Unintended pregnancy is a widespread public health issue globally, but there is evidence that it has decreased over the years. The recent estimate of unintended pregnancy is 121 million pregnancies per year in 2015-2019, which is less than two-fifths of the previous estimate of 210 million unintended pregnancies per year (Bearak et al., Reference Bearak, Popinchalk, Ganatra, Moller, Tunçalp, Beavin, Kwok and Alkema2020). Furthermore, it is estimated about two-fifths are mistimed or unwanted at the time of conception (Bearak et al., Reference Bearak, Popinchalk, Alkema and Sedgh2018). Of those unintended pregnancies, 61% ended in abortion with remaining either miscarried or carried to term (Bearak et al., Reference Bearak, Popinchalk, Ganatra, Moller, Tunçalp, Beavin, Kwok and Alkema2020). There is no doubt that unintended pregnancy, which includes either unwanted or mistimed pregnancies, pose health risks to both mother and baby. Because unintended pregnancy is often a surprise, there is a high likelihood of prenatal care being neglected or foregone (Hajizadeh & Nghiem, Reference Hajizadeh and Nghiem2020) due to economic or psychological reasons (Heaman et al. Reference Heaman, Moffatt, Elliott, Sword, Helewa, Morris, Gregory, Tjaden and Cook2014). Studies have shown that births resulting from unintended pregnancies are associated with higher rates of adverse health, social, and economic outcomes for mother and baby (Oyediran et al., Reference Oyediran, Feyisetan and Akpan2011; Logan et al., Reference Logan, Holcombe, Manlove and Ryan2007; Gipson et al., Reference Gipson, Koenig and Hindin2008; Filonenka et al., Reference Filonenka, Law, Purser, Mader and Graham2019). Such adverse health outcomes including low birthweight, shorter duration of breastfeeding, parenting stress, and negative physical and mental impacts for children such as poor development and potential behavioural health issues (Brieger et al., Reference Brieger, Delano, Lane, Oladepo and Oyediran2001; Logan et al., Reference Logan, Holcombe, Manlove and Ryan2007; Gipson et al., Reference Gipson, Koenig and Hindin2008; Dude et al., Reference Dude, Miller, Garcia and Yee2021). Furthermore, unwanted pregnancies are more likely to end in abortion than intended pregnancies (Finer & Lindberg, Reference Finer and Lindberg2010), and given that many abortions in restrictive legal context are unsafe, this is problematic for the health and safety of women (Bankole et al., Reference Bankole, Remez, Owolabi, Philbin and Williams2020).
Several women and couples wish to plan the timing and spacing of birth or pregnancy for health, social, and economic reasons, thus, timely access to highly effective contraception would lower the rate of unintended pregnancy among those who are not trying to become pregnant. But access to family planning remains a challenge in most countries in sub-Saharan Africa, particularly in West African nations where contraceptive uptake remains low. Many households are living in abject poverty and find it difficult to access health services except through out-of-pocket health expenditures (Kagaigai et al., Reference Kagaigai, Anaeli, Mori and Grepperud2021). Thus, state-actors and stakeholders are in search of approaches to increase access to essential health services including the uptake and quality of family planning services, and the most important one has been the institutionalization of health insurance schemes.
Family planning is central to sustainable development goal 3 (SDG 3), particularly targets 3.1, reduced maternal mortality and 3.8 achieving universal health coverage. In addition, family planning will contribute to accomplishing SDG 5 on gender equality, with target 5.6 specifically targeted at ensuring universal access to sexual, reproductive health, and reproductive rights. There are signs and efforts by countries demonstrating strong commitment to achieving universal health coverage, and health insurance is being advancing as a strategy of meeting the goal of universal health coverage (Naik, Morgan & Wright, Reference Naik, Morgan and Wright2014; Perehudoff, Alexandrov & Hogerzeil, Reference Perehudoff, Alexandrov and Hogerzeil2019; Warner et al., Reference Warner, Benjamin, Churchwell, Firestone, Gardner, Harrington, Johnson, Ng-Osorio, Rodriguez, Todman, Yaffe and Yancy2020). One of the ways that these countries work to achieve their goal of universal health coverage is through the implementation of the health insurance scheme. Health insurance can empower women to access health services, including contraception by defraying their financial constraint which is often a major barrier cited for non-use of family planning (Guttmacher Institute & International Planned Parenthood Federation, 2010; Fox & Barfield, Reference Fox and Barfield2016).
