Open access Research Is enrolment in the national health insurance scheme in Ghana pro-poor? Evidence from the Ghana Living Standards Survey Eric Nsiah-Boateng,  1 Jennifer Prah Ruger,2 Justice Nonvignon3 To cite: Nsiah-Boateng E, AbstrACt Prah Ruger J, Nonvignon J. Objectives This article examines equity in enrolment in strengths and limitations of this study Is enrolment in the national the Ghana National Health Insurance Scheme (NHIS) to health insurance scheme in ► Our study is the first to use data from the Ghana inform policy decisions on progress towards realisation of Ghana pro-poor? Evidence Living Standards Survey to examine equity in en- universal health coverage (UHC). from the Ghana Living rolment in the National Health Insurance Scheme Standards Survey. BMJ Open Design Secondary analysis of data from the sixth round of (NHIS). 2019;9:e029419. doi:10.1136/ the Ghana Living Standards Survey (GLSS 6). ► We developed concentration curves and multivariate bmjopen-2019-029419 setting Household based. logistic regression models to produce new findings Participants A total of 16 774 household heads ► Prepublication history and to inform decision-making.participated in the GLSS 6 which was conducted between additional material for this ► Unlike previous studies, this study found that enrol- 18 October 2012 and 17 October 2013. paper are available online. To ment in the NHIS is slightly concentrated among the view these files, please visit Analysis Equity in enrolment was assessed using poor; however, the odds of enrolling increases with the journal online (http:// dx. doi. concentration curves and bivariate and multivariate wealth quintile, level of education and age. org/1 0. 1136/b mjopen-2 019- analyses to determine associated factors. ► As a secondary analysis, the data used for the study 029419). Main outcome measure Equity in NHIS enrolment. lack a number of important factors including trust results Survey participants had a mean age of 46 years Received 26 January 2019 in scheme management, perceived quality of care, and mean household size of four persons. About 71% of Revised 3 June 2019 ease of enrolment, etc., which would be useful for households interviewed had at least one person enrolled in Accepted 14 June 2019 better understanding NHIS enrolment.the NHIS. Households in the poorest wealth quintile (73%) had enrolled significantly (p<0.001) more than those in the richest quintile (67%). The concentration curves further poor.1–3 Prepayment schemes such as social showed that enrolment was slightly disproportionally health insurance, if implemented effectively, concentrated among poor households, particularly those headed by males. However, multivariate logistic can reduce out-of-pocket payments (OOP) analyses showed that the likelihood of NHIS enrolment and associated catastrophic effects on house- increased from poorer to richest quintile, low to high holds. 4 The quest to ensure equity in access to level of education and young adults to older adults. Other healthcare services and to achieve Universal factors including sex, household size, household setting Health Coverage (UHC) has become more © Author(s) (or their employer(s)) 2019. Re-use and geographic region were significantly associated with imperative, following adoption of the Sustain- permitted under CC BY-NC. No enrolment. able Development Goals (SDGs) by member commercial re-use. See rights Conclusions From 2012 to 2013, enrolment in the countries of the United Nations. Equity in and permissions. Published by NHIS was higher among poor households, particularly prepayment schemes is also recognised by BMJ. male-headed households, although multivariate analyses WHO as one of the fundamental elements of 1Health Policy, Planning and demonstrated that the likelihood of NHIS enrolment UHC.5 Management, University of increased from poorer to richest quintile and from low Ghana School of Public Health, to high level of education. Policy-makers need to ensure Ghana had a free healthcare system after Accra, Ghana equity within and across gender as they strive to achieve independence in the 1950s, financed by 2School of Social Policy & UHC. general taxation. However, this system of Practice and Perelman School healthcare changed when the economy of Medicine, University of started declining and user fees were partially Pennsylvania, Philadelphia, USA 3Health Policy, Planning and introduced in the 1970s and 1980s to offset IntrODuCtIOn costs of healthcare services delivery.6–8Management, University of Ghana School of Public Health, Many low-income and middle-income coun- Although the