IMPACT OF NHIS AMONG PREGNANT WOMEN IN GHANA

DISCUSSION OF RESULTS

5.1 Introduction

The main objective of the present study was to evaluate the impact of National Health Insurance on antenatal care attendance among pregnant women in Ghana. This chapter is concerned with the discussion of results, deduction of insights garnered from subjecting survey data to descriptive and inferential statistics and discussion of those insights in the light of what is known from literature.

5.2 Demography of Study Participants

This section explores the characteristics of the study participants.

Age

From Table 1, it can be observed that the mean age of study participants was 28.70 years with a standard deviation of ±6.81 years, suggesting that the actual average age of study participants was likely to lie between 21.89 years and 35.51 years.

Parity

The computed mean parity of study participants was 3.18, with a standard deviation of 2.13 (Table 1), indicative of the fact that the true parity of study participants lies somewhere between 1.05 and 5.31 children. This further presupposes that Ghanaian women as at 2014 may be less willing to give birth to more than 5 children.

Ad: We offer a range of writing and research services. Find out more!

Income

The computed monthly income of study participants was GHS 102.50, with a standard deviation of GHS 656.75, suggesting wide variations in the amount earned by study participants in a month (Table 1). Clearly, the earning power of the pregnant women greatly differed, a situation which likely may influence decisions to enroll in NHIS. A standard deviation of GHS656.75 further suggests some of the pregnant women earn very little or no incomes. In fact, further analysis showed that a massive majority of 2,126 of the 2,587 pregnant women (representing 82.18%) earn between 0 and GHS 99, with 8.16 % (n = 352) earning between GHS 100 and GHS 499. Only 67 out of the 2,587 pregnant women (2.57 %) were reportedly earning between GHS 500 and GHS 999.

Education

Data sampled showed that majority of the pregnant women representing 52.61 % (n= 1,361) had only a basic education, whilst 191 (7.38 %) and 93 (3.59 %) respectively had secondary and tertiary education (Table 1). Meanwhile, 36.37 % of respondents said they had no formal education. The foregoing shows that majority of pregnant women in Ghana have basic or no education as those with basic or no education accounted for a whopping 88.98 % of the study participants.

Area of Residence

A large chunk of the pregnant women sampled for the study were found to be based in rural areas (66.33 %; n = 1,716) compared to the 33.67 percentage frequency registered for the urban area. The foregoing appears to presuppose that there may be more pregnant women in rural areas than in urban areas.

Furthermore, the majority of the study participants were found to be in Eastern Region (17.36 %; n = 446), followed by Ashanti Region (17.16 %; n = 444).

Employment Status

Most of the study participants indicated that they were employed. Those who were employed accounted for 90.3 % (n = 2,336) of the study participants, while the remaining 9.7 % indicated that they were unemployed.

Marital Status

Most of the pregnant women were observed to be married (72.4 %; n = 1,873). Consensus was the second most frequent marital status which accounted for a percentage frequency of 18.28 % (n = 473). Widowhood as a marital status registered the lowest percentage frequency of 0.43%.

Religion

The pregnant women sampled for the study were mostly Protestants accounting for 52.42 % (n = 1,356) of study participants. This was followed by Islam which accounted for a percentage frequency of 23.85% of study participants.

Ethnicity

Akans constituted the greatest proportion (33.82 %; n = 875) of the study participants, followed by Mole-Dagbani (27.99%; n = 724) and then Ewe (10.90%; n = 282). The Guan ethnic group registered the lowest percentage frequency.

5.3 Estimation of Proportion of Pregnant Women with a valid NHIS Card

This section seeks to estimate the proportion of pregnant women with a valid NHIS card. Table 2 below shows the proportion of pregnant women with valid NHIS.

