Knowledge quintiles of pregnancy danger signs attributable to - PDF document

Knowledge quintiles of pregnancy danger signs attributable to Community Health Workers: evidence from 646 women in three rural districts of Tanzania Amon Exavery 1 , Almamy Malick Kanté 1,2 , Kassimu Tani 1 , James F. Phillips 2 Correspondence: Amon Exavery – aexavery@gmail.com or aexavery@ihi.or.tz 1 Ifakara Health Institute, P.O. Box 78373, Dar es Salaam, Tanzania 2 Heilbrunn Department of Population and Family Health, Mailman School of Public Health, Columbia University, New York, NY, USA 1

Abstract Background This analysis hypothesizes that presence of a trained, paid and supported Community Health Workers (CHW) in a village enhances women’s knowledge of pregnancy danger signs. Methodology This analysis is based on cross – sectional household survey data collected between May and July 2015 for the Connect Project’s end line household survey in Rufiji, Kilombero, and Ulanga districts of Tanzania. Knowledge quintiles of pregnancy danger signs were constructed using principal component analysis (PCA) of pregnancy danger signs reported by the women. Ordinal logistic regression model was used. Results Of the 646 eligible women analyzed, their mean age was 28.4 years. After adjusting for socio – demographic factors women living in villages with CHW were significantly 38% more likely to belong in higher knowledge quintiles of pregnancy danger signs than those living in villages without CHW (OR=1.38, 95% CI 1.05 – 1.83). Conclusion The data supports the hypothesis, thus highlighting the need to train and deploy CHW in our communities in order to enhance women’s knowledge of pregnancy danger signs among others. 2

Background To ensure safe motherhood, we cannot overlook raising women’s awareness on danger signs during pregnancy, childbirth, and postpartum period. This is partly because pregnancy – related complications remain highly ranked among the leading causes of maternal mortality and morbidity in most of the less developed countries. However, most of these complications are preventable if appropriate precautionary measures were taken. In Tanzania, a country with one of the highest maternal mortality ratios according to the recent demographic and health survey report (TDHS-MIS, 2016), knowledge of women on obstetric danger signs requires further research and interventions. This analysis hypothesizes that presence of a trained, paid and supported Community Health Workers (CHW) in a village enhances women’s knowledge of pregnancy danger signs. Methodology This analysis is based on cross – sectional household survey data collected between May and July 2015 for the Connect Project’s end line household survey in Rufiji, Kilombero, and Ulanga districts of Tanzania. Women ages 15 – 49 years who had given birth within two years preceding the survey were included. During the baseline household survey, sampling was conducted using probability proportionate to size (PPS) and details about this sample have been published elsewhere (Exavery et al 2012). Households interviewed for the baseline survey are the same interviewed for the endline. Therefore, the same PPS sampling technique remains valid for the current analysis. 3

The outcome variable for this analysis was women’s knowledge quintile of pregnancy danger signs. The knowledge quintiles of pregnancy danger signs were constructed using principal component analysis (PCA) of pregnancy danger signs reported by the women. This approach, PCA, was opted over looking at each danger sign in isolation because it reveals the internal structure or pattern of the data that best explains the variance in the data. Six pregnancy danger signs were included in the PCA process – reduced/absence of fetal movement, high blood pressure, edema of the face/hands, convulsions, excessive vaginal bleeding, and severe lower abdominal pain. The main independent variable was presence or absence of CHAs in a village. Villages that had CHAs (intervention) were represented by 1 and those that did not have CHAs (control) were represented by 0. The design or the trial randomly assigned 50 villages to intervention and 51 controls making a total of 101 villages studied in three districts. The number of CHAs (for interventions villages) were distributed based on village population sizes, ranging from two CHAs for villages that had less than 1,000 population to four CHAs for villages that had at least 7,000 population. Details about this distribution of CHA is available elsewhere (Kate et al., 2013). Ordinal logistic regression model was used for multivariate analysis. The level of significance was 5%. 4