OOP somewhat helped public Accra, Ghana tries are increasingly implementing prepay- healthcare services providers to recover Correspondence to ment schemes to provide financial risk partial costs of essential medicines and Dr Eric Nsiah-Boateng; protection and equitable access to healthcare other pharmaceutical products and to raise e nsiah-b oateng@s t.u g.e du.g h services for their populations, particularly the revenue, the system-created inequity in access Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 1 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access to healthcare and in some cases led to avoidable deaths.6 8 9 Evidence shows that the NHIS has made progress in This situation resulted in the introduction of a National population coverage and contributed to utilisation of Health Insurance Scheme (NHIS) in 2003 to replace healthcare services and to expansion of healthcare facil- OOP and ensure equity in healthcare access.10 The NHIS ities in its short period of existence.17 A report of the is managed by the National Health Insurance Authority, a NHIS shows that the scheme has covered 36% (10.8 body mandated by law to regulate both public and private million) of the population as of December 2018.18 It has health insurance schemes in the country.11 166 district offices and a network of over 4000 healthcare Membership in the NHIS is broadly categorised into providers comprising both public and private healthcare exempt and non-exempt groups.11 The exempt groups facilities across the country. The benefits package report- are members who are exempted from paying premiums edly covers 95% of the disease conditions afflicting the to the scheme and they include persons below 18 years population. It broadly covers outpatient services, inpa- of age, persons aged 70 years and above, pregnant tient services, oral health, eye care services, maternity women, indigent (extreme poor), formal sector workers care and emergencies.19 Preventive services, for example, who contribute to Social Security and National Insur- immunisation and service that have the potential to pose ance Trust (SSNIT) and beneficiaries of the Livelihood sustainability challenges are excluded from the benefit Empowerment Against Poverty (LEAP) programme. package.9 11 The non-exempt group includes members who pay There are few equity-oriented studies of the NHIS in premiums and enrolment processing fees to the scheme Ghana. A mixed-method study that evaluated equity in and these are workers in the informal sector of the NHIS enrolment in two regions (Central and Eastern) economy. The NHIS is tax-funded through the National found that more males had registered in the scheme Health Insurance Fund which is based on 2.5% levy on than females and households in the richest quintile selected goods and services. Other sources of funding were significantly more likely to enrol than those in are a 2.5% deduction from formal sector workers’ SSNIT the poorest quintile.1 The study also found that old contributions, premiums from informal sector workers, age, higher education, female-headed households and funds allocated by parliament, interest from invest- perceived NHIS benefits were significantly associated ments and donor funds and gifts.11 The premium and with NHIS enrolment. Another mixed-method study enrolment processing fee from the non-exempt group examining why the NHIS is not reaching the poor used is GHS30.00 (US$6.33) per year. However, the exempt the same two regions and found fewer of the poor to be group only pays a processing fee of GHS8.00 (US$1.69) covered due to poverty and policy-makers’ and imple- for new enrolment and GHS5.00 (US$1.05) for renewal menters’ lack of commitment to pursue NHIS’s equity of membership per year. Relative to the per capital goal.20 Kusi et al,21 in examining affordability of NHIS income of GHS8863 (US$2035),12 the NHIS premium contribution, used three districts from the southern, and processing fee represent 0.34%. Again, reference middle and northern ecological zones of Ghana and to the daily minimum wage of GHS10.65 (US$2.25)13 or also found that significantly more of the rich were GHS2769.00 (US$584.18) per year, the NHIS premium enrolled in the NHIS than the poor. These three studies and processing fee constitute 0.38%. were conducted in 2008 and 2011 and employed bivar- Like many health systems around the globe, Ghana’s iate and logistic regression analyses to examine enrol- health system is hierarchical with the Ministry of Health ment equity. Other studies that also examined equity (MoH) as apex body mandated to formulate policies to in