Ghana’s National Health Insurance Scheme (NIHS) was established in 2003 and operationalized in 2005 with the goal of providing a broad range of healthcare services while working to increase affordability and utilization among the more vulnerable populations across the country. The NIHS benefit package which covers “95% of disease conditions that afflict Ghanaians.” This package includes a variety of outpatient services including diagnostic testing and basic procedures as well as most in-patient services including most surgeries, hospital accommodation, and specialist care. It also includes a provision for oral health, all maternity care related services, and all drugs on the official medicines list but no specificity of sexual and reproductive health services (Blanchet et al., Reference Blanchet, Fink and Osei-Akoto2012).
The impact of the NIHS has been documented in multiple studies. For instance, Mensah and colleagues found that pregnant women covered under the NHIS were more likely to receive prenatal care and give birth at a health facility (Mensah et al., Reference Mensah, Oppong and Schmidt2010). Evidence regarding the impact of health insurance on sexual and reproductive health outcome is important in the context of debate around universal health coverage from both health and public policy perspectives. Prior research suggests that health insurance is associated with an increased likelihood of health service coverage including contraception (Kavanaugh et al., Reference Kavanaugh, Douglas-Hall and Finn2019). However, scant research has examined whether health insurance coverage is a predictor of unintended pregnancy in the West Africa sub-region, particularly in Ghana where health insurance coverage is high among the population. Thus, with increased attention on universal health coverage and health insurance schemes, it is important to examine potential relationship between health insurance coverage and various health outcomes, such as the links between health insurance coverage and unintended pregnancy among women of reproductive age in Ghana.
Materials and methods
Data
This is a further analysis of the 2014 Ghana Demographic and Health Survey (GDHS) (2014 GDHS) (Ghana Statistical Service, Ghana Health Service & ICF International, 2015); a national survey conducted every five years. The survey was funded by numerous donors including, United States Agency for International Development (USAID), United Nations Development Program (UNDP), United Nations Fund Population Activities (UNFPA), United Nations Children Fund (UNICEF), among others with technical oversight from ICF. The GDHS covers issues such as unintended pregnancy, family planning, nutrition, fertility, health insurance, prenatal care, choice of place of delivery and postnatal care, women empowerment, infectious diseases, and domestic violence (Ghana Statistical Service, Ghana Health Service & ICF International, 2015). The 2014 GDHS interviewed 9,396 women aged 15–49 years in 11,835 households selected from 427 clusters using a two-stage sample design (Ghana Statistical Service, Ghana Health Service & ICF International, 2015). The goal was to derive estimates for the key indicators at the national and ten regions as well as for the urban and rural areas. The survey used the updated frame from the 2010 Population and Housing Census (PHC) and had a response rate of 97%. The 2014 GDHS dataset was accessed from the DHS website following approval from the program (Ghana Statistical Service, Ghana Health Service & ICF International, 2015).
Statistical analysis
Descriptive statistics
The descriptive analysis was conducted to show the distribution of women by the key variables. The descriptive statistics include percentages and means for categorical and continuous variables, respectively. The analytical techniques involve logistic regression and propensity scoring matching (PSM). PSM was employed to examine the effect of health insurance coverage on women’s experiencing an unintended pregnancy (Zhehui, Gardiner & Bradley, Reference Zhehui Luo and Bradley2010). Weight was applied to adjust for differences in the probability of selection and to adjust for non-response to produce the proper representation. Individual weights were used for descriptive statistics.
