Menu
Menu
Menu
Menu
Background The contraceptive prevalence rate in Mozambique was estimated as 11.3% in the last Demographic and Health Survey. The impact of family planning (FP) on women’s health and on the reduction of maternal mortality is well known. Methods Acknowledging the importance of user satisfaction in the utilisation of health services, exit interviews were used to assess women’s satisfaction with FP services in Mozambique. The survey, conducted in 174 health facilities, was representative at the national level, covered all provinces, and both urban and rural areas. Results Overall, 86% of respondents were satisfied with FP services, but issues such as insufficient supplies of oral contraceptives and the low quality of healthcare provider/client interactions were given as reasons for women’s dissatisfaction. Conclusion Defined actions at the level of health service provision are needed to tackle the identified issues and ensure improved satisfaction with, and better utilisation of, FP services in Mozambique.
This study was conducted in public primary level health facilities in both urban and rural areas across 11 provinces throughout the country. In 2011, Mozambique had a total of 1386 healthcare facilities, of which 1338 offered FP services on a regular basis. The sample of health facilities was drawn from all facilities that routinely offered FP services. A representative sample was selected, stratified by province and type of health facility. The health facilities were divided into four groups: ‘Class 1’ – District Hospitals, ‘Class 2’ – Health Centers Type I, ‘Class 3’ – Health Centers Type II and ‘Class 4’ – Health Posts. Of the total eligible health facilities, all 23 facilities from Class I were included and 151 facilities were selected from the remaining 1315 facilities, producing a final sample size of 174 health facilities. According to the relative weighting of the different types of health facilities providing FP services in each province, the total sample size was proportionally allocated in each of the 11 provinces. Class 2, 3 and 4 health facilities were randomly selected within each province. Women were sampled by selecting users who had left or were about to leave the FP service site on the interview day. The estimated number of women selected and interviewed in each facility was calculated according to the average number of clients seen in the last 7 days. The women to be interviewed were selected systematically by selecting every mth woman who was about to leave the FP services, where ‘m’ was based on the average number of consultations per day over the previous 7 days. A questionnaire was adapted from one used by the United Nations Population Fund and pre-tested locally. Data collectors were selected from maternal and child health nurses and trained in the use of the tool and data collection techniques. The data collection occurred in December 2011. A database was created using the Statistical Package for the Social Sciences (SPSS) V.17.0 (SPSS Inc., Chicago, IL, USA); this software was used for data entry. Ethical approval for the study was granted by the National Health Ethical Committee of Mozambique. To measure women’s satisfaction a composite variable ‘general satisfaction’, with an ordinal scale, was constructed by summing the ratings of the following three questions: (1) Did the woman feel well treated by the health provider? (2) Was the woman satisfied with the overall services and care in this health facility? (3) In the woman’s opinion, besides the nurse or doctor, did the other staff treat her well? These questions were rated as follows: −1 if No, 0 if failed to answer, and 1 if Yes. If the sum resulted in a negative value, then the user was considered to be ‘dissatisfied’ (No), between 0 and 1 as ‘somewhat satisfied’ and above 1 as ‘satisfied’ (Yes). Satisfaction was then analysed against the client and facility factors such as type of healthcare facility, the woman’s age, quality of provider/client interaction, user’s occupation, access to information on FP, duration of using contraceptive method, and waiting time at the health facility. The quality of provider/client interaction was a composite variable obtained by summing 10 different questions which were all binary coded. The resulting values were categorised into low (0–3), medium (4–6) and high quality of interaction (7–10) based on the number of positive answers. Although ordinary baseline logistic models can be applied to an analysis of the probability of categorical outcomes, they are not appropriate for ordinal outcomes such as satisfaction since they ignore the natural order of the categories or levels of the variable, resulting in a considerable loss of statistical power. Walker and Duncan,15 and later McCullagh,16 proposed the use of proportional odds models for the analysis of ordinal outcomes. A multiple logistic regression using a proportional odds assumption was applied to assess the effect of the covariates on general satisfaction (Table 3). Probabilities were summed from No to Yes across the satisfaction scale. We tested the proportional odds assumption using a Score test, which failed to reject it at the 5% significance level. Odds ratio estimates for the general dissatisfaction of family planning service users in Mozambique CI, confidence interval; OR, odds ratio; Ref., reference; STI, sexually transmitted infection. In this model, cumulative probabilities are expressed as functions of the explanatory variables. One key assumption of this model is that the regression coefficients or odds ratios (ORs) are identical for each of the k (k=1, 2, 3) cumulative probabilities, hence the name ‘proportional odds model’. The model for the kth category can be expressed as follows, for k=1, 2, where y represents the outcome variable (satisfaction) with three levels; αk represents the intercept term, which is ordered to reflect the order of the cumulative probabilities, x is a matrix containing the covariates or factors, and β is a parameter vector containing regression coefficients associated with the covariates in x. In this model, an OR can be obtained by exponentiation of the estimates in the vector β. This model was fitted using Statistical Analysis System (SAS) V.9.2.
N/A