NHIS enrolment, using data from the 2008 Ghana improve health of the population.14 The MoH has about Demographic Health Survey, employed concentration 12 agencies, comprising the public, quasi-government curves and logistic regression and found that coverage and private health facilities, as well health education insti- was highest among the educated, households in the tutions. The biggest agency is the Ghana Health Service, richest quintile, and urban residents.22 23 which is charged with the responsibility of delivering This study examines equity in enrolment in Ghana’s healthcare to the population, as well as implementing NHIS to inform policy decisions regarding attainment policies of the MoH. The Ghana Health Service has a of UHC. It is necessary now to study equity to assess decentralised system of healthcare delivery with a consid- major NHIS policy reforms instituted in recent years to erable number of healthcare facilities located across the make the scheme more attractive to the general public. country. The lowest level of the healthcare delivery system One such policy is the intersectoral collaboration with is the community-based and health planning services state-owned social protection institutions, for example, compound and the highest being the tertiary or teaching Ministry of Gender and Social Protection, Ministry of hospitals at the national level. The number of health- Education, LEAP Secretariat and Savannah Accelerated care facilities and professionals are unevenly distributed Development Authority, to increase the population of the across the country, with the majority located in the urban poor and vulnerable in the NHIS and to improve equity. areas.15 16 On the other hand, many of the private health- Findings from this study can inform policy-making on care facilities particularly the faith-based ones are located UHC attainment and contribute to the body of knowl- in remote areas, where they provide about 40% of health- edge on equity in NHIS enrolment and progress towards care services to the population.14 achieving the SDGs. 2 Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access MethODs of household head, household head employment status, study design and setting household setting and geographic region of residence. This study analyses secondary data from the sixth round of Age of household head was categorised based on the the Ghana Living Standards Survey conducted between 18 October 2012 and 17 October 2013. The survey covered Table 1 Individual and household characteristics a representative sample of 18 000 households in 1200 enumeration areas across the 10 administrative regions of Variable % (n=16 772) the country.24 Survey participants had an average age of NHIS status 44 years and 48 years for males and females, respectively. C overed 70.5 In the 2010 Population and Housing Census, Ghana had N ot covered 29.5 a population of 24 658 823, with 51.2% being females. The majority of the population resided in the Ashanti Highest Education (19.4%) and Greater Accra (16.3%) regions, the two most None 50.7 urbanised regions25 of the country. These two regions Primary 30.5 also have the lowest poverty rates, while those in the Secondary 8.5 northern savannah ecological zones (Northern, Upper T ertiary 10.3 East, Upper West, Brong-Ahafo, Volta) have the highest poverty rates.26 Online supplementary appendices 1 and Employment status 2 provide more details on the population distribution E mployed 89.5 and poverty profile of Ghana. U nemployed 10.5 Wealth quintile Data collection and analysis P oorest 20.1 Data were sourced from the Ghana Statistical Service (GSS) and had already been cleaned and managed Poorer 17.6 including creation of sampling weights and wealth quin- M iddle 17.9 tiles. The GSS constructed the wealth quintiles using R icher 20.1 household expenditure as a proxy.24 The household R ichest 24.3 expenditure is composed of food and non-food items. Sex of household head The total number of households covered in the survey was divided into five groups by their total household F emale 71.8 consumption expenditure. The quintile ranking was then M ale 28.2 constructed using the household members total expen- Age of household head diture per capital. Bivariate analyses examined unad- 1 9–24 4.9 justed relationships between socio-demographic factors 25–44 47.1 and wealth quintiles. Equity in enrolment was assessed using concentration curves and indices, and multivariate 45–64 33.3 logistic regression models to determine factors associated 65–79 11.7 with enrolment.1 22 27 28 While the concertation curve anal- 8 0+ 3 