Propensity score matching
A propensity score is the probability of being exposed to a treatment or an intervention given a set of observed covariates. This method was developed as a means to balance the treatment and control units so that a direct comparison would make a valid conclusion. PSM therefore makes it feasible to create an unexposed comparison group that is statistically equivalent on the average to subjects who are exposed to the treatment (Rosenbaum & Rubin, Reference Rosenbaum and Rubin1983; Heckman, Ichimura & Todd, Reference Heckman, Ichimura and Todd1998). For research survey, a single score for matching is generated using statistically regressing exposure on all of the variables that determine exposure and also may be related to the outcome variable (Kincaid & Do, Reference Kincaid and Do2006). This technique requires a two-stage process. Stage 1 involves the use of a logistic regression model to calculate all respondents’ propensity for experiencing a treatment of interest, in this case, enrolled in health insurance coverage. The propensity score is defined as follows (Rosenbaum & Rubin, Reference Rosenbaum and Rubin1983):
$$\rm p(T) \equiv Pr2 = E\{T \,|\, S\}$$
Where p(T) is the propensity to be enrolled into health insurance, T indicates that a woman was uninsured, and S is a vector of covariates influencing whether the woman did or did not have health insurance coverage. In Stage 2, propensity scores estimated in stage 1 was used to match women who are insured and uninsured. The analysis applies the stratification matching which uses all treatment and control cases to obtain the full sample. Using the STATA 16.0 “psmatch2” command, the full range of sample members’ propensity scores is divided into propensity score strata, or blocks, each of which includes treatment and control cases with the same or nearly the same propensities for receiving the treatment. The number of appropriate strata depends on the number necessary to gain a balanced propensity score. Within each of these strata, the ATT is calculated, and then the ATT’s across strata are averaged to produce a final ATT.
The standard error for each of the five strata identified follows Shoemaker’s (Reference Shoemaker1972) method of calculating unbiased parameters of sub-classification of propensity scores was used to obtain the standard errors for the adjusted probabilities in order to completely eliminate any initial bias that may have been caused by the covariates used in the matching.
The propensity score analysis was used to stimulates randomisation by matching the women by outcome status (unintended pregnancy vs. intended pregnancy) for the predictors that reduce the likelihood of bias in the treatment effects (Heckman et al., Reference Heckman, Ichimura, Smith and Todd1996). The approach tries to reduce the bias associated with confounding factors that may exist in an estimate of the treatment effect obtained from simply comparing outcomes among subjects that were exposed to the treatment and those that were not. The differences in the characteristics of women that experienced unintended pregnancy with and without health insurance coverage were accounted for by using the propensity score technique. The sample included woman between the ages of 15-49 who provided a birth history within the five-year period preceding the survey, and the analysis was restricted to 4,544 women who were pregnant or had a birth. The women were further ask about the planning status of their last child in the 2 years preceding the survey or current pregnancy. The breakdown showed that 3,097 of the women either desired their last child or pregnancy when it occurred while the remaining (1,447) women reported the birth or pregnancy as on unintended.
Logistic regression
Logistic regression was used as a propensity score estimation algorithm and nearest neighbour matching as matching algorithm. Each woman was matched with health insurance coverage with the closest propensity score on a ratio of 1.5 using the nearest neighbour algorithm with no replacement. Women’s sociodemographic characteristics were presented as percentages in the first step of the analysis. The health insurance coverage according to the women’s socio-economic and demographic characteristics. In addition, chi-square (χ2) test was performed to check for significant associations between the explanatory variables and unintended pregnancy. Variables that were found to be significantly associated in the χ2 tests (at p < 0.25) were selected for final regression analysis. In the final step, binary logistic regression model was used to calculate the odds ratios (OR) of the associations between unintended pregnancy and health insurance coverage. STATA 16.0 package was used for the complete analysis.
The logistic regression was used to determine the explanatory variables with a significant effect on the outcome variable. Below is the regression model equation.
Where, where p is the probability that the event Y occurs, coefficient ‘b’ is the factor by which the odds change with unit increase in independent variable. If “b” is positive, odds ratio will increase, as this factor will be greater than 1. Contrary to that if ‘b’ is negative, odd ratio will decrease. When ‘b’ is 0, the factor exponential of ‘b’ equal to 1 and, therefore the odds remain unchanged.
$$Logit\,p = \log {p \over {1 - p}} = \log \Omega = {b_0} + {b_1}{x_1} + {b_2}{x_2} + \cdots \ldots + {b_k}{x_k}$$
Dependent and independent variables
Dependent variable
The outcome variable was unintended pregnancy that captured the pregnancy intention at the conception. During the survey, women reported retrospectively if their last pregnancy (the index child) was wanted then (planned), wanted later (mistimed) or not wanted at all (unwanted). These three responses were reclassified into a dichotomous variable in two ways; that is using unintended pregnancy vs. intended pregnancy and using unwanted pregnancy vs. wanted pregnancy categories. In the first scenario, women who reported that their most recent pregnancy (in the last two years) was wanted later or not wanted at all were categorised as unintended pregnancies and intended if they reported otherwise. In the second instance, women who reported that their most recent pregnancy as wanted then or wanted later were categorised as wanted pregnancies and unwanted if they reported the pregnancy was not wanted at all.
