Multilevel analysis of factors predicting pregnancy loss experiences among pregnant women in Ghana: a further analysis of nationally representative data : Global Reproductive Health

Journal Logo

Original Article

Multilevel analysis of factors predicting pregnancy loss experiences among pregnant women in Ghana: a further analysis of nationally representative data

Klu, Desmond

Author Information
Global Reproductive Health 7(4):p e63, Winter 2022. | DOI: 10.1097/GRH.0000000000000063
  • Open

Abstract

Pregnancy loss can occur through miscarriage, stillbirth, and abortion (spontaneous or induced). Global statistics indicate that nearly 15% of clinically identified pregnancies end in recognized miscarriage, a pregnancy loss before 20 weeks of gestation1,2. More than one-third of all conceptions that can be identified hormonally may end in loss when taking into account unrecognized pregnancies3. It is estimated that fewer than 5% of women will experience 2 consecutive miscarriages, and only 1% experience 3 or more4.

Over 3 million stillbirths occur globally each year, nearly all of which are in low-income countries5–8. Of the 130 million babies born worldwide every year, ~4 million are stillborn, and more than 98% of these occur in developing countries9. Stillbirth accounts for more than half of perinatal mortality in developing countries by ~26 per 1000 live births, ~5 times higher than in developed countries. One-fourth to one-third of all stillbirths are estimated to take place during delivery.

Within the sub-Saharan African context, stillbirth rates are particularly high, where up to 14% of deliveries could result in stillbirths. According to the World Health Organization’s Opportunities for Africa’s Newborns 2006 report, 98% of stillbirths occur in developing countries, especially sub-Saharan Africa.

The term “pregnancy loss” includes stillbirths of at least 20 weeks’ gestation and reported abortions of under 20 weeks gestation, but not neonatal deaths. It is inherently difficult to obtain complete data on pregnancy loss in a population. This difficulty arises in connection with abortions rather than with stillbirth10. Again, pregnancy loss refers to the unexpected loss of an unborn baby.

In 2011, 1.06 million abortions were performed, down 13% from 1.21 million in 2008. From 1973 through 2011, nearly 53 million legal abortions occurred in the United States11. Pregnancy loss can reduce effective fecundity, lengthen birth intervals, and decrease fertility. “Indeed, variation in fecundability may be more sensitive to heterogeneity in risk of fetal loss than it is to variation in coital frequency”12.

In the case of Ghana, according to the Ghanaian constitution, abortions are illegal despite having one of the most liberal abortion laws in sub-Saharan Africa13. However, it is estimated that 20% of pregnancies in Ghana are aborted14. In Ghana, it has also been established that abortion-related deaths are the most frequent cause of maternal mortality15. A community-based study in Ghana in 1998 found that only 12% of the women obtaining induced abortion utilized physician services for the procedure16. A hospital-based study also found that 18% of gynecology admissions in 2000 were related to complications of induced abortion. Furthermore, out of a total of 105 maternal deaths recorded at the Korle Bu Teaching Hospital, 14% were due to complications of induced abortion17. Again, according to the 2007 Ghana Maternal Health report, more than 1 in 10 maternal deaths results from complications of induced abortion, which is also the second leading cause of maternal death in Ghana18.

Concerning stillbirth experiences among women in Ghana, an autopsy conducted at the Korle Bu Teaching Hospital (KBTH) found that out of a total of 3761 deliveries, 93 of those deliveries ended up in stillbirths, where 54.8% were males19. A hospital-based study found that 54.8% of stillborns were males and 45.2% were females. In addition, 55.9% of the stillbirths were macerated, and 44.1% were fresh stillbirths20. The same study found that low maternal educational level, low socioeconomic status, and late and irregular antenatal attendance were present in the majority of the mothers who had stillbirths. The general lack of proper and rigorous medical examination for stillbirths in Ghana could easily lead to a recurrence loss of pregnancy21.

Another hospital-based study at the KBTH found that 42.0% of pregnancy cases among women had experienced at least one miscarriage in their lifetime22. From the introduction above, it could be clearly seen that although globally the incidence and prevalence of pregnancy loss (stillbirth, miscarriage, and abortion) is low, the situation seems to be different on the side in developing countries, in sub-Saharan Africa and Ghana in particular. Most hospital-based studies on pregnancy loss experiences among women in Ghana have clearly shown that there is a high rate of pregnancy loss experience among women20–22.