yses equity in NHIS enrolment between the poor and the Household size, M (SD) 4.3 (2.78) rich, the logistic regression model shows factors associ- Household setting ated with enrolment in the scheme. The use of these two analytical techniques is therefore meant to produce reli- Rural 44.4 able findings for informed policy decision-making. Urban 55.6 The unit of analysis was the household, and we exam- Geographic region ined cumulative proportion of enrolment by wealth quin- Western 10.2 tiles, decomposed by sex, within and across male-headed C entral 9.6 and female-headed households. A multivariate logistic regression model was employed to assess whether lower G reater Accra 11.5 wealth groups were more likely to enrol in the NHIS Volta 9.4 than higher wealth groups, holding the other socio-de- Eastern 10.8 mographic variables constant. The outcome or depen- A shanti 11.8 dent variable ‘NHIS enrolment status’ was labelled 1 Brong Ahafo 9.7 for active card-bearing members and 0 for inactive card- bearing members or those who had never enrolled in Northern 10.2 the scheme. The main independent variable was ‘wealth Upper East 8.6 quintile’ and the others (control variables) were socio-de- Upper West 8.3 mographic characteristics such as age of household head, M, mean; NHIS, National Health Insurance Scheme. sex of household head, household size, education level Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 3 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access Figure 1 Concentration curves for enrolment in National Health Insurance Scheme (NHIS). Medical Subject Headings age definition.29 30 Microsoft equity in enrolment Excel 2016 and STATA V.13 were used for all analyses. Results of the concentration curve analyses demonstrate that enrolment was slightly more concentrated among Patient and public involvement poor households (figure 1). Enrolment by sex also Patients were not involved in this study. showed that enrolment was more concentrated among households headed by males compared with those results headed by females. The concentration indices further Characteristics of study participants revealed that among the study participants, equity was A total of 16 772 household heads with an average age more pronounced in the insured than the uninsured of 46 years (SD=15.58) and household size of 4 persons and within male-headed households than female-headed (SD=2.78) responded to questions on NHIS in the survey households (table 2). (table 1). Majority of the household heads (47%) were in the age bracket of 25–44 years. Out of the total number relationship between household characteristics and wealth of survey participants, 72% were females; 51% had no quintiles formal education; 90% were employed; 24% were in There were significant differences in all household the richest quintile; 56% lived in urban areas; and 12% characteristics by wealth quintiles, except employment resided in the Ashanti region. About 71% of households status (table 3). The poorest households (73%) enrolled had at least one person enrolled in the NHIS. in the NHIS more than the richest households (67%). Table 2 Concentration index (CI) showing inequity in National Health Insurance Scheme (NHIS) enrolment Total Within households (HH) Between HH Wealth Not Female-headed HH Male-headed HH quintile Enrolled enrolled Enrolled Not enrolled Enrolled Not enrolled Female Male Poorest −0.0009 0.0021 −0.0023 0.0060 −0.0009 0.0020 0.0073 −0.0029 Poorer −0.0014 0.0035 0.0061 −0.0153 −0.0010 0.0026 0.0096 −0.0039 Middle 0.0018 −0.0039 −0.0085 0.0234 −0.0002 0.0011 0.0268 −0.0108 Richer −0.0116 0.0290 0.0000 0.0000 −0.0135 0.0321 0.0455 −0.0185 Richest 0.0000 0.0000 −0.0056 0.0167 0.0000 0.0000 0.0000 0.0000 Total −0.0120 0.0307 −0.0103 0.0307 −0.0156 0.0378 0.0891 −0.0362 4 Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access Table 3 Differences in household characteristics by wealth quintile (n=16 772) Variable Q1 (poorest) Q2 (poorer) Q3 (middle) Q4 (richer) Q5 (richest) Total Pearson’s χ2 NHIS status 0.000 Enrolled 72.6 70.9 70.3 72.4 67.0 70.5 Not enrolled 27.4 29.1 29.7 27.6 33.0 29.5 Highest education 0.000 None 80.1 62.7 51.9 39.8 25.8 50.7 P rimary 16.3 28.7 34.8 38.7 33.6 30.5 S econdary 2.1 5.4 7.0 10.1 15.8 8.5 Tertiary 1.5 3.2 6.3 11.4 24.8 10.3 Employment status 0.065 Employed 90.9 89.6 90.5 88.7 88.8 89.5 Unemployed 9.1 10.4 9.5 11.3 11.2 10.5 Sex 0.000 Female 79.1 73.1 71.7 69.4 66.9 71.8 Male 20.9 26.9 28.3 30.6 33.1 28.2 Age of household head 0.000 19–24 2.4 4.1 4.4 5.9 7 4.9 25–44 40.9 43.8 46.6 48.4 54.1 47.1 4 5–64 37.5 35 34.7 32 28.7 33.3 65–79 14.9 14.2 11.4 