Results The analysis included 646 women of reproductive age who had given birth two years preceding the survey. Their mean age was 28.4 years, and the majority were in the age group 20 – 34 (57%). Majority also were married (74.8%) and about two – thirds had primary education (68%). With respect to district of residence, 61.2% resided in Kilombero, 22% resided in Rufiji and 16.9% resided in Ulanga (Table 1). Seventy – six percent of the women had knowledge of at least one danger sign, and the rest 24% knew none of the danger signs (Figure 1). Excessive vaginal bleeding was the danger sign that most of the women knew about (35.1%), and the least known danger sign was reduced/absent of fetal movement (17.2%) (Figure 2) The proportion of women belonging to the highest knowledge quintile (Q5) was 21% and 18% in CHW and non CHW villages respectively, whereas so was 20% and 27% in the lowest knowledge quintile (Q1). CHW effect on knowledge quintiles was more obvious in the two highest quintiles – Q4 and Q5 as shown in Figure 3. In the multivariate analysis (Table 2) where we adjusted for socio – demographic factors (i.e. age, district of residence, education, marital status, religion, gravidity, and antenatal care attendance during pregnancy), women living in villages with CHW were significantly 38% more likely to belong in higher knowledge quintiles of pregnancy danger signs than those living in villages without CHW (OR=1.38, 95% CI 1.05 – 1.83). Ever married women (currently divorced or widowed) were 45% less likely than those currently married to be in higher knowledge quintiles of pregnancy danger signs (OR = 0.55, 95% CI 0.32 — 0.97 ). Another significant factor that positively predicted knowledge quintile was education. Women with secondary education were 5

more than two times more likely to be in higher knowledge quintiles than those that had never been to school (OR = 2.14, 95% CI 1.14 – 3.99 ). Discussion and Conclusion The data supports the hypothesis that presence of a trained, paid and supported Community Health Workers (CHW) in a village enhances women’s knowledge of pregnancy danger signs – thus highlighting the need to train and deploy CHW in our communities in order to enhance wome n’s knowledge of pregnancy danger signs among others. This is an imperative component of the birth and emergency planning priorities towards enhanced maternal and newborn survival in Tanzania. Figure 1. % of women by the number of pregnancy danger signs they know (n = 646) 40 35 30 25 % of women 20 15 10 5 0 0 1 2 3 4 5 6 Number of danger signs known 6

Figure 2. % of women with knowledge of each danger sign (n = 646 ) 35.1 29.7 21.4 19.8 18.4 17.2 Excessive Edema of the Severe lower High blood Convulsions Reduced/absent Vaginal face/hands abdominal pain pressure fetal movement Bleeding Figure 3. % of women in each knowledge quintile of pregnancy danger signs by CHW status in Rufiji, Kilombero, and Ulanga districts in Tanzania (n = 646) 30 25 % Women 20 15 Village had no CHW 10 Village had CHW Overall 5 0 Lowest Q2 Q3 Q4 Highest (Q1) (Q5) Knowledge quintiles of pregancy danger signs 7

Table 1. Frequency distribution of respondents Number of Variable Percent (%) respondents (n) Overall 646 100.0 Age <20 103 15.9 20-34 368 57.0 35-49 175 27.1 Marital status Married 483 74.8 Ever married 48 7.4 Single 115 17.8 Education Never been to school 149 23.1 primary 439 68.0 Secondary+ 58 9.0 Religion Muslim 334 51.7 Christian 259 40.1 Trad/unspecified 53 8.2 Ethnicity Ndengereko 103 15.9 Pogoro 86 13.3 Others 457 70.7 Wealth quintile (Household socioeconomic status) Poorest 167 25.9 Poorer 125 19.4 Poor 136 21.1 Less poor 121 18.7 Least poor 97 15.0 Gravidity 1 167 25.9 2-4 274 42.4 5+ 205 31.7 District Kilombero 395 61.2 Ulanga 109 16.9 Rufiji 142 22.0 8

Table 2. Multivariate ordered logistic regression model of the influence of Community Health Workers (CHW) on knowledge quintiles of pregnancy dangers signs among women who recently gave birth in three districts of Tanzania (n=646) 95% Confidence Covariate Odds Ratios (OR) Interval (CI) CHW status Village had no CHW 1.00 — Village had CHW 1.38** 1.05 — 1.83 District of residence Kilombero 1.00 Ulanga 1.06 0.71 — 1.59 Rufiji 1.16 0.78 — 1.71 Education attainment Never been to school 1.00 — Primary 1.45* 0.98-2.13 Secondary+ 2.14** 1.14-3.99 Age group (years) <20 1.00 — 20-34 1.53* 0.96 — 2.43 35-49 1.84* 0.998 — 3.38 Marital status Married 1.00 — Ever married 0.55** 0.32 — 0.97 Single 0.90 0.58 — 1.39 Religion Muslim 1.00 — Christian 0.69** 0.50 — 0.96 Traditional 0.60 0.32 — 1.11 Gravidity 0 1.00 — 1 0.84 0.46 — 1.51 2-4 0.96 0.64 — 1.46 5+ — — Attended antenatal care at least once Yes 1.00 — No 0.41* 0.15-1.14 /cut1 (0.60) (1.32) — 0.12 /cut2 0.23 (0.48) — 0.95 /cut3 1.07 0.35 — 1.79 /cut4 2.08 1.35 — 2.81 **P<0.05, *P<0.10 9