Although previous studies have examined factors influencing pregnancy loss experiences among pregnant women across different geographical locations, including Ghana13–22, to the best of my knowledge, none of the studies have focused on how individual, proximate, and household factors interact to influence the experience of pregnant women losing their pregnancy either through stillbirth, miscarriages or abortion. This paper, therefore, seeks to examine the individual, proximate, and household factors at the multilevel level on pregnancy loss experiences among pregnant women in Ghana using the 2014 Ghana Demographic and Health Survey (GDHS).

Methods

Study design and population

The study is based on data obtained from the 2014 GDHS. The GDHS is a nationally representative cross-sectional survey that collects information on issues such as housing characteristics and household population, marriage and sexuality, fertility and fertility preferences, family planning, infant and child mortality, maternal health, and child and early development. It further includes issues on nutrition of children and women, malaria, HIV and AIDS related knowledge, attitudes, and behavior, HIV prevalence, adult health and lifestyle, women empowerment, and demographic and health outcomes. This study focuses on women within their reproductive ages 15–49 who had ever had a pregnancy. The study used information on the demographic and social characteristics of the women.

Sample size

In the 2014 GDHS, women who had ever had a pregnancy in their reproductive ages 15–49 years. The women’s data file utilized in this study was weighted to obtain a sample of ever pregnant women of 2321.

Variable definitions and measurements

Outcome variable

The outcome variable comprises pregnancy loss among women. This study considers ever pregnant women who have ever experienced pregnancy loss through stillbirth, miscarriage or abortion regardless of the number of pregnancies they ever had. The following question was used to determine pregnancy loss among women: “Have you ever had a pregnancy that was miscarried, was aborted, or ended in a stillbirth?” A sample of women who had ever been pregnant before the period within which the survey was conducted was filtered out of the weighted data set. Women who lost their pregnancy were coded “1” and “0” otherwise.

Predictor variables

Proximate factors

The proximate (intermediate) factor considered in this study comprised the number of antenatal care visits and access to health care facilities. The number of antenatal care visits was measured as the number of times women visited health facilities for antenatal care during pregnancy and was categorized as (1=no visit) (2=1–3 visits) (3=4 or more visits). Access to health care facilities was measured by the convenience of the location of the health facility from the pregnant women’s house and was categorized as 1=easy access and 2=difficulty access.

Individual factors

The individual-level sociodemographic factors included the age of women (15–29, 30–39, 40–49), educational level (no education, primary, secondary/higher), religion (catholic, protestant, Moslem, Pentecostal/Charismatic, other Christians, no religion), ecological zones of residence (coastal zone, middle belt, northern zone), place of residence (urban, rural), marital status (never married, currently married, ever married), ethnicity (Akan, Ga/Dangme, Ewe, Mole-Dagbani, Others) and parity (1–3, 4–6, 7 or more) and occupational type (no occupation, professional/managerial/technical/clerical, sales, services, agriculture, and skilled/unskilled manual job).

Household factors

The following household factors were considered in the study: sex of household head (male and female), age of household head (20–29,30–39, 40–49, 50–59, 60–69, 70+) and household wealth quintile (poorest, poorer, middle, richer, richest). The other variables included the household source of drinking water the and type of toilet facility used by the household. The measurement and classification of the variable “household source of drinking water” was guided by the WHO/United Nations International Children’s Emergency Fund Joint Monitoring Program for Water Supply, Sanitation and Hygiene (WHO/UNICEF-JMP) classification of source of drinking water. For this study, the variable was classified into 2 categories: improved and unimproved sources of drinking water. In this study, the improved source of drinking water comprised pipe-borne water inside the dwelling, piped into the dwelling, pipe to yard/plot, piped to the neighbor’s house/compound, tube well water, borehole, protected dug well, protected well, protected spring and rainwater collection, bottled water, and sachet water. The unimproved source of drinking water in this study included unprotected wells, surfaces from spring, unprotected springs, rivers/dam, tanker trucks, and carts with small tanks. The type of toilet facility was also categorized into 2 improved and unimproved. The classification of improved toilet facilities was also guided by the WHO/UNICEF-JMP classification of sanitation technologies. The improved toilet facilities in this study comprised flushing to pipe sewers, flushing to septic tanks, flushing to pit latrines, flushing to unknown places, flushing to bio-digesters, ventilated improved pit latrines, pit latrines with slabs, pit toilet latrines, and composting toilets. The unimproved toilet facility included flush to somewhere else, a pit without slab/open pit, no facility, bush/field, and hanging toilet/latrine.