10.9 7.9 11.7 8 0+ 4.3 2.9 2.9 2.8 2.3 3 Household size 20.1 17.7 17.9 20.1 24.3 100 0.000 Household setting 0.000 R ural 13.7 31.1 43.6 56.0 70.4 44.4 U rban 86.3 68.9 56.4 44.0 29.6 55.6 Geographic region 0.000 W estern 5.6 9.4 11.1 12.8 11.9 10.2 Central 5.1 10.6 12.2 10.8 9.5 9.6 Greater Accra 2.3 4.4 8.0 14.0 24.7 11.5 Volta 8.7 11.0 9.8 10.1 7.9 9.4 E astern 7.0 11.8 13.8 13.1 8.9 10.8 Ashanti 4.0 9.2 11.9 14.6 17.7 11.8 Brong-Ahafo 8.4 11.9 11.0 9.6 8.2 9.7 Northern 20.0 12.9 9.6 6.4 3.6 10.2 Upper East 14.7 11.4 8.3 6.2 3.8 8.6 Upper West 24.2 7.4 4.2 2.5 3.8 8.3 NHIS, National Health Insurance Scheme. Interestingly, the richer households had the second than those in the richest quintile, and more households highest enrolment (72.4%) in the scheme. Majority of in the poorest quintile (86%) living in urban settings the poorest households (80.1%) had no formal educa- than households in the richest quintiles (30%). tion compared with about 25% of the richest house- Results of the multivariate logistic regression showed holds with tertiary level education. Similarly, majority of that the likelihood of enrolling in the NHIS increases the poorest households (91%) were more employed as from poorer to richest quintile, low to high level of educa- were the richest households (89%), and there were more tion and young adults to older adults (table 4). Females females (79%) in the poorest quintile than in the richest (OR: 1.52; 95% CI: 1.39–1.65) and persons living in the quintile (67%). There were also significantly more house- Upper East (OR: 5.99; 95% CI: 4.91–7.31), Upper West hold heads aged 45 years or more in the poorest quintile (OR: 5.04; 95% CI: 4.14–6.15), Brong-Ahafo (OR: 3.06; Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 5 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access Table 4 Multivariate logistic regression model of enrolling in the National Health Insurance Scheme Variable Unadjusted OR 95% CI Adjusted OR 95% CI Wealth quintile P oorest 1.00 1.00 Poorer 0.92 0.82 to 1.02 1.33*** 1.17 to 1.50 Middle 0.89* 0.79 to 0.99 1.54*** 1.36 to 1.75 Richer 0.98 0.88 to 1.09 1.94*** 1.70 to 2.22 Richest 0.76*** 0.69 to 0.84 1.67*** 1.45 to 1.91 Highest education None 1.00 1.00 Primary 1.05 0.98 to 1.14 1.65*** 1.51 to 1.80 S econdary 1.27*** 1.12 to 1.44 2.35*** 2.03 to 2.72 T ertiary 1.75*** 1.55 to 1.99 2.87*** 2.48 to 3.32 Employment status Unemployed 1.00 1.00 Employed 0.85** 0.76 to 0.95 0.99 0.87 to 1.12 Sex of household head M ale 1.00 1.00 Female 1.11** 1.03 to 1.19 1.52*** 1.39 to 1.65 Age of household head 1 9–24 1.00 1.00 25–44 1.99*** 1.72 to 2.31 1.53*** 1.31 to 1.79 45–64 2.38*** 2.05 to 2.77 1.69*** 1.43 to 1.99 65–79 3.43*** 2.87 to 4.08 3.05*** 2.51 to 3.69 80+ 3.18*** 2.47 to 4.08 3.28*** 2.49 to 4.34 Household size 1.17*** 1.15 to 1.18 1.23*** 1.20 to 1.25 Household setting Urban 1.00 1.00 Rural 0.97 0.91 to 1.04 0.75*** 0.69 to 0.82 Geographic region Western 1.00 1.00 C entral 0.64*** 0.55 to 0.73 0.63*** 0.54 to 0.73 Greater Accra 0.79** 0.69 to 0.90 0.63*** 0.52 to 0.69 Volta 1.89*** 1.62 to 2.21 2.04*** 1.73 to 2.39 E astern 1.34*** 1.16 to 1.53 1.39*** 1.20 to 1.62 Ashanti 1.16* 1.01 to 1.32 1.08 0.94 to 1.25 Brong-Ahafo 2.68*** 2.27 to 3.15 3.06*** 2.58 to 3.62 N orthern 1.07 0.92 to 1.22 1.32*** 1.13 to 1.54 Upper East 4.30*** 3.56 to 5.19 5.99*** 4.91 to 7.31 U pper West 3.62*** 3.01 to 4.33 5.04*** 4.14 to 6.15 _cons 0.23*** 0.18 to 0.29 Number of obs. 16 693 LR chi2(24) 2236.6 Prob>chi2 0.0000 Pseudo R2 0.1106 Control variables: education of household head, employment status of household head, sex of household, age of household head, household size, household setting and geographic region. ***P<0.001; **P<0.01; *P<0.05. 95% CI: 2.58–3.62), Volta (OR: 2.04; 95% CI:1.74–2.39) employed were less likely to enrol in the NHIS (OR=0.99; and Northern (OR: 1.32; 95% CI: 1.13–1.54) regions 95% CI 0.87–1.12) although not significantly so. The were significantly more likely to enrol in the NHIS than unadjusted odds ratios (OR) showed similar associations their respective reference categories. Surprisingly, the except for wealth quintile, the explanatory variable of 6 Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access interest, which showed a decreased likelihood of enrolling study by Jehu-Appiah et al,1 One reason may be due to in the NHIS from poorer to richest. poverty; prior studies showed that the majority of rural dwellers are unable to afford the NHIS premium and processing or renewal fee.20 31 34 36–38 This study’s findings DIsCussIOn also show that the odds of enrolling in the NHIS increases This