From Table 2, out of the 2,587 study participants, 1,643 of them indicated that they do hold a valid NHIS card, while 144 said they don't. This means that the proportion of pregnant women with valid NHIS card was found to be 63.51 %. This figure was lower than that observed in a study by Sakeah et al. (2017) who found that the proportion of pregnant women in Kintampo who hold a valid NHIS card was 90.1%. In another study to assess the effect of NHIS on community pharmacies, it was found that less than 25 % of the customers of the majority of the accredited pharmacies hold a valid NHIS card (Adjei, 2012), a figure lower than that reported in this study.

5.4 Determination of Factors Affecting Antenatal Care Attendance

The factors affecting antenatal care attendance is wide and varied. This section seeks to discuss the factors affecting antenatal care attendance. The predicting factors explored were NHIS and socio-demographic/economic factors such as age, parity, income, education, area of residence, employment status, marital status, religion, and ethnicity.

NHIS vs. ANC

From Table 3 above, a chi square of 11.56 was registered for NHIS at a p-value of 0.001. Since the p-value was less than 0.05, the association between NHIS enrollment and ANC was thus considered as significant. This could have been due to the lack of upfront payment that a valid NHIS card affords holders as the NHIS takes care of the cost factor. This finding agrees with the results of a study in which it was observed among other things that for 49% of the women sampled cost was an important factor influencing their antenatal care attendances (Asundep et al., 2013).

Age vs. ANC

A chi-square of 8.41 was recorded for age at a p-value of 0.038. Since the computed p-value was less than 0.05, the association between age and ANC was thus regarded as significant (Table 3). This finding corroborates with the results of Gupta et al. (2014) who found that there exists a significant association between age and antenatal care attendances.

The adjusted odds ratio for age in relation to ANC was found to be 0.94 (95% CI: 0.89-0.98) and it was significant (p = 0.007) as seen in Table 5. This means that age had a significant effect on antenatal care attendance among pregnant women in Ghana.

Parity vs. ANC

A chi-square of 17.64 was recorded for parity at a p-value of 0.224 (Table 3). Since the calculated p-value was less than 0.05, the association between age and antenatal care attendances was thus regarded as insignificant. Nonetheless, parity was found to have a significant effect on antenatal care attendance as evinced by an adjusted odds ratio 1.22 (95% CI: 1.03-1.45; p = 0.007) (Table 5).

Income vs. ANC, Employment Status vs. ANC and Marital Status vs. ANC

A chi-square of 9.38 was observed for income at a p-value of 0.153 (Table 3). The foregoing means that the association between income and antenatal care attendance was insignificant. The association between employment status of the pregnant women and ANC was also found to be insignificant (X² = 0.0006; p > 0.05). These two aforementioned observations could be ascribed to the male centric framework set up in Africa, in which men are generally the heads of family units, with the power and authority to decide for their family members not excluding those concerning their spouses' utilization of maternal wellbeing administrations (White et al., 2013), regardless of the income or employment status of the pregnant women. The foregoing reason perhaps also explains why there was no significant association between marital status and antenatal care attendance as evinced by a chi-square of 8.70 (p > 0.05).

The adjusted odds ratio in relation to ANC attendance for income and employment respectively were 0.99 (95% CI: 0.99-0.99; p = 0.016) and 0.54 (95% CI: 0.21-1.43; p = 0.216) respectively (Table 5). The foregoing therefore suggests that income as a predictor had a significant effect on antenatal care attendance, whilst employment was found to have registered no significant effect on antenatal care attendance.

Consensual as a marital status was observed to have registered a significant effect on antenatal care attendance as evinced by an adjusted odds ratio of 0.40 (95% CI: 1.98-0.82; p = 0.012) (Table 5). Furthermore, the AOR of 0.40 for consensual in relation to ANC attendance suggest that pregnant women in consensual relationships were significantly less likely to attend antenatal care since the AOR of 0.40 was less than 1.