Data analysis

Data were analyzed with SPSS version 25. In analyzing the data, 3 stages were followed. The first stage was the use of simple descriptive statistics to describe the outcome and predictor variables. The second stage involved a cross-tabulation of all the individual and household and proximate factors against the pregnancy loss experiences among ever pregnant women. In the third stage, 4 different models were developed that involved multilevel binary logistic regression analyses to examine the effect of individual and household and proximate factors on ever pregnant women’s experiences in losing their pregnancies. Model I analyzed the effect of only individual factors, model II analyzed the effect of only household factors, model III analyzed the effect of only proximate factors, and model IV analyzed the combined effect of individual, household and proximate factors on pregnancy loss experiences among ever pregnant women. For all 4 models, the adjusted odds ratios (AORs) and their associated 95% CIs are presented. In addition, sample weight (v005/1,000,000) in weighting the entire data to correct possible over- and under sampling issues was applied.

Results

Descriptive

Pregnancy loss among ever pregnant women in Ghana

Figure 1 illustrates the percentage distribution of pregnancy loss among ever pregnant women in Ghana. Pregnancy loss experience in this study incorporates termination of pregnancy through miscarriages, stillbirths, and both induced and spontaneous abortion by women. Little than half (52%) of the ever pregnant women had never experienced any form of pregnancy loss, while 4 of 10 women had ever experienced a form of pregnancy loss.

F1
Figure 1:
Pregnancy loss experiences among ever pregnant women in Ghana.

Description of predictor variables in the study

Table 1 shows the proximate factors and the individual sociodemographic characteristics of ever pregnant women in Ghana. Approximately 54% of ever pregnant women never attended antenatal care, while 42.5% had 4 or more antenatal visits. More than half (58.2%) of the ever pregnant women had difficulty accessing health facilities, while the remaining 41.8% easily accessed health facilities during pregnancy. Approximately 39% of the respondents were aged 30–39 years, constituting the highest proportion of ever pregnant women of the age categories. With regard to residence, a higher proportion (60.0%) of ever pregnant women resided in the coastal/southern zone compared with other ecological zones of the country. More than half (61.4%) of the ever pregnant women resided in urban areas, with a higher proportion (53.8%) belonging to the Akan ethnic group. More than half (63.0%) had attained a secondary or higher level of education. Concerning religion, a higher proportion (45.5%) of ever pregnant women belonged to Pentecostal/Charismatic faith compared with other religious faith. Approximately 7 of 10 ever pregnant women in Ghana are currently married, while most (47.9%) of them are into sale as their occupational type.

Table 1 - Proximate and individual sociodemographic characteristics of ever-pregnant women in Ghana
Variables Weighted sample, n=2 321(%)
Proximate factors
 Number of antenatal visits
  No ANC visits 1259 (54.2)
  1–3 visits 75 (3.3)
  4+ visits 987 (42.5)
 Access to health facility
  Easy access 971 (41.8)
  Difficult access 1350 (58.2)
Individual factors
 Age
  15–29 803 (34.6)
  30–39 909 (39.2)
  40–49 609 (26.2)
 Place of residence
  Urban 1425 (61.4)
  Rural 896 (38.6)
 Education level
  No education 411 (17.7)
  Primary 448 (19.3)
  Secondary/Higher 1463 (63.0)
 Ecological zones of residence
  Coastal zone 1392 (60.0)
  Middle belt 762 (32.8)
  Northern zone 168 (7.2)
 Ethnicity
  Akan 1249 (53.8)
  Ga/Dangme 211 (9.1)
  Ewe 436 (18.8)
  Mole-Dagbani 316 (13.6)
  Other 110 (4.7)
 Religion
  Catholic 193 (8.3)
  Protestant 317 (13.6)
  Pentecostal/Charismatic 1055 (45.5)
  Other Christians 369 (15.9)
  Moslem 277 (11.9)
  No religion 110 (4.7)
 Marital status
  Never married 378 (16.3)
  Currently married 1615 (69.6)
  Formerly married 328 (14.1)
 Occupational type
  Not working 270 (11.6)
  Professional/Technical/Managerial/Clerical 155 (6.7)
  Sales 1113 (47.9)
  Services 61 (2.6)
  Agriculture 391 (16.9)
  Skilled/unskilled manual 331 (14.3)
Source: Computed from 2014 Ghana Demographic and Health Surveys (GDHS).

