Menu
Menu
Menu
Menu
Background. Although deworming pregnant women is one of the strategies to reduce parasites (roundworms and hookworms) causing anemia and related perinatal and maternal complications, utilization of deworming medication among pregnant women in Cameroon is suboptimal. Comprehensive assessment of individual, household (including women’s autonomy), and community-level factors associated with utilization of deworming medication has not been done so far. Therefore, we investigated the individual/household and community-level factors associated with deworming among pregnant married women in Cameroon. Methods. Our study was limited to pregnant women because they have a greater risk due to increased chances of anemia. We used data from the 2018/19 Cameroon Demographic and Health Survey. Analysis on 5,013 pregnant married women was carried out using multilevel logistic regression. Odds ratios with a 95% confidence interval (CI) were reported. Results. Our findings showed that about 29.8% of pregnant married women received deworming medications. The individual/household level predictors of deworming medications utilization identified in this study were women’s educational level, wealth quintile, and skilled antenatal care. Distance to health facility and region were identified as community-level predictors of deworming medications utilization. Higher odds of receiving deworming medication occurred among educated and wealthier pregnant married women as well as among pregnant married women who had skilled antenatal care or lived in the south region, whereas lower odds were observed among pregnant married women living in the north region. Conclusion. Access to education and economic empowerment of pregnant married women in remote areas and the north region should be the primary focus of the Cameroon government to enhance deworming coverage in the country.
The data used for the analysis in this study were extracted from the 2018/19 Cameroon Demographic and Health Survey (CDHS), which is carried out by the Cameroon National Institute of Statistics (CNIS) in collaboration with the Ministry of Health (MOH) with financial and technical support from the United States Agency for International Development (USAID) and ICF International [38]. The CDHS collects data to produce evidence for monitoring vital population and several health indicators including utilization of deworming medications [38]. In the CDHS, the two-stage stratified cluster sampling technique was applied. In the first stage, primary sampling units (PSUs) or enumeration areas (EAs) were selected from the sampling frame, which was prepared from the recent population census using probability proportional to size [38]. In the second stage, a fixed number of households [25–30] were selected from the selected EAs using a systematic sampling technique [38]. A total of 14,677 women aged 15-49 and 6,978 men aged 15-64 were interviewed from 11,710 households [38]. Detailed descriptions of the methodology used in the survey are explained in the final report of the 2018/19 CDHS [38]. For this study, we used the Individual Recode (IR) file and limited the analysis to a sample of 5,013 pregnant married women. We used IR file because datasets for measuring women’s health indicators such as the utilization of deworming medication are found in that file [39]. In addition, we limited the study to married women because female empowerment factors such as decision-making power were confined to only married women [39–41]. Utilization of deworming medication was the outcome variable for this study. The WHO recommends that pregnant women take a single dose of mebendazole (500 mg) or albendazole (400 mg) after the second trimester [18]. The DHS asked pregnant women who took deworming medication with a birth in the last five years [18, 39, 42]. We categorized and coded responses to binary as “yes” if they took and “no” if they did not take the deworming medication. We incorporated several individual/household and community level explanatory variables based on available evidence on the uptake of deworming medication among pregnant women [19–25, 28, 43]. We incorporated the following individual/household level predictors and coded them as follows: maternal educational level (no education, primary, secondary, higher), husband’s educational level (no education, primary, secondary, higher), women’s occupation (not working, clerical, sales, agricultural self-employed, services, skilled manual, unskilled manual), husband’s occupation (not working, professional or technical or managerial, clerical, sales, agricultural self-employed, service, skilled manual, unskilled manual) religion (Catholic, Protestant, Other Christians, Muslim, Other), sex of household head (male, female), and skilled antenatal care (ANC) (no, yes). The wealth index was coded as poorest, poorer, middle, richer, and richest. In DHS, for measuring households’ economic status, the wealth index is usually computed using durable goods, household characteristics, and basic services following the methodology explained elsewhere [44], and we followed the same procedure. Regarding media exposure (yes, no), we coded yes if the women read newspaper, listened radio, or watched television for at least less than once a week, and no for otherwise. Women’s decision-making power was coded as yes versus no. If the women decided, either alone or together with their husband on all three of decision-making parameters; their own health, to purchase large household expenses, to visit families or relatives, the women considered as having decision-making power. However, if the woman did not decide, either alone or together with her husband, on at least one of the three abovementioned decision-making parameters, the woman was considered as having no decision-making power. The community-level factors included in this study were as follows: distance to health facility (big problem, not a big problem) place of residence (urban, rural), and region (Adamawa, Centre [without Yaoun], Douala, East, Far-North, Littoral [without Dou], North, North-West, West, South, South-West, Yaounde). Others were community literacy level (low, medium, high) and community socioeconomic status (low, moderate, high). In this study, a big problem indicates that the distance from women’s home to health facility (could it be a health center or hospital) to get medical help for herself was problematic. If the women responded as the distance was a big problem, we coded as 1 if the women reported as not a big problem and coded as 0 if the women were reported as a big problem. The socioeconomic status variable was an aggregation from occupation, wealth, and education of research participants who resided in a given community. We further applied principal component analysis to estimate women who were unemployed, uneducated, and poor. A standardized score was derived with a mean score (0) and standard deviation [1]. The scores were then segregated into tertile 1 (least disadvantaged), tertile 2, and tertile 3 (most disadvantaged), where the least score (tertile 1) denoted greater socioeconomic status and the highest score (tertile 3) denoting lower socioeconomic status. Community literacy level was derived from women who could read and write (or not read and write) at all. First, descriptive analysis including frequency distribution of respondents, utilization of deworming medication, and utilization across explanatory variables was conducted. Then, a chi-square test of independence was carried out to select variables that had a significant association with utilization of deworming medications at P value 0.05 cut point. Subsequently, a multicollinearity test was done using variance inflation factor (VIF) for all statistically significant variables at the chi-square test, and we found no evidence of high collinearity among the explanatory variables (Mean VIF = 1.71, Min VIF = 1.03, Max VIF = 3.51). Based on available evidence, a mean VIF less than 10 is acceptable [45, 46]. Finally, four different models were constructed using the multilevel logistic regression (MLLR) technique to assess whether or not the individual/household and community level predictors had significant associations with the outcome variable (utilization of deworming medication). The first model was a null model, which had no explanatory variables, and it displayed variance in the coverage of deworming medication, attributed to PSU. The second model called model I incorporated only the individual/household level predictors and the third model (Model II) included community-level predictors only. The final model, (Model III), comprised both the individual/household and community level predictors. All four MLLR models included fixed and random effects [47–49]. The fixed effects indicated the association between the explanatory variables and the outcome variable and the random effects signified measure of variation in the outcome variable based on PSU, which is measured by intracluster correlation (ICC) [50]. Finally, the model fitness, or how the different models were fitted with the data, was examined using Akaike’s Information Criterion (AIC) [51]. We used the “mlogit” command to run the MLLR models. Weighting was done to take into account the complex nature of DHS data, while the “svyset” command was used for adjusting for disproportionate sampling and nonresponse. The analysis was conducted using the Stata version-14 software (Stata Corp, College Station, Texas, USA). We used publicly available DHS data from MEASURE DHS for analysis of this study. Since the institution commissioned, funded, and managed the survey, further ethical clearance is not required. ICF international ensured that the protocol of the survey was compliant with the U.S. department of health and human service regulations to protect human subjects.
N/A