study examined equity in NHIS enrolment with household size, consistent with other studies,22 33 34 employing data from the Ghana Living Standards Survey because larger households may be risk averse and thus (round 6) which was conducted between October 2012 would enrol in the NHIS to seek financial risk protection and October 2013. The findings show inequity in enrol- against their healthcare costs and to avoid catastrophic ment and significant associations between socio-demo- OOP. Our findings also reveal that individuals residing graphic factors and NHIS enrolment. Among households in less developed regions of the country are significantly surveyed, enrolment is disproportionally concentrated more likely to enrol in the scheme compared with those among poor households especially those headed by in developed regions. Again, this may be attributed to males. The possible explanation relates to policy changes policy reforms focused on enrolling individuals living made over the last few years to increase enrolment in in deprived regions, particularly those in the northern the scheme. One such policy is the deliberate attempt savannah ecological zones, comprising the Northern, to increase numbers of the poor and vulnerable in the Upper East, Upper West and some parts of Brong-Ahafo scheme through enrolment of the LEAP beneficia- and Volta regions,24 consistent with some studies22 23 and ries, students in secondary and tertiary institutions in contradicting other.35 Ghana, prisoners and individuals living in less developed Our study’s primary limitation is that the data lacked geographic regions, particularly those in the northern several important factors (such as trust in scheme manage- savannah ecological zone, where there is high preva- ment, perceived quality of care, ease of enrolment, etc) lence of poverty. The disproportionate concentration of which would be useful for better understanding NHIS enrolment among poor households contradicts previous enrolment. Nonetheless, the variables used in the multi- studies on the NHIS,1 20–22 31 32 due possibly to the years in variate logistic regression modelling did not significantly which those studies were conducted (2008 and 2011), as affect model robustness. well as the limited regional scope (three administrative regions except the 2008 Demographic Health Survey that covered the entire country). This present study employs a COnClusIOn nationally representative survey. The study reveals that from 2012 to 2013, enrolment in Our study also shows that a number of socio-demo- the NHIS was higher among poor households, particu- graphic factors are significantly associated with NHIS larly male-headed households, although the multivar- enrolment. Although unadjusted findings illustrate that iate analyses demonstrated that the likelihood of NHIS enrolment is concentrated among poor households, enrolment increased from poorer to richest quintile, multivariate findings illustrate that the odds of enrolling low to high level of education and young adults to older in the scheme increases with wealth quintiles, that is, the adults. While the NHIS strives to achieve its pro-poor goal rich are more likely to enrol than the poor. This may of providing financial risk protection for the poor and be attributed to evidence that the rich are more able vulnerable in society, equity must be addressed within to afford the cost of enrolling in the health insurance and across the entire population. Adequate funds are programme than the poor.1 20 33 34 Besides, as explained also required to cover the anticipated increase in medical earlier, the policy decision to deliberately enrol the poor claims costs because as more poor and vulnerable groups might have contributed to their higher numbers in the enrol in the scheme, the claims cost is likely to escalate NHIS, but voluntarily other factors other than being poor and threaten the scheme’s sustainability. Thus, policy contribute to enrolment in the scheme. Individuals with decisions to ensure equity in enrolment must also ensure higher levels of education are more likely to enrol in the commensurate funding to avoid financial uncertainty NHIS compared with those with no formal education; and collapse. Further research on equity in healthcare females are more likely to enrol than males; and older services utilisation, expenditures and accreditation of adults are more likely to enrol than young adults, consis- healthcare providers is needed to provide a fuller picture tent with previous studies.1 22 32–35 The employed are of equity assessment in the NHIS. less likely to enrol compared