Table 2 shows the household characteristics of ever pregnant women. The results showed that 60.3% of ever pregnant women belonged to male-headed households. The highest proportion (33.9%) of heads of household were between the ages of 30 and 39 years. Approximately 8 of 10 ever pregnant women belong to households that access improved sources of drinking water as well as improved toilet facilities. The highest proportion of (26.0%) ever pregnant women belonged to the richer household wealth quintile category, with the lowest proportion (10.0%) belonging to the poorest wealth quintile.

Table 2 - Household characteristics of ever pregnant women in Ghana.
Household level factors Weighted sample, n=2321 (%)
Sex of household head
 Male 1399 (60.3)
 Female 922 (39.7)
Age of household head
 20–29 344 (14.8)
 30–39 783 (33.7)
 40–49 757 (32.6)
 50–59 254 (10.9)
 60–69 126 (5.4)
 70+ 57 (2.4)
Household source of drinking water
 Improved source of drinking water 2068 (89.1)
 Unimproved source of drinking water 254 (10.9)
Household type of toilet facility
 Improved toilet facility 1779 (76.7)
 Unimproved toilet facility 542 (23.3)
Household wealth quintile
 Poorest 232 (10.0)
 Poorer 340 (14.7)
 Middle 517 (22.3)
 Richer 604 (26.0)
 Richest 628 (27.1)
Source: Computed from 2014 Ghana Demographic and Health Surveys (GDHS).

Association between individual, proximate and household level factors and pregnancy loss experiences among ever pregnant women in Ghana

Tables 3 and 4 shows the strength of association with χ2 analyses between the individual, proximate, and household factors and pregnancy loss experiences among ever pregnant women in Ghana. Proximate factors such as the number of antenatal visits (P=0.005) had a significant association with pregnancy loss experiences. Individual sociodemographic factors, including women’s age (P=0.000), educational level (P=0.000), place of residence (P=0.001), religion (P=0.026), marital status (P=0.000), and occupational type (P=0.000), were found to be significantly associated with pregnancy loss at P<0.05. With regard to household factors, age of household head (P=0.000) and wealth index (P=0.000) were significantly associated with pregnancy loss experiences among ever pregnant women in Ghana at P<0.05.

