Menu
Menu
Menu
Menu
Background: Ghana’s National Health Insurance Scheme has improved access to care, although equity and sustainability issues remain. This study examined health insurance coverage, type of payment for health insurance and reasons for being uninsured under the National Health Insurance Scheme in Ghana. Methods: The 2014 Ghana Demographic Health Survey datasets with information for 9396 women and 3855 men were analyzed. The study employed cross-sectional national representative data. The frequency distribution of socio-demographics and health insurance coverage differentials among men and women is first presented. Further statistical analysis applies a two-stage probit Hackman selection model to determine socio-demographic factors associated with type of payment for insurance and reasons for not insured among men and women under the National Health insurance Scheme in Ghana. The selection equation in the Hackman selection model also shows the association between insurance status and socio-demographic factors. Results: About 66.0% of women and 52.6% of men were covered by health insurance. Wealth status determined insurance status, with poorest, poorer and middle-income groups being less likely to pay themselves for insurance. Women never in union and widowed women were less likely to be covered relative to married women although this group was more likely to pay NHIS premiums themselves. Wealth status (poorest, poorer and middle-income) was associated with non-affordability as a reason for being not insured. Geographic disparities were also found. Rural men and nulliparous women were also more likely to mention no need of insurance as a reason of being uninsured. Conclusion: Tailored policies to reduce delays in membership enrolment, improve positive perceptions and awareness of National Health Insurance Scheme in reducing catastrophic spending and addressing financial barriers for enrolment among some groups can be positive precursors to improve trust and enrolments and address broad equity concerns regarding the National Health Insurance Scheme.
The 2014 GDHS datasets for women and men were analyzed. Demographic Health Surveys (DHS) ensure national representation by employing a two-stage sample design across all geographical regions. A total of 427 clusters were selected for the survey which comprises 216 urban and 211 rural areas from enumeration areas (EAs) defined by the 2010 Population and Housing Census. A total of 12,831 households across the 10 regions in Ghana were selected for the 2014 survey. For the individual-level data, women in reproductive-age women (15–49 years) were included as well as men aged 15–59 years. Survey data for women aged 15–49 were collected in 11,835 occupied households, while data for men aged 15–59 were collected in half of all sampled households [19]. No reason was provided in the original data report why more women were sampled than men. Both women and men datasets contain information on respondents’ background characteristics, HIV testing and knowledge, anthropometric measures (height/weight), anemia status, fertility preferences, child health outcomes and health insurance measures. The women dataset also contains data on reproductive history and maternal and child health outcomes not included in the male dataset [19]. In the interviewed households, a total of 9656 eligible women were identified. Interviews were however conducted among a total of 9396 women, providing a response rate of 97%. In addition, 4609 eligible men were identified while 4388 of them were interviewed, providing a response rate of 95% [19]. To ensure data comparability men and women aged 15–49 were analyzed, i.e. men aged 50–59 were excluded. Thus, the total study samples consisted of 9396 women and 3855 men. Three outcomes were assessed: health insurance coverage, type of payment for insurance and reasons why some individuals were not registered for health insurance. The three outcomes were of interest because previous literature indicated that NHIS coverage and not being enrolling to the NHIS vary across population groups and regions [1, 20], which we investigated further in our analysis. Health insurance coverage was dichotomized as 0 = no for not covered and 1 = yes for those covered. Regarding type of payment, the question, who paid for national health insurance membership was applied. This was recoded into 2 categories based on the original dataset; 1 = paid self; 0 = paid by others (relative/friend; employer; the state/exempted). Regarding the third outcome measure, the question, why not registered with national health insurance was asked. Three dummy variables were created: cannot afford premium (0 = no, 1 = yes), do not trust the national health insurance (0 = no, 1 = yes) and do not need health insurance (0 = no, 1 = yes). Thus, we used 5 dependent variables in total. The inclusion of independent variables is based on previous literature on NHIS enrolments in Ghana [6, 13–17]. Eight socio-demographic level variables were included in the analysis: age, marital status, a wealth index, region, educational level, religion, place of residence and parity levels. Maternal age was categorized in five-year group intervals (15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49). Male age groups included 2 additional groups; 50–54, and 55–59 in addition to what is categorized for women. To ensure uniformity in comparisons, only respondents aged between 15 and 49 years were included. For both men and women datasets, marital status was recoded into three responses; never in union, married and widowed. An available wealth index in the dataset was used (poorest, poorer, middle, richer and richest). Region (10 geographical divisions) and educational level (no education, primary, secondary and higher) as coded in the dataset was used. Religion was recoded into responses (Christianity, Islam, Traditional and no religion), parity was recoded into responses; nulliparous, 1–2 births, 3–4 births and 5+ births. Urban and rural classification was used for place of residence. First, descriptive analysis for socio-demographic characteristics for both male and female and health insurance coverage was performed and results were presented in the form of frequencies and a graph. Further statistical analysis was performed on the dependent outcome variables of interest, namely type of payment for NHIS (paid by self and paid by others) and the three reasons for not being insured under the NHIS (cannot afford NHIS, yes/no, do not trust NHIS, yes/no and do not need NHIS yes/no). We used a two-stage probit Hackman selection model. We applied Heckman selection in this estimation to control for selection bias for type of payment (insured group) and reasons not insured (uninsured group). Thus, the selection equations for type of payment was NHIS status (1 = covered and 0 = not covered) and for reasons for being uninsured, NHIS status was recoded as 1 = not covered and 0 = covered. Statistical significance threshold of p < .05 and lower was applied in all analysis. Education status was used as instriúmental variable based on the preliminary anaylsis. Software package Stata version 14 was used to perform the analyses.
N/A