with the unemployed. The plausible explanation is that the employed may be able Acknowledgements We are grateful to the Ghana Statistical Service for the to afford OOP for healthcare services because they are provision of the GLSS data for this study. We also thank all contributors and more economically resourced than the unemployed. This reviewers for their comments and time. result runs counter to earlier studies.21 35 Contributors ENB, JPR and JN conceived and designed the study. JN retrieved the data and ENB analysed the data and drafted the manuscript. JPR and JN provided Findings from this study also reveal that individuals intellectual contributions to develop and revise the manuscript. All the authors read residing in rural settings are significantly less likely to enrol and approved the manuscript for publication. in the NHIS compared with those living in urban areas, Funding The authors have not declared a specific grant for this research from any consistent with previous studies,32 35 but contradicting a funding agency in the public, commercial or not-for-profit sectors. Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 7 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright. Open access Competing interests ENB is an employee of the National Health Insurance 17. National Development Planning Commission. 2008 Citizens’ Authority, however his affiliation did not influence findings of this study. JPR and JN assessment of the national health insurance scheme: towards a declare no competing interests. sustainable health care financing arrangement that protects the poor.  Accra, 2009. Patient consent for publication Not required. 18. National Health Insurance Authority. Active membership report for December 2018 (final as at April 1 2019. Accra, 2019. ethics approval This study is a secondary analysis of the Ghana Living Standard 19. National Health Insurance Authority. Benefits Package. National Survey (round 6) data, however formal approval was obtained from the Ghana Health Insurance Scheme. 2019 http://www. nhis. gov. gh/ benefits. Statistical Service to use the data. aspx (Accessed 8 Apr 2019). Provenance and peer review Not commissioned; externally peer reviewed. 20. Kotoh AM, Van der Geest S. Why are the poor less covered in Ghana’s national health insurance? A critical analysis of policy and Data sharing statement No additional data is available. practice. Int J Equity Health 2016;15:1–11. 21. Kusi A, Enemark U, Hansen KS, et al. Refusal to enrol in Ghana’s Open access This is an open access article distributed in accordance with the National Health Insurance Scheme: is affordability the problem? Int J Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which Equity Health 2015;14:1–14. permits others to distribute, remix, adapt, build upon this work non-commercially, 22. Dake FAA. Examining equity in health insurance coverage: an and license their derivative works on different terms, provided the original work is analysis of Ghana’s National Health Insurance Scheme. Int J Equity properly cited, appropriate credit is given, any changes made indicated, and the use Health 2018;17:1–10. is non-commercial. See: http:// creativecommons.o rg/ licenses/b y- nc/4 . 0/. 23. Dixon J, Tenkorang EY, Luginaah I. Ghana's National Health Insurance Scheme: Helping the Poor or Leaving Them Behind? Environment and Planning C: Government and Policy 2011;29:1102–15. 24. Ghana Statistical Service. Ghana Living Standards Survey Round 6 reFerenCes (GLSS 6): Main Report. Accra 2014 http://www. statsghana.g ov.g h/ 1. Jehu-Appiah C, Aryeetey G, Spaan E, et al. Equity aspects of the docfiles/ glss6/ GLSS6_ Main R eport.p df (Accessed 21 Aug 2017). National Health Insurance Scheme in Ghana: Who is enrolling, who is 25. Ghana Statistical Service. Population & housing census: summary not and why? Soc Sci Med 2011;72:157–65. report of final results. Accra. 2012 http:// statsghana. gov. gh/ docfiles/ 2. Wiseman V, Thabrany H, Asante A, et al. An evaluation of health 2010phc/ Census2010_ Summary_r eport_ of_ final_ results. pdf. systems equity in Indonesia: study protocol. Int J Equity Health 26. Ghana Statistical Service. Ghana Living Standards Survey Round 6 2018;17:1–9. (GLSS 6): Poverty Profile in Ghana (2005-2013). Accra. 2014 http:// 3. Asante A, Price J, Hayen A, et al. Equity in health care financing in www. statsghana.g ov. gh/ docfiles/g lss6/ GLSS6_P overty Profile in low- and middle-income countries: a systematic review of evidence Ghana.p df (Accessed 29 Jan 2017). from studies using benefit and financing incidence analyses. PLoS 27. Allin S, Masseria C, Sorenson C, et al. Measuring inequalities in One 2016;11:e0152866. access to health care. A review of the indices. London Sch Econ 4. Ruger JP. An alternative framework for analyzing financial protection Polit Sci 2007:1–24. in health. PLoS Med 2012;9:e1001294. 