Table 3 - Association between individual, proximate and household factors and pregnancy loss experience among ever pregnant women in Ghana.
Pregnancy loss experience
Factors Not loss Loss P
Proximate factors
 No antenatal visits 0.005**
  No ANC visits 49.0 51.0
  1–3 antenatal visits 50 50.0
  4+ antenatal visits 55.9 44.1
 Access to health facility 0.067
  Easy access 54.2 45.8
  Difficult access 50.4 49.6
Individual factors
 Age 0.000***
  15–29 81.0 19.0
  30–39 45.9 54.1
  40–49 22.8 77.2
 Educational level 0.000***
  No education 41.2 58.8
  Primary 48.1 51.9
  Secondary/higher 56.2 43.8
 Place of residence 0.001**
  Urban 54.7 45.3
  Rural 47.8 52.2
 Ecological zone
  Coastal zone 53.4 46.6 0.170
  Middle belt 49.2 50.8
  Northern zone 53.3 46.7
 Ethnicity
  Akan 53.9 46.1 0.054
  Ga/Dangme 46.9 53.1
  Ewe 47.1 52.9
  Mole-Dagbani 55.1 44.9
  Other 51.4 48.6
 Religion
  Catholic 52.8 47.2 0.026*
  Protestant 53.3 46.7
  Pentecostal/Charismatic 48.2 52.8
  Other Christians 57.5 42.5
  Moslem 55.8 44.2
  No religion 55.5 44.5
 Marital status
  Never married 80.7 19.3 0.000***
  Currently married 49.6 50.4
  Formerly married 30.5 69.5
 Occupational type
  Not working 68.1 31.9 0.000***
  Professional/Technical/Managerial/Clerical 54.8 45.2
  Sales 49.9 50.1
  Services 40.0 60.0
  Agriculture 44.6 55.4
  Skilled/unskilled manual 55.3 44.7
Household factors
 Sex of household head 0.253
  Male 53.0 47.0
  Female 50.5 49.5
 Age of household head
  20–29 81.1 18.9 0.000***
  30–39 55.6 44.4
  40–49 37.8 62.2
  50–59 41.7 58.3
  60–69 52.0 48.0
  70+ 60.7 39.3
 Source of drinking water
  Improved 51.4 48.6 0.084
  Unimproved 57.1 42.9
 Type of toilet facility
  Improved 51.8 48.2 0.684
  Unimproved 52.8 47.2
 Wealth index
  Poorest 51.9 48.1 0.034*
  Poorer 44.6 55.4
  Middle 51.6 48.4
  Richer 53.3 46.7
  Richest 55.1 44.9
Source: Computed from 2014 Ghana Demographic and Health Surveys (GDHS).
*P<0.05.
**P<0.01.
***P<0.001.

Table 4 - Multilevel logistic regression of individual, proximate and household level factors influencing pregnancy loss experience among ever pregnant women in Ghana.
Variables Model I AOR [95% CI] Model II AOR [95% CI] Model III AOR [95% CI] Model IV AOR [95% CI]
Individual level factors
 Age
  15–24 Ref Ref
  25–34 4.20 *** [3.32–5.33] 3.81 *** [2.88–5.03]
  35–49 10.91 *** [8.20–14.51] 10.50 ** [7.25–15.21]
 Educational level
  No education Ref Ref
  Primary 0.83 [0.601.15] 0.81 [0.581.13]
  Secondary + 0.72 * [0.53–0.98] 0.71 * [0.52–0.97]
 Place of residence
  Urban 0.67 *** [0.53–0.83] 0.69 ** [0.53–0.91]
  Rural Ref Ref
 Ecological zone
  Coastal zone 0.91 [0.571.45] 0.82 [0.491.37]
  Middle belt 1.27 [0.792.03] 1.12 [0.671.85]
  Northern zone Ref Ref
 Ethnicity
  Akan 0.63 [0.381.04] 0.60 [0.361.00]
  Ga/Dangme 0.79 [0.441.42] 0.79 [0.441.43]
  Ewe 0.81 [0.481.38] 0.78 [0.461.34]
  Mole-Dagbani 0.67 [0.401.13] 0.62 [0.371.05]
  Other Ref Ref
 Religion
  Catholic 1.57 [0.912.72] 1.57 [0.912.73]
  Protestant 1.69 * [1.00–2.85] 1.76 * [1.04–2.98]
  Pentecostal/Charismatic 1.78 * [1.12–2.84] 1.79 * [1.12–2.87]
  Other Christian 1.22 [0.742.02] 1.23 [0.742.04]
  Moslem 1.14 [0.661.97] 1.12 [0.641.95]
  No religion Ref Ref
 Marital status
  Never married Ref Ref
  Currently married 1.80 *** [1.32–2.46] 1.89 ** [1.30–2.75]
  Formerly married 3.22 *** [2.18–4.76] 3.06 *** [2.05–4.55]
 Occupational type
  Not working Ref Ref
  Professional/Technical/Managerial/Clerical 1.56 [0.972.52] 1.79 * [1.09–2.94]
  Sales 1.55 ** [1.11–2.14] 1.65 ** [1.18–2.29]
  Services 2.56 ** [1.31–4.99] 2.82 *** [1.43–5.53]
  Agriculture 1.01 [0.681.50] 1.07 [0.711.61]
  Skilled/unskilled manual 1.14 [0.781.69] 1.22 [0.821.81]
Household level factors
 Sex of household head
  Male Ref Ref
  Female 1.40 *** [1.16–1.69] 1.26 [0.971.65]
 Age of household head
  2029 0.35 ** [0.19–0.64] 0.92 [0.451.91]
  3039 1.36 [0.772.40] 1.17 [0.602.30]
  4049 2.80 *** [1.59–4.93] 1.25 [0.642.44]
  5059 2.52 ** [1.38–4.61] 1.40 [0.692.87]
  6069 1.64 [0.853.14] 1.04 [0.482.24]
  70+ Ref Ref
 Household Source of drinking water
  Improved 1.59 ** [1.18–2.16] 1.52 * [1.08–2.13]
  Unimproved Ref Ref
 Household type of toilet facility
  Improved toilet facility 1.02 [0.801.29] 0.96 [0.741.26]
  Unimproved toilet facility Ref Ref
 Household wealth index
  Poorest 1.29 [0.901.85] 0.98 [0.571.68]
  Poorer 1.71 *** [1.27–2.30] 1.39 [0.912.13]
  Middle 1.24 [0.971.60] 1.05 [0.751.47]
  Richer 1.20 [0.951.52] 1.10 [0.831.45]
  Richest Ref Ref
Proximate factors
 No. antenatal visits
  No ANC visits Ref Ref
  1–3 visits 0.96 [0.611.53] 1.80 * [1.03–3.13]
  4+ visits 0.77 ** [0.65–0.92] 1.28 * [1.02–1.61]
 Access to health facility
  Easy access 0.89 [0.751.05] 0.84 [0.691.02]
  Difficult access Ref Ref
Model I= individual sociodemographic variables.
Model II=household variables.
Model III= proximate variables.
Model IV= final model adjusted for individual, proximate and household variables.
AOR indicates adjusted odds ratio; Ref, reference.
Source: Computed from 2014 Ghana Demographic and Health Survey (GDHS).
*P<0.05.
**P<0.01.
***P<0.001.