Education vs. ANC

A chi-square of 23.96 was recorded for education (p < 0.05) as seen in Table 3. Since the computed p-value was less than 0.05, the association between education and ANC was thus considered as significant. It stands to reason that the more enlightened a pregnant woman is by virtue of education, the more likely it is that that woman would appreciate the merits of attending antenatal care. Without doubt, this finding lends some credence to findings made elsewhere where lesser academic credentials (Simkhada et al., 2008a) were observed to be connected with reduced ANC visits in reproductive females in rural centres (Gyimah et al., 2006).

A computation of the adjusted odds ratios for the different levels of education investigated showed that secondary education had a significant effect on antenatal care attendance as evinced by an AOR of 6.87 (95% CI: 1.04-45.52; p < 0.05) (Table 5). This presupposes therefore that pregnant women with secondary education as their highest qualification attained were significantly more likely to attend antenatal care than those pregnant women with their highest academic qualification being basic and tertiary education.

Area of Residence vs. ANC

Area of residence vis-à-vis rural-urban classification registered a significant association with ANC (Table 3). This was because the corresponding p-value of the computed chi-square (7.097) was less than 0.05.

Furthermore, adjusted odds ratio computed for rural-dwelling pregnant women in relation to antenatal care attendance was observed to be 0.84 (95% CI: 0.43-1.62; p = 0.596) (Table 5), which implied that area of residence had no significant effect on antenatal care attendance. This result further suggests that pregnant women living in rural areas were less likely to attend antenatal care since the AOR for rural in relation to ANC attendance was less than 1. The foregoing observation agrees with Comfort, Peterson, & Hatt (2013) who observed in their study that pregnant women dwelling in urban centres have a greater propensity to register ANC visits than rural dwellers.

Religion vs. ANC

A chi-square of 15.34 was recorded for religion at a p-value of 0.004 (Table 3). Since the computed p-value was less than 0.05, the association between religion and ANC was thus regarded as significant. This finding corroborates with the results of Makate & Makate (2017) who observed that there exists a significant association between religious beliefs and antenatal care attendances.

Using unadjusted odds ratio, Catholic, Protestant and Islam as religious beliefs were found to be significant predictors of antenatal care attendances. However, after adjusting for odds ratio, only Islam was observed to have a significant effect on antenatal care attendances (AOR = 4.23; 95% CI: 1.36-13.17; p = 0.013).

Ethnicity vs. ANC

Ethnicity registered an insignificant association with ANC (Table 3). This was because the corresponding p-value of the computed chi-square (10.47) was less than 0.05.

The adjusted odds ratios in relation to antenatal care attendance for Gurma (AOR = 0.23; 95% CI: 0.78-0.71; p = 0.010), Mole-Dagbani (AOR = 0.24; 95% CI: 0.90-0.65; p = 0.005), Grusi (AOR = 0.16; 95% CI: 0.38-0.70; p = 0.015) and other ethnic groups (AOR = 0.12; 95% CI: 0.32-0.42; p = 0.001) was found to be significant, suggesting that ethnicity does have a significant effect on antenatal care attendances. This further implies that pregnant women belonging to the Gurma, Mole-Dagbani and Grusi ethnic groups are significantly less likely to attend antenatal care as their respective AORs was less than 1.

In sum, the factors that were observed to be significantly associated with antenatal care attendances were found to be NHIS enrollment, age, education, and area of residence. Meanwhile, factors that were identified to have a significant effect on antenatal care attendance among pregnant women in Ghana were age, parity, income, marital status (specifically consensual relationships), educational status (specifically secondary education), ethnicity (specifically Gurma, Mole-Dagbani, Grusi).

5.5 Determination of Factors Associated with NHIS Enrollment

Although there are various factors associated with NHIS enrollment, some factors are more influential than others. A multivariable logistic regression analysis was thus conducted. This section therefore sought to determine the factors associated with NHIS enrollment. Table 4 examines the effect of socio-demographic/economic factors on National Health Insurance Scheme enrollment using multivariable logistic regression analysis.