28. O’Donnell O, Van Doorslaer DE, Wagstaff A, et al. Analyzing health 5. Evans DB, Etienne C. Health systems financing and the path to equity using household survey data: a guide to techniques and their universal coverage. Bull World Health Organ 2010;88:402–128. implementation. Washington DC: The World Bank, 2008. www. 6. Waddington C, Enyimayew KA. A price to pay, part 2: The impact of worldbank.o rg. (Accessed 22 Sep 2018). user charges in the Volta region of Ghana. Int J Health Plann Manage 29. Geifman N, Cohen R, Rubin E. Redefining meaningful age groups in 1990;5:287–312. the context of disease. Age 2013;35:2357–66. 7. Asenso-Okyere WK, Anum A, Osei-Akoto I, et al. Cost recovery in 30. Kastner M, Wilczynski NL, Walker-Dilks C, et al. Age-specific search Ghana: are there any changes in health care seeking behaviour? strategies for Medline. J Med Internet Res 2006;8:e25. Health Policy Plan 1998;13:181–8. 31. Kotoh AM, Aryeetey GC, Van der Geest S. Factors that influence 8. Nyonator F, Kutzin J. Health for some? The effects of user fees in the enrolment and retention in Ghana’ National health insurance scheme. Volta Region of Ghana. Health Policy Plan 1999;14:329–41. Int J Health Policy Manag 2018;7:443–54. 9. Ministry of Health. National Health Insurance Policy framework 32. Akazili J, Welaga P, Bawah A, et al. Is Ghana’s pro-poor health for Ghana: Revised Version. Accra. 2004 https://www. insurance scheme really for the poor? Evidence from Northern ghanahealthservice.o rg/ downloads/N HI_ policy framework. pdf Ghana. BMC Health Serv Res 2014;14:1–9. (Accessed: 17 Dec 2017). 33. Dror DM, Hossain SA, Majumdar A, et al. What factors affect 10. Parliament of Ghana. National Health Insurance Act, 2003 (Act 650). voluntary uptake of community-based health insurance schemes in Insurance Act, 650. Accra, Ghana, 2003:1–39. low- and middle-income countries? A systematic review and meta- 11. Parliament of Ghana. National Health Insurance Act, 2012 (Act 852. analysis. PLoS One 2016;11:e0160479. Accra, Ghana: Parliament of Ghana, 2012:1–55. 34. Adebayo EF, Uthman OA, Wiysonge CS, et al. A systematic review 12. Attah Yeboah O. Per capita GDP increases to US$2,035 after of factors that affect uptake of community-based health insurance rebasing …but has little effect on poverty. Business and Financial in low-income and middle-income countries. BMC Health Serv Res Times Online. 2018 https://t hebftonline.c om/2 018/ economy/ per- 2015;15:1–13. capita-g dp-i ncreases-t o- us2035-a fter-r ebasing- but-h as- little- effect- 35. van der Wielen N, Falkingham J, Channon AA. Determinants of on-p overty/ (Accessed 8 Apr 2019). National Health Insurance enrolment in Ghana across the life course: 13. Nyavor G. Minimum wage goes up by 10%; now GHS10.65. 2018 Are the results consistent between surveys? Int J Equity Health https://www.m yjoyonline. com/b usiness/ 2018/J uly- 27th/m inimum- 2018;17:1–14. wage-g oes- up- by- now- gh1065. php (Accessed 8 Apr 2019). 36. Fenny AP, Kusi A, Arhinful DK, et al. Factors contributing to low 14. Aseweh Abor P, Abekah‐Nkrumah G, Abor J. An examination of uptake and renewal of health insurance: a qualitative study in Ghana. hospital governance in Ghana. Leadersh Health Serv 2008;21:47–60. Glob Health Res Policy 2016;1:1–10. 15. Andoh-Adjei FX, Nsiah-Boateng E, Asante FA, et al. Perception 37. Agyepong IA, Abankwah DNY, Abroso A, et al. The “Universal” in of quality health care delivery under capitation payment: a cross- UHC and Ghana’s National Health Insurance Scheme: policy and sectional survey of health insurance subscribers and providers in implementation challenges and dilemmas of a lower middle income Ghana. BMC Fam Pract 2018;19:1–12. country. BMC Health Serv Res 2016;16:1–14. 16. Duku SKO, Nketiah-Amponsah E, Janssens W, et al. Perceptions of 38. Atinga RA, Abiiro GA, Kuganab-Lem RB. Factors influencing the healthcare quality in Ghana: Does health insurance status matter? decision to drop out of health insurance enrolment among urban PLoS One 2018;13:e0190911. slum dwellers in Ghana. Trop Med Int Health 2015;20:312–21. 8 Nsiah-Boateng E, et al. BMJ Open 2019;9:e029419. doi:10.1136/bmjopen-2019-029419 BMJ Open: first published as 10.1136/bmjopen-2019-029419 on 1 July 2019. Downloaded from http://bmjopen.bmj.com/ on September 11, 2019 by guest. Protected by copyright.