Individual, proximate and household factors predicting pregnancy loss experiences among ever pregnant women in Ghana

Table 4 presents the results on the individual, proximate and household factors that influence the experiences of pregnancy loss among ever pregnant women in Ghana. In model 1 (only individual factors), the age of ever pregnant women, educational level, place of residence, religion, marital status, and occupational type significantly predicted pregnancy loss experiences among pregnant women in Ghana. In terms of age, the results show that as the age of ever pregnant woman increases, the likelihood of pregnancy loss experience also increases. For instance, the ever pregnant women aged 25–34 years (AOR=4.20; CI: 3.32–5.33) and 35–49 years (AOR=10.91; CI: 8.20–14.51) have a higher probability of ever losing their pregnancy compared with those aged 15–24 years. Regarding educational level, the results showed that ever pregnant women with secondary/higher education (AOR=0.72; CI: 0.53–0.98) had lower odds of experiencing pregnancy loss than those with no formal education. With place of residence, compared with those who resided in rural areas, ever pregnant women who resided in urban areas (AOR=0.67; CI: 0.53–0.98) had lower odds of losing their pregnancy. Ever pregnant women who are protestants (AOR=1.69; CI: 1.00–2.85) and those belonging to the Pentecostal/Charismatic faith (AOR=1.78; CI: 1.12–2.84) are more likely to lose their pregnancy compared with those with no religion. Again, ever pregnant women who are currently married (AOR=1.80; CI: 1.32–2.46) and those formerly married (AOR=3.2; CI: 2.18–4.76) have higher odds of experiencing pregnancy loss than never married women. Again, ever pregnant women who are engaged in sales work (AOR=1.55; CI: 1.11–2.14) and services (AOR=2.56; CI: 1.31–4.99) had a higher likelihood of losing their pregnancy compared with those who are not working.