Comparing age, parity and income as predictors of NHIS enrolment, the results presented in Table 4 shows that the adjusted odds ratios (AOR) of age, parity and income respectively were 1.01 (95 % CI: 1.00-1.02; p < 0.05), 0.99 (95% CI: 0.95-1.04; p > 0.05) and 0.99 (95% CI: 0.99-1.00; p>0.05). The foregoing implies that age as a predictor was significantly more likely to engender NHIS enrollment among pregnant women in Ghana than parity and income.

From Table 4, the unadjusted odds ratios (UOR) for basic, secondary and tertiary education respectively were 0.82 (95% CI: 0.70-0.95; p = 0.009), 0.78 (95% CI: 0.63-0.97; p = 0.027) and 0.70 (95% CI: 0.52-0.94; p = 0.018). This means that OR for education when left unadjusted is a significant predictor of NHIS enrollment. Among the three levels of education, using UOR basic education was found to be the strongest predictor of NHIS enrollment, followed by secondary and then tertiary education. This means that women with basic education are more likely to enroll in NHIS than those with secondary education and tertiary education. However, using the AOR, basic, secondary and tertiary educational levels were observed to be insignificant predictors of NHIS enrollment.

The UOR for rural area of residence was found to be 1.29 (95% CI: 1.04-1.62; p = 0.024) (Table 4). This presupposes that using unadjusted odds ratio pregnant women living in rural areas are more likely to enroll in NHIS than urban dwellers since UOR was greater than 1 and significant. However, when the odds ratio was adjusted for, the AOR for rural area of residence was 1.16 (95% CI: 0.96-1.47; p = 0.23) which was insignificant.

The adjusted odds ratio for unemployed pregnant women was 1.24 (95% CI: 0.96-1.61; p = 0.093) which was statistically insignificant (Table 4). This means that unemployed pregnant women are more likely to enroll in NHIS than employed pregnant women, but then this difference in likelihood is not significant.

From Table 4, the unadjusted odds ratio for pregnant women who have never been married was 0.80 (95% CI: 0.72-0.92; p = 0.001). This means that using UOR, pregnant women who have never been married were significantly less likely to enroll in NHIS than pregnant women who were married, in consensual relationships, divorced or widowed. Using the AOR, pregnant women in consensual relationships (AOR = 1.18; 95% CI: 0.92-1.53; p = 0.232) were insignificantly more likely to enroll in NHIS than pregnant women who were married, have never been married, divorced or widowed.

Data collected in Table 4 shows that the UOR for Mole Dagbani, Grusi and other ethnic groups respectively was 1.47 (95% CI: 1.15-1.890; p = 0.002), 2.08 (95 % CI: 1.47-2.93; p <0.001) and 1.50 (95% CI: 1.07-2.11; p = 0.020). The foregoing suggests that pregnant women belonging to the ethnic groups, Mole Dagbani and Grusi are significantly more likely to enroll in NHIS than Akan, Ga-Dangbe, Ewe, Guan, or Gurma pregnant women. Between Mole-Dagbani and Grusi pregnant women, Grusi pregnant women exhibited a greater propensity to enroll in NHIS. This is because the UOR for Grusi (UOR = 2.08) was greater than that for the Mole-Dagbani (UOR = 1.47). None of the computed AORs for the different ethnic groups was found to be significant. Using the AOR, Ga-Dangbe pregnant women were observed to be the least likely to enroll in NHIS as it registered the lowest AOR.

