Examining the relationships between contraception and fertility rate in Ghana: Evidence from the 1988 to 2014 Ghana Demographic and Health Surveys

Study design

This study is a repeated cross-sectional study of six distinct national surveys from Ghana. These surveys were conducted between 1988 and 2014 among women of reproductive age (15–49 years old). This study design helped us to investigate the pattern of association between contraceptive use and fertility within a 26-year period in Ghana.

Data collection

We used secondary data from all the six Ghana Demographic and Health Surveys of 1988, 1993, 1998, 2003, 2008, and 2014.^12–17 All the surveys used standardised DHS model questionnaire which was developed by the Measure DHS programme. The Ghana Demographic and Health Survey is a nationwide survey executed every five years, except in 2014 which was delayed by one year. The survey is implemented by the Ghana Statistical Service and the Ghana Health Service in collaboration with the ICF International, which offers technical support through MEASURE DHS Programme. The focus of the survey is on maternal and child health, and it gathers relevant data for monitoring the population and health situation of Ghana. It collates data on an array of demographic and health issues such as contraceptive use, fertility, child health, nutrition, malaria, HIV/AIDS, health insurance, antenatal care, delivery care, and post-natal care. In this present study, sample included sexually active women from the six surveys.

Variables and measurement

Women’s cumulative fertility (i.e., number of children ever born) was the outcome variable for this study. This was a count variable. The main independent variable was contraceptive use. All women who reported that they were using contraceptives, either traditional or modern were categorized as “yes”, whilst those who reported that they were not using contraceptives at all were categorized as “no”. Besides, eight covariates were included, namely age (15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49), education (no education, primary, secondary, higher), union (never in a union, currently in union/living with a man, formerly in union/living with a man), wealth (poorest, poorer, middle, richer, richest), residence (urban, rural), region (Western, Central, Greater Accra, Volta, Eastern, Ashanti, Brong Ahafo, Northern, Upper East, Upper West), knowledge of fertile period within the ovulatory cycle (during her period, after period ended, middle of the cycle, before the period begins, at any time, don’t know) and religion (Catholic, Other Christian, Islam, Traditionalist, no religion). No data were collected on wealth status and knowledge of the fertile period within the ovulatory cycle in 1988 and 1998, respectively. These variables were selected as covariates as a result of their significant association with fertility as reported by previous studies.^18–20

Data analysis

The data were managed and analysed with Stata (version 14). Descriptive analyses were conducted to compute the socio-demographic characteristics of the research participants (see Table 1) and the trend of fertility from 1988 to 2014 (see Figure 1). Considering that the outcome of interest was a count variable without evidence of overdispersion, we conducted Poisson regression to investigate the factors that have underpinned Ghana’s fertility between 1988 and 2014 at a 95% confidence interval. Model 1 is for the first survey (i.e., 1988) and follows chronologically the model for the 2014 survey (Model 6). We exponentiated the Poisson regression coefficients and reported findings as Incidence Rate Ratio (IRR). The data were weighted with the complex survey design to adjust for the multi-stage sampling nature and to ensure that we generate nationally representative and accurate results. Variance inflation factor was used to investigate multicollinearity among the independent variables.

Ethical considerations

In all the surveys,^12–17 ethical approval for the data collection was provided by the Ethical Review Committee of Ghana Health Service, Accra, Ghana, and the Institutional Review Board of ICF International, USA.^12–17 We noticed from these ethics’ approval documents that, each participant engaged in the various GDHS provided verbal or written informed consent prior to data collection. In the case of participants below 18 years (the statutory age for adulthood), parental or legal guardian’s consent was secured in addition to the respondent’s assent.^12–17

While the original work that provided the data for this work had ethical review and approval,^12–17 this report in its own right also obtained ethical approval from the Institutional Review Board and Ethical Review Committee of the Ghana Health Service, Accra, Ghana (GHS-ERC 009/12/21) because we reported on health related information which was predominantly generated through the Ghana Health Service, hence a requirement to seek approval in this particular study as well. However, because we used a secondary data, informed consent was not obtained from individual participants; rather the consent for data use was obtained from the Ghana Statistical Service in line with the use of GDHS data.

Socio-demographic characteristics

Table 1 presents the socio-demographic characteristics of the research participants. The analysis of age revealed that women aged 25–29 years old dominated in all the surveys as evidenced in 1988 (21.4%) and 2014 (19.0%). However, those aged 40–44 and 15–19 were the least in the 1988 (9.0%) and 2014 (8.5%) surveys, respectively. More than half of the participants for the 1988 survey had primary education (50.1%), meanwhile, those with secondary education dominated in 2014 (54.7%). Those with higher education were consistently few, as noted in the 1988 (0.9%) and 2014 (6.3%) surveys. Women in union or living with a woman persistently dominated. For instance, almost eight out of 10 of the participants were in union/living with a man (76.5%), whilst 64.8% were in union/living with a man in 2014. Only the 1988 survey lacked wealth data and for the surveys with wealth data, and the analysis revealed dominance of “richest” women in 1993 (24.5%) and 2014 (22.9%) correspondingly.