Independent variables
The main independent variable included in the analyses was health insurance coverage. The GDHS asked respondents whether they were covered at the time of interview and, if covered, the type of health insurance by which they are covered. A dichotomous measure of health insurance coverage was constructed with women who reported to be registered with the National Health Insurance Scheme (NHIS) was equal to one and zero otherwise. Other covariates of unintended pregnancy included in the analysis were informed by the previous studies (Ahinkorah, Reference Ahinkorah2020; Nyarko, Reference Nyarko2019; Ameyaw et al., Reference Ameyaw, Budu, Sambah, Baatiema, Appiah, Seidu and Ahinkorah2019; Bain, Zweekhorst & de Cock Buning, Reference Bain, Zweekhorst and de Cock Buning2020) and availability in the GDHS datasets. Other covariates included were women’s age, education, religious affiliation, type of marriage, place of residence, household wealth status, number of living children, and administrative region of residence. In addition, women’s position and status related factors such as working status, mass media exposure to family planning message and decision-maker for contraceptive use and health-related issues.
Women’s education was categorized into “no education,” “primary education,” and “secondary and above.” The employment by earning status was categorized as “not working”, “working receiving cash”, working and receiving cash/in-kind”, and “working but not compensated in cash or in-kind.”
To explore the relationship between mass media exposure and unintended pregnancy, a composite score of the media exposure was developed from women’s responses to their exposure to the family planning messages through listening to the radio, watching television, and reading newspapers: if a woman read newspapers more than once a week, she scored “1” and “0” otherwise. A woman who had listened to the radio more than once a week was scored “1” and “0” otherwise. A woman who watched television more than once radio, a week was scored “1” and “0” otherwise. Thereafter, the score for each woman was aggregated and ranged between 0 and 3, with a score of 0 meaning that the woman had no exposure to any mass media and a score of 3 meaning that she had access to all three. Scores were categorized into 0 “no media exposure,” 1 “exposed to one source of media,” 2 “exposed to two sources of media,” and 3 “exposed to three sources of media.”
Geographical variables included are: place of residence (urban vs. rural) and administrative region. Ghana consists of twelve distinct administrative regions. Household wealth is an ordinal measure describing the poorest, poorer, middle, richer, and the richest households. Some fertility-related variables including women’s age (15–24 years, 25–34 years, >=35 years) and type of marital union (never married, monogamous, polygamous, and previously married) are included in the analysis. The number of living children (parity) was grouped into four categories: 0, 1–2, 3–4, and 5 or more. It is expected that women in polygamous relationship and those who have more living children are more likely to have an unintended pregnancy than their monogamous counterparts and those with fewer number of living children, respectively.
Women’s status as a variable was constructed from respondent’s decision-making capacity within the household from four items (person who usually decides on the respondent’s health care, person who usually decides on large household purchases, person who usually decides on visits to family or relatives, and person who usually decides what to do with husband’s earning). For each item, each woman reported whether the decision was made by the respondent alone, respondent with husband/partner, husband/partner alone, someone else or other. A dichotomous variable was constructed from each item by re-coding the responses as 1 for those who reported “respondent alone” or “respondent and husband/partner” and 0 for those who reported “husband/partner alone” or someone else or other.” An index with a value of 0-4 was created from these dichotomous variables with 0 indicating “no decision-making power” in household decision-making and 1-4 indicating participation in the household decision-making process.
Results
Sample profile
The socio-demographic characteristics of the studied women were depicted in Table 1. Sixty-six percent of the 9,396 women interviewed during the 2014 GDHS reported health insurance coverage. Nearly half (48.4%) of the 9,396 women interviewed reported ever being pregnant, 31.8% of whom reported that at least one of their pregnancies was unintended. Among those women who reported an unintended pregnancy, 65.0% had health insurance covered as compared to the 72.1% of women who did not have an unintended pregnancy who had health insurance.
Table 1. Socioeconomic characteristics of the women and background attribute as predictor of health insurance coverage, 2014 Ghana demographic and Health Survey
*p < 0.05; **p < 0.01; ***p < 0.001.
The first panel of Table 1 shows the distribution of the women by selected socio-demographic and behavioural attributes. The mean ages of all women in the sample were 29.7 years and 29.9 years for those who had health insurance coverage. The women are almost equally distributed by the age group with nearly two-thirds of women were under 35 years of age. Over half (57.2%) of women reported having secondary or higher education. Regarding economic status, 72.0% of the women reported they were currently working, and about 43.6% reported to be living in the poorest households (first and second wealth quintiles). The results showed that a fairly high percentage (35.5 %) of women reported living in female headed households, and 46.6% of the women were in monogamous unions. An overwhelming majority (76.3%) of the women were Christians. Nearly equal percentages of women lived in rural and urban areas (51.0% vs. 49.0%). A significant number of women (54.2%) reported to be participating in household decision-making, and 63.7% were married at 18 years or older.
Panel 2 of Table 1 showed that the prevalence of health insurance coverage varied significantly across women’s population sub-groups. The prevalence of health insurance coverage had an inverted U-shaped association with age. Women aged 15-24 years were significantly less likely to have health insurance coverage than those aged 25 years and above. The findings further revealed that women who participated in household decision-making, women who are Muslims, and women who were exposed to family planning messages through three mediums of mass media were more likely than women who have not participated in household decision-making, women who are Christians, and women who were not exposed to family planning message to report health insurance coverage. See Table 1 for more details on the relationship between women’s characteristics and health insurance enrolment.
Table 2 details the logistic regression analysis of the correlation of women’s health insurance coverage in Ghana. The findings showed that women’s age, education, religious affiliation, region of residence, household wealth index, type of marital union, and exposure to mass media were significant predictors of the health insurance enrolments among women in Ghana. For instance, women with post-secondary education are more likely to be enrolled in health insurance (OR=2.68, p <.0001) compared with those who are not educated. Also, women in the household on the fifth quintile of the wealth index are more likely to have health insurance coverage (OR=1.82, p=.001) compared with those in the first wealth quintile. Religious affiliation of the women also shown to be an important predictor, with Muslim women significantly more likely than their Christian counterparts to enrol for health insurance. For the type of marital union, polygamous married women (OR=0.81, p=.033) and previously married (OR=0.49, p, .0001) were statistically significantly more likely than those who are monogamously married to have health insurance coverage.
Table 2. Logistic Regression Analysis of the Factors Influencing Health Insurance Coverage among women in Ghana
Differentials in unintended pregnancy and health insurance coverage status
Table 3 report the percentage of women that experienced unintended pregnancy according to the health insurance coverage status at the time of the interview as defined by their socio-demographic and behavioural characteristics. The results showed that about 30% of women who had health insurance coverage experienced unintended pregnancy compared with 37% among women who had no health insurance coverage, meaning the proportion of women reporting unintended pregnancy was lower among insured women than uninsured women. The analysis revealed that women that reported unintended pregnancy according to the health insurance coverage status varied significantly across the population sub-groups, as defined by their socio-demographic and behavioural attributes (See Table 3). Irrespective of the health insurance coverage, unintended pregnancy appears to differ by age. For instance, among those with health insurance coverage, women aged 15-24 years were significantly more likely to experience unintended pregnancy than those aged 25 years and above. The same pattern was observed among women without health insurance coverage. Results further revealed that unintended pregnancy was associated with use of modern family planning method among women without health insurance coverage, while unintended pregnancy did not appear to differ by the use of the modern family methods among women with health insurance coverage. Other characteristics associated with unintended pregnancy irrespective of the health insurance coverage are: education, religious affiliation, ethnicity, region of residence, household wealth quintile, and women status related factors such as participation in household decision-making, working etc.
Table 3. Prevalence of unintended pregnancy among women of reproductive age according to the socioeconomic characteristics by health insurance coverage, 2014 GDHS
*p < 0.05; **p < 0.01; ***p < 0.001.
Effect of health insurance coverage on unintended pregnancy
Women’s demographic, socio-economic and behaviour factors before and after propensity score matching are shown in Table 4. Before matching, the analysis revealed that the unintended pregnancy according to the women’s health insurance status varied significantly across the population subgroups, as defined by their demographic, socio-economic, and behavioural factors. Those who were insured with unintended pregnancy were younger women compared those who were uninsured with unintended pregnancy. Insured women with unintended pregnancies were more likely to be living in poorer households, urban dwellers, and attained secondary education and above. The results showed that 2,867 insured women with unintended pregnancies were successfully matched to 1,204 women who were uninsured with unintended pregnancies. The findings revealed that after propensity score matching, the difference in the unintended pregnancy among those who were insured and uninsured across the population subgroups by their socio-demographic and behavioural characteristics was reduced. In addition, the absolute standardized differences of the propensity score after matching were below 10% for most of the variables, indicating good quality matches (Table 4).
Table 4. Standardised Differences across Covariates: before Matching and after Matching
Note: HIC = Health Insurance Coverage.
*The standardised percentage bias. The standardised % bias is the % difference of the sample means in the treatment and control (full or matched) sub-samples as a percentage of the square root of the average of the sample variances in the treated and non-treated groups (McLemore et al., Reference McLemore, Berkowitz, Oltman, Baer, Franck, Fuchs, Karasek, Kuppermann, McKenzie-Sampson, Melbourne, Taylor, Williams, Rand, Chambers, Scott and Jelliffe-Pawlowski2021).
The estimated likelihood of an unintended pregnancy for each identified strata of insured and uninsured women was calculated using propensity score matching. Women were divided into strata groups based on the propensity score to predict the probability of unintended pregnancy by health insurance enrolment (Table 5). The overall probability of unintended pregnancy was 0.312 among insured women and 0.351 for those women who are uninsured. According to the table, insured women have a lower probability of unintended pregnancy than uninsured women across all strata groups except for those in Strata One. Both insured and uninsured women in Strata One have equal probability to experience unintended pregnancy. Each strata group had an equal amount of women, with a difference of eight women between the strata with the highest and lowest numbers of women. The directly adjusted probability of unintended pregnancy was an average of the subclass total weight for insured and uninsured women.
Table 5. Subclass probabilities of unintended pregnancy among insured and uninsured women in Ghana, GDHS, 2014
Note: CI=confidence interval.
a Based on estimated propensity score.
b Standard errors for the adjusted probabilities were calculated following the method described by David M. Shoemaker (Reference Shoemaker1972).
Discussion
Unintended pregnancy remains an important public health and social issues for policy makers and donors as well as program implementers in sub-Saharan African (SSA) nations, especially among young women (Singh & Darroch, Reference Singh and Darroch2000; Oyediran, Ishola & Adewuyi, Reference Oyediran, Ishola and Adewuyi2002). Poor health outcomes and socio-economic consequences have been associated with an unintended pregnancy. High preterm birth and low birthweight have been associated with unintended pregnancies (Cleland et al., Reference Cleland, Peipert, Westhoff, Spear and Trussell2011; Cheng et al., Reference Cheng, Schwarz, Douglas and Horon2009; Oyediran, Ishola & Bankole, Reference Oyediran, Ishola and Bankole2020). In addition, an unintended pregnancy often obstructs educational and career goals that may severely affect future earning potential and family well-being. On the other hand, studies have demonstrated that health insurance increased access to health care and decreased poor sexual and reproductive health outcomes (Kavanaugh, Douglas-Hall & Finn, Reference Kavanaugh, Douglas-Hall and Finn2019; Sommers et al., Reference Sommers, Gunja, Finegold and Musco2015; Dixon et al., Reference Dixon, Tenkorang, Luginaah, Kuuire and Boateng2014). No study has investigated the relationship between unintended pregnancies and health insurance coverage in SSA despite that unintended pregnancy has become endemic in the region. The paper examined effect of health insurance coverage on unintended pregnancy with the analysis of the 2014 GDHS.
The important finding of this study is that insured women were less likely to experience unintended pregnancy as compared to uninsured women regardless of their socio-demographic characteristics. The study demonstrated the effect of health insurance coverage of reducing unintended pregnancy if those without health insurance had insurance coverage. For instance, the study showed that probability of insured women experienced an unintended pregnancy was 0.312 compared with 0.351 among uninsured women. Studies have demonstrated that health insurance coverage influence contraceptive use because enrollment in NHIS have found to reduce inequalities in access to sexual and reproductive health services including contraception (Nearns, Reference Nearns2009; Kost, Finer & Singh, Reference Kost, Finer and Singh2012; Arthur, Reference Arthur2012; Finlayson & Downe, Reference Finlayson and Downe2013). This result implies that health insurance coverage may facilitate women access to contraceptive methods that women can use to prevent pregnancy, especially among those at risk of pregnancy but may not be able to meet the cost of contraception. More importantly, women with insurance coverage may access methods that require interaction with the health care system like long-acting reversible contraception (LARC) methods and shorter-acting hormonal methods such as the pill, injectable (Pickle, Wu & Burbank-Schmitt, Reference Pickle, Wu and Burbank-Schmitt2014). This study demonstrates that health insurance coverage can play a significant role in reducing unintended pregnancy in Ghana and sub-Saharan African in general. Because the implementation of health insurance scheme has the potential of aiding access to contraceptive methods, especially those methods that are more effective, costly, or provide long-term coverage but require consultation with a health care provider in order to access. In addition, health insurance coverage is a mechanism that works to remove the financial barriers, and improves access to birth control, in turn, reducing the burden of unplanned pregnancies experienced (Peipert et al., Reference Peipert, Madden, Allsworth and Secura2012).
The study showed that a significant percentage (66%) of Ghanaians had health insurance coverage for women who had had births in the previous 2 years, in contrast to the reported health insurance coverage of less than 5 percent among women and men interviewed in thirty low-middle-income countries (Wang, Temsah & Mallick, Reference Wang, Temsah and Mallick2013). The results show that health insurance coverage varied according to the women’s individual and household level factors. For instance, the findings revealed an inverted U-shaped relationship between women’s age and health insurance coverage. For instance, the health insurance coverage peaked (69.6%) among women aged 25-34 years. The findings are consistent with the previous studies that demonstrated an increased probability of enrolment for health insurance with advanced age (Salari et al., Reference Salari, Akweongo, Aikins and Tediosi2019; Mulenga et al., Reference Mulenga, Mulenga, Musonda and Phiri2021; Kazungu & Barasa, Reference Kazungu and Barasa2017). An inverted U-shaped relationship between women’s age and health insurance enrolment may be attributed to the likelihood of women in 25-34 years to be working with employer health insurance coverage or working spouse with employer insurance coverage. In addition, the increased risk of developing health conditions associated with aging could further explain higher rates of enrolment in health insurance as compared with those below 25 years of age (Kiplagat, Muriithi & Kioko, Reference Kiplagat, Muriithi and Kioko2013).
Another key factor that increases the probability of signing-up for health insurance is the educational level attained by women. This may be because education is an enabling and opportunistic variable. For instance, educated women would be informed and understand their health needs, thus, able to make informed choices (Alesane et al., Reference Alesane and Anang2018), including a strategy to reduce out-of-pocket expenses during health care services. In addition, education creates the opportunity to get jobs that provides employer health insurance coverage or to get well-paying jobs that give a higher propensity to purchase private health insurance coverage (Mulenga et al., Reference Mulenga, Mulenga, Musonda and Phiri2021). The present study found an increased probability of health insurance coverage with the household wealth index. This implies that women in the highest household wealth quintile have the highest health insurance coverage. This result is consistent with a previous study carried out in low and middle-income countries (Wang, Temsah & Mallick, Reference Wang, Temsah and Mallick2013) and four sub-Saharan African countries (Amu et al., Reference Amu, Dickson, Kumi-Kyereme and Darteh2018).
The health insurance coverage was higher among urban dwellers than among the women residing in rural area. This may be due to the high likelihood of many urban dwellers working in blue-collar jobs that provides health insurance cover compared to rural dwellers that are mostly working in the informal sector. The informal sector such as artisan and farmers may find it challenging to sign-up for health insurance coverage due to financial costs (Ibok, Reference Ibok2012). In addition, people in rural areas are less educated and potentially working in low pay jobs.
Furthermore, the study showed a regional variation in the health insurance coverage among women in Ghana. The regional variation in the health insurance sign-up can be attributed to the differences in the demographic, socio-economic, and structural composition of the population. For instance, fertility and use of modern contraception varies by residence and region (Ghana Statistical Service, Ghana Health Service & ICF International, 2018). It means that the individual’s plan on how to meet their reproductive health needs, including pregnancy goals, varies by region. Moreover, sexual and reproductive health interventions vary by region in Ghana, especially with donors supported activities not uniformly implemented at the regional level. Additional research is needed to understand the contextual factors that can enable or hinder health insurance enrolment and the ability to prevent unwanted pregnancies. Other women’s individual or exposure variables that increased the probability of having health insurance coverage include: being monogamously married, married at age 18 years or older, ability to participate in household decision-making, and exposure to the more mass media channels. These attributes are the agency of women’s ability to make decisions about sexual and reproductive health and rights. Thus, they can be described as an enabler or opportunistic factor that can increase women to sign-up for health insurance coverage.
Limitations and future research
This is the first study that examines the nexus between health insurance coverage and unintended pregnancy in Ghana. Other existing studies on health insurance coverage have focused on examining the prevalence of and factors associated with its sign-up. While few studies have examined the nexus between health insurance coverage and utilization of health care. This study established that health insurance coverage plays an important role in reducing unwanted pregnancy among women. Despite its strengths, the study had limitations. Many factors that influence health insurance enrolments, such as premiums, benefits covered under health insurance, quality of health, among others, are not available in the GDHS. Thus, the analysis was restricted to the variables found in the 2016 GDHS despite that DHS has been widely adjudged a golden survey to measure fertility and health-related indicators. Although, the wealth index can be used as a proxy for premiums. So, future research on health insurance coverage should consider collecting the data that will help to capture premiums, quality of health, and benefits covered. In addition, the GDHS was cross-sectional, which can lead to a lack of clear understanding in the temporal ordering among variables. For example, if unintended pregnancy or ill-health has a causal relationship with health insurance coverage, or if health insurance coverage is a coping mechanism utilized after experiencing unintended pregnancy. There is a need to establish if experience of unintended pregnancy precedes the sign-up for health insurance coverage. In addition to the current study, there is a need for more research to examine the relationship between health insurance and sexual reproductive health, including contraceptive.
Despite the limitation, the study findings are significant because education negatively and positively associated with unintended pregnancy and health insurance coverage respectively is worth noting. Although higher education was not the only factor found in this study to have such a relationship with both unintended pregnancy and health insurance coverage, it is the only factor may help individuals and enables them to make informed choices. Hence, health insurance coverage would be boosted and lead to decline in unintended pregnancy by aiding access to contraceptive methods. In addition, implementing community health insurance schemes will enhance individuals and households working in the informal sector to sign-up for coverage
Conclusion
The study highlighted the importance of examining the nexus between health insurance coverage and sexual and reproductive health in countries in Sub-Sahara Africa, such as Ghana. Nearly two thirds of women reported health insurance with 30% of those with health insurance coverage reported unintended pregnancy and 37% of those without insurance coverage reported unwanted pregnancies. A key result is that health insurance coverage is significantly negatively associated with unintended pregnancy among women, indicating that health insurance coverage has a protective effect against unintended pregnancy. The result underscores the role of health insurance in aiding access to preventive measures that can be hindered by financial costs. Hence policy to strengthen health insurance schemes to ensure everyone participates in it would help meet universal health insurance coverage. For instance, those in the formal sector should be enrolled in health insurance schemes through their employer. While those engaged in the informal sector can be enrolled in the health insurance program through community-based societies. The insurance premium can be paid in-kind by such as agricultural products to encourage those who may not be able to pay cash to participate. The study further demonstrates that there are variations in the women’s characteristics and factors influencing health insurance sign-up, the significant predictors, including age, education, religious affiliation, region of residence, household wealth, type of marital union, age at first cohabitation, and exposure to mass media. Similar factors and characteristics aforementioned associated with unintended pregnancies among Ghanaian women after controlling for health insurance coverage. This points to the fact that there are differentials in the factors that influence health insurance coverage and unintended pregnancies, and policymakers and donors should consider this in formulating policy and design of an intervention to ensure universal health coverage.
Acknowledgments
The authors acknowledge the permission granted by the Measures DHS to use the 2014 Ghana Demographic and Health Survey data. We would like to thank Dr. Akin Bankole, Senior Fellow at the Guttmacher Institute, New York, USA, for the guidance on the analysis and thoughtful review of an earlier draft. More importantly, the authors acknowledge the paper reviewers, who provided new insights to improve this work for publication. The views and opinions expressed in this paper are the authors’ alone and do not necessary reflect that of the John Snow Inc.
Funding
No financial support was received by the author(s) for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
