Tablets) before taking part in pre-testing of the questionnaires. All the questionnaires were then installed into the designed application in the mobile phone. All selected primary health facilities had GPS coordinates and all the data enumerators used tablets with GPS sensors so that to increase data integrity [26]. Mobile phone (Tablets) had a web-based interface that allows real-time gathering of data and the first author to monitor the data collection exercise on daily basis. After the actual field survey, collected data was then sent directly to the Gmail account app (which acted as a server) after being filtered in the field. For this study, the database (data collection software) was developed to which all the data obtained from the study units were entered. The patient survey data was captured on mobile phone then entered into a pre-designed database. The collected data were transferred into a Microsoft excel Database, and then exported to Statistical package for Social Sciences (SPSS) version 25 for statistical analysis. Data cleaning was undertaken before statistical analyses. In this study socio-demographic and health system responsiveness variables were measured from the questionnaire as detailed below. These included; age, sex, educational status, marital status, number of visits to the primary health care facilities, distance covered by a patient to get the health facility, and family size. In this study, age was measured in years, sex was categorized into male or female and marital status was categorized into single or married. We measured distance covered by a patient to access health care in kilometres whereas family size was measured by number of household members and number of health facility visitation was measured in days (Table 1). Respondents’ Social Demographic characteristics (n = 422) Measures the non-health aspect of care relating to the environment and the way healthcare services are provided and relates to a system’s ability to respond to the legitimate expectations of potential users or clients. In this study health system Responsiveness mainly focused on the seven domains that are: attention, autonomy, and basic amenities of care, access to care, clear communication, confidentiality, and respect of dignity. All questions were Likert scale in nature and grouped under seven domains. Each domain was measured by using a mean score and then they were compared among health facilities and among regions (Table 2). The internal consistency reliability of the overall responsiveness scale (37 items) was calculated and average Cronbach’s alpha for all seven domains was 0.827. Health care responsiveness performance criteria and their Categorization The first step did include conducting descriptive statistics (frequency, percentage, mean and standard deviation) analysis of the health system responsiveness of all seven domains. Analyses of health system responsiveness scores for all variables of socio demographic were conducted. The Health System Responsiveness was analyzed basing on the primary health facilities user’s experiences as shown in the four points Likert scales. Each point of the Likert scale was in the percentage and the answers were then dichotomized for further analysis for example good and very good as ‘Good’ and bad and very bad as ‘Bad’. The Likert scale rating for each domain was matched with the responsiveness performance categories as ‘unacceptable’ (Fail) and ‘acceptable’ (Good and Very Good) (Table ​(Table2).2). For instance, the corresponding code for response for basic amenities domain was four that was multiplied by ten (the number of questions in the domain) that produced a cut-off score of 40 that makes a maximum score (“Acceptable”) for an individual for this Domain, and one multiplied by 10 (number of questions) making minimum score (“Fail”) of 10 for each individual who responds to all questions for this Domain (Table ​(Table2).2). This approach is similar to that was used in another study conducted in Ethiopia in 2017 [31]. A total of 37 questions were included to assess health system responsiveness in the primary health care facilities in Tanzania. Four points Likert scales question items ranging from 0 to 3 for five domains (attention, dignity, communication, autonomy and confidentiality) in which 0 represented absence of the assessed feature of HSR and 3 denoting the highest level of its availability. On the other hand, 1 to 4 points were used for two domains (access to care and quality of basic amenities) with 1 score indicating the least performance of the assessed HSR feature and 4 for the highest level of the availability of the features. In total, a minimum of HSR score was computed as 14 out of the maximum score of 125 for all 37 questions (Table ​(Table2).2). Total score for each domain was computed in percentage by taking the actual score obtained from each respondent divided by the maximum possible score multiplied by 100%. Similarly, the overall HSR score was computed by dividing the overall total scores over the maximum possible value (125) multiplied by 100%. The second step was to conduct the inferential statistics. In order to conduct some inferential statistics, the basic assumptions for normality test were conducted. Visual inspection of histogram Q-Q and Box plots (graphs) was done. In addition, skewness and kurtosis z-values and Shapiro-Wilk test for dependent variables were also conducted. Visual inspection of histograms indicated that dependent data distribution were along the straight line for Q-Q plots for both dispensary and health centre whereas symmetrical feature was observed on box plots for both dispensary and health centre. Shapiro Wilk test showed p ≥ .005 (p = .694 and .828 for dispensary and health centre, respectively). Skewness and kurtosis z-values were within the range of − 1.96 to + 1.96 for both dispensary and health centre (dispensary = −.0599, .543 and health centre = −.245, .543). All this suggests that data were approximately reasonably normally distributed. Therefore, parametric tests for inferential statistics were considered relevant for performance comparison. Was used to compare the means of more than two groups especially the regional level comparisons that has tried to display the mean of each region allowing for comparison of overall performance on different assessed aspects (domains). Was used to explore the predictor power of each independent variable on a dependent variables specifically demographic information and perceived responsiveness. It was also used to assess the power of predictors for institutional factors (e.g. population, staffing level, number of beds etc) with the perceived responsiveness.