In the second model (only household variables), sex and age of household heads were significant predictors of pregnancy loss experiences among ever pregnant women in Ghana. Pregnant women who belonged to female-headed households (AOR=1.40; CI: 1.16–1.69) had a higher likelihood of losing their pregnancy relative to those who belonged to male-headed households. Ever pregnant women whose household heads were aged between 20 and 29 years were less likely (AOR=0.35; CI: 0.19–0.64) to lose their pregnancy compared with those whose household heads were aged 70 or more years. However, the odds of pregnancy loss experience were higher among household heads aged 40–49 years (AOR=2.80; CI: 1.59–4.93) and 50–59 years (AOR=2.52; CI: 1.38–4.61) than among household heads aged 70 years or older. In addition, ever-pregnant women whose households have access to improved drinking water sources (AOR=1.59; CI: 1.18–2.16) have higher odds of experiencing pregnancy loss than those accessing unimproved drinking water sources. Ever pregnant women who belong to the poorer household wealth quintile (AOR=1.71; CI: 1.27–2.30) have a higher probability of losing their pregnancy relative to those belonging to the richest wealth category. The third model (only proximate factors) shows that ever pregnant women who attended antenatal care four or more times during pregnancy (AOR=0.77; CI: 0.65–0.92) were less likely to lose their pregnancy compared with those who did not utilize antenatal care services.

Finally, the fourth model, which combined individual, proximate and household-level factors, showed that the age of ever pregnant women, their educational level, place of residence, religion, marital status, occupational type, household source of drinking water and number of antenatal visits were significant in predicting pregnancy loss experiences among ever pregnant women in Ghana. Ever pregnant women aged 25–34 years (AOR=3.81, CI: 2.88–5.03) and 35–49 years (AOR=10.50; CI: 7.25–15.21) had a higher likelihood of losing pregnancy than those aged 15–24 years. Ever pregnant women with secondary/higher education were less likely (AOR=0.71, CI: 0.52–0.97) to lose their pregnancy than those with no formal education. In addition, ever pregnant women who reside in urban areas (AOR=0.69; CI: 0.53–0.91) compared with those living in rural areas. Compared with ever pregnant women with no religion, those who are protestant (AOR=1.76; CI: 1.04–2.98) and those belonging to Pentecostal/Charismatic faith (AOR=1.79; CI: 1.12–2.87) had higher odds of losing their pregnancy. Ever pregnant women who were currently married (AOR=1.89; CI: 1.30–2.75) and those who were formerly married (AOR=3.06; CI: 2.05–4.55) were more likely to experience pregnancy loss than never married women. Regarding the occupation of ever pregnant women, those who were in professional/technical/managerial/clerical work (AOR=1.79; CI: 1.09–2.94), sales work (AOR=1.65; CI: 1.18–2.29) and services (AOR=2.82; CI: 1.43–5.53) were all more likely to lose their pregnancies compared with those who were not working.

Ever pregnant women who had access to an improved source of drinking water (AOR=1.52; CI: 1.08–2.13) had a higher probability of losing their pregnancy than those who had access to unimproved drinking water. Interestingly, ever pregnant women who attended antenatal care services 1–3 times (AOR=1.80; CI: 1.03–3.13) and 4 or more times (AOR=1.28; CI: 1.02–1.61) were more likely to experience pregnancy loss than those who did not attend antenatal care services.

Discussion

Using the 2014 GDHS data, the study examined the effect of individual, proximate, and household factors on pregnancy loss experiences among ever pregnant women in Ghana. The results show that pregnancy loss experience among ever pregnant women is relatively high, with ~4 of 10 (48%) losing their pregnancy either through miscarriages, stillbirths, spontaneous or induced abortion compared with other studies conducted in Manitoba (11.3%)23, the United Kingdom (10.8%)24, and the United States (13.5%)25.