The AORs for Central (AOR = 2.39; 95% CI: 1.34-4.25; p = 0.003), Volta (AOR = 2.70; 95 % CI: 1.55- 4.71; p < 0.001), Eastern (AOR = 1.96; 95% CI: 1.16- 3.31; p = 0.012), Brong-Ahafo (AOR = 1.69; 95% CI: 1.07-2.680; p = 0.025), Northern (AOR = 1.73; 95 % CI = 1.02- 2.94; p = 0.044), Upper East (AOR = 3.39; 95 % CI = 1.93-5.97; p < 0.001) and Upper West (AOR = 3.27; 95 % CI = 1.82- 5.88; p < 0.001) regions were statistically significant. This implies that pregnant women from Central, Volta, Eastern, Brong-Ahafo, Northern, Upper East and Upper West regions were found to be more likely to enroll in NHIS than those pregnant women from Greater Accra and Ashanti Region. This observation could be due to the ability of most pregnant women from these regions to afford upfront payment for healthcare services considering the fact that these two regions are arguably the richest in the country.

Of the different forms of religions, only traditional religion was found to have recorded a significant UOR in relation to NHIS (Table 4). The traditional religion showed an UOR of 1.77 (95% CI: 1.08-2.89; p = 0.023). This presupposes that pregnant women observing traditional religion were significantly more likely to enroll for NHIS than those pregnant women who are Protestants, Catholics or Islamic. Using AOR, Protestant pregnant women were the least likely to enroll for NHIS as the Protestant predictor recorded AOR of 0.88 (95% CI: 0.56- 1.38; p < 0.05), but then this likelihood was not statistically significant.

In sum the factors significantly associated with NHIS enrollment were observed to be education, area of residence, marital status, ethnicity, region and religion.

5.6 Estimation of Impact of NHIS on Antenatal Care Attendance

From Table 5, it can be observed that the adjusted odds ratio for NHIS was found to be 2.19 (95 % CI: 1.31-3.67; p = 0.003), which suggests that NHIS enrollment had a significant effect on antenatal care attendance. The foregoing result further implies that pregnant women who are NHIS cards holders are significantly far more likely to attend antenatal care than those who are not. This finding agrees in part with the adjusted odds ratio of 1.64 (95% CI: 1.14–2.38) observed by Sakeah et al. (2017) in a study on the factors that determine antenatal care attendance in rural parts of Ghana.

The impact of NHIS on antenatal care attendance was subjected to further tests using the propensity score matching to determine the causal effect of NHIS enrollment alone on antenatal care attendance. Propensity score matching thus seeks to eliminate bias in observational studies by accounting for confounding variables (such as religious beliefs, marital status, age, parity, area of residence, income level, employment status, educational status, and ethnicity) whilst at the same time seeking to ensure that observations in the treatment groups are truly comparable to observations in the comparison group (Mensah, Oppong, & Schmidt, 2010). To achieve this, the observations in both treatment group and comparison were matched on the basis of the following characteristics: religious beliefs, marital status, age, parity, area of residence, income level, employment status, educational status and ethnicity. A total of 240 observations were matched using propensity score matching analysis, with 120 observations assigned to each treatment and comparison groups.

The fundamental concept underlying statistical matching (Rubin, 1974) is to pin-point in a large reservoir of potential comparison observations a sufficient number of candidates that closely resemble the treated units. "If, conditional on observable characteristics such as marital status, parity and the like, selection into treatment can be regarded as a random event, the contrast between treatment and comparison observations yields, on average, an estimate of the treatment parameter, most specifically the "average effect of treatment on the treated" (ATT)" (Mensah et al., 2010, p.11). For the purpose of this study however, the control group was all observations that have not enrolled for the NHIS, whereas the treatment group was observations that have enrolled for the NHIS. The treatment, NHIS enrollment, was being tested on the outcome variable, antenatal care attendances.

The significant propensity score for NHIS enrollment (No vs Yes) was -0.681 (p < 0.05) (Table 8). The foregoing implies that NHIS enrollment had a significant effect on ANC attendances. This further means that non members of NHIS were significantly less likely to attend antenatal care than comparable NHIS members. Conversely, NHIS members were significantly more likely to antenatal care than comparable non members of NHIS. This finding also agrees with the results from the odds ratio analysis indicated earlier in this section.

We offer a range of writing and research services. Find out more!