Most of the research participants were in rural locations in 1988 (67.3%). However, more than half of the women were in urban locations in 2014 (53.2%). During the 1988 survey, two of the current administrative regions of Ghana were part of the Northern region (Upper East and Upper West Regions), and that is why no results were recorded for those regions under the 1988 survey (see Table 1). As displayed in the table, a significant proportion of the research participants resided in the Ashanti region (17.6%). However, in 2014, a greater proportion of the research participants were residents of the Greater Accra region (20.0%). Regarding women’s knowledge of their fertile period during the ovulatory cycle, most of them reported that they did not know the fertile period during the cycle in 1988 (42.4%), whilst 38.4% admitted that it was the middle of the cycle (38.4%). Lastly, on religion, a greater proportion of the participants were other Christians both in 1988 (52.1%) and 2014 (70.5%), as shown in Table 1.

Fertility trend: 1988–2014

As shown in Figure 1, Total Fertility Rate (TFR) in Ghana declined from 6.4 births per woman to 4.2 births per woman between 1988 and 2014. Yet, it is worth noting that there was a decline from 1988 to 2008 (4.0 births per woman), before it slightly increased to 4.2 births per woman in 2014.

Figure 1. Fertility trend: 1988-2014.

Source: GDHS 1988, 1993, 1998, 2003, 2008, 2014.

TFR = Total Fertility Rate.

Inferential results

Tables 2 and 3 present the inferential results of the study. Table 2 presents the bivariate results, whilst Table 3 presents the multivariate results. From the bivariate models, we noted that between 1988 and 2014, women who reported that they were using contraception had higher prospects of bearing more children relative to women who were not using contraception. This manifested in 1988 (IRR = 1.16, 95% CI = 1.11, 1.22) and 2014 (IRR = 1.20, 95% CI = 1.12, 1.29) as shown in Table 2.

The multivariable Poisson regression models revealed the same pattern as women who were using contraception showed chances of higher fertility in all the survey rounds, including that of 1988 (IRR = 1.12, 95% CI = 1.09, 1.19) and 2014 (IRR = 1.13, 95% CI = 1.09, 1.17). Throughout the period, women in all age categories had a higher incidence of fertility relative to 15–19-year-old women, especially among 45–49-year-old women in 2003 (IRR = 13.36, 95% CI = 11.16, 15.99).

The analysis revealed that highly educated women had a lower incidence of fertility compared to those without formal education, with the least incidence in 1998 (IRR = 0.52, 95% CI = 0.46, 0.61). Notably, women who reported that they were currently in a union or living with a man had a higher fertility incidence relative to those who had never been in a union (IRR = 11.58, 95% CI = 8.90, 15.06) in 1993. In 2014, wealthy women had a lower fertility incidence than poor women (IRR = 0.58, 95% CI = 0.55, 0.62). From 1988 to 2014, rural women were noted to have a higher fertility incidence compared to urban women, particularly in 1998 (IRR = 1.19, 95% CI = 1.13, 1.25).

Analysis of fertility across the erstwhile 10 administrative regions showed that, relative to the Western region, the incidence of fertility was lower in the Upper West region both in 1998 (IRR = 0.85, 95% CI = 0.77, 0.92) and 2003 (IRR = 0.85, 95% CI = 0.77, 0.93). Women of other religions had a higher fertility incidence in 2003 relative to Catholics (IRR = 1.43, 95% CI = 1.28, 1.60) and this trend permeated all the survey rounds, as shown in Table 3.

Weaknesses and strengths

The study has a number of limitations and outstanding strengths. This study employed rigorous analytical procedures to investigate the association between contraception and fertility ratio over a 26-year period. It offers insights into the critical factors to be prioritized regarding fertility regulation in Ghana. Also, the representativeness of the sample strengthens the generalisability of our findings to all Ghanaian women within the reproductive age bracket. In spite of these strengths, caution is required in the interpretation of our findings as a causal inference cannot be made between contraception, women’s socio-demographic characteristics, and fertility.

Implications for practice

This study investigated the relationship between Contraceptive Prevalence Rate (CPR) and Total Fertility Rate (TFR) between 1988 and 2014. This study did not show an inverse relationship between CPR and TFR, as reported by earlier studies. Findings from the study suggest that contraception use alone does not necessarily translate into reduced fertility in Ghana, but that other several factors might account for Ghana’s fertility trend between CPR and TFR.

Ethical approval

The study did not use human subjects hence; ethical approval was not required. However, the paper was written in accordance with the relevant guidelines and regulations of the journal. Permission was obtained from all persons and institutions that contributed information to the study in various forms and they are also duly acknowledged as deemed necessary.