The study showed that ever pregnant women aged 25–34 and 35–49 years had higher odds of losing their pregnancies than those aged 15–24 years before and after controlling for proximate and household factors. This means that women in the later reproductive age group (30–49 y) have experienced a relatively higher loss of pregnancy than women in their early reproductive age group (15–24 y). The reason might be that women in their later reproductive ages have experienced more conception in their entire reproductive years compared with women who are now entering their reproductive years. Previous studies found that at the age of 42 years, more than half of such pregnancies resulted in fetal loss due to spontaneous abortion, ectopic pregnancy, placenta previa, pregestational diabetes, eclampsia, and pregnancy-induced hypertension26–30. For instance, a study by Andersen et al30 revealed that the risk of a spontaneous abortion was 8.9% in women aged 20–24 years and 74.7% in those aged 45 years or more. Again, high maternal age was a significant risk factor for spontaneous abortion irrespective of the number of previous miscarriages, parity, or calendar period. The risk of ectopic pregnancy and stillbirth also increased with increasing maternal age. However, other studies found contrary results, where pregnancy loss was predominant among women whose age was younger than 29 years31,32.

This study also found that ever pregnant women with secondary or higher education had lower odds of losing their pregnancies compared with those who had no formal education with and without controlling for the effect of household and proximate factors. This finding confirms that of earlier studies21,33–36. These studies found that likelihood that a woman will report having had any form of pregnancy loss rises steadily as educational attainment increases, with a particularly sharp increase from women who have lower-level to those who have upper-level secondary schooling, followed by a decrease for university educated women.

The study also showed that ever pregnant women residing in urban areas had lower odds of losing their pregnancies relative to those in rural areas before and after controlling for the effect of other household and proximate factors. Similar findings were observed in a study in Uganda where the prevalence of stillbirths among rural women was very high compared with their urban counterparts37. Another study found that women in rural agricultural areas are at increased risk for adverse spontaneous miscarriage38.

Another important finding in this study is that ever pregnant women who are protestants and belong to the Pentecostal/Charismatic faith have a higher probability of experiencing pregnancy loss before and after the inclusion of proximate and household factors in the model. The plausible explanation for this is that during pregnancy, highly religious women intensify their prayers to God for protection, safe delivery and blessings for fear of losing their pregnancies39–41. Religious beliefs and practices, irrespective of the type, enable women who have experienced a pregnancy loss not only to cope better with the devastating effects of the loss but also to adapt to their loss42.

The result further shows the higher probability of pregnancy loss experiences among currently married and formerly married women compared with the never married women with and without the inclusion of other variables in the model. The possible logical reason for this occurrence was that it is assumed that since most conceptions and births occur in marriage, women are more at risk of losing their pregnancies because of the frequency of conception and birth. In other words, there are shorter pregnancy and birth intervals in marriage than in marriages. A study in India indicated that the likelihood of aborting increases monotonically among unmarried women43. However, other studies have shown that abortion is more likely among unmarried than married women44. They argued that although differences between married and unmarried women’s propensity to abort are affected by the greater likelihood of unwanted pregnancies among unmarried than married women, differences in the costs of carrying a pregnancy to term also influence abortion levels. For example, a birth to an unmarried woman is likely to be more disruptive, more socially stigmatizing, and more costly in terms of lost opportunities than a birth to a married woman.

Furthermore, ever pregnant women who are engaged in sales and services as a form of occupation were more likely to lose their pregnancy; however, after controlling for proximate and household factors, women who are engaged in professional/technical/managerial/clerical work together with sales and services workers had a higher likelihood of losing their pregnancy. This clearly shows the effect of other variables on the occupation of women and their pregnancy outcomes. The type of occupation women engage in sometimes influences their chances of experiencing pregnancy loss. Women who work in certain professional fields lead to conflict between their childbearing and career roles; as a result, they are more likely to make attempts to prevent unintended pregnancies45. However, once an unintended pregnancy occurs, it may result in pregnancy loss46. This loss could be in the form of induced or spontaneous abortion as a result of the stressful and demanding nature of the job they do.

In terms of household factors, sex, age of household head, household source of drinking water and wealth index significantly predicted pregnancy loss experiences among ever pregnant women. However, after including individual and proximate factors in the model, only household sources of drinking water predicted pregnancy loss experiences. The results show high possibilities of pregnancy loss experiences among women who dwell in female-headed households, relatively older household heads (40–49 and 50–59 y), households with access to improved drinking water sources and poorer households.

Proximate determinants, such as the number of antenatal visits by ever pregnant women, strongly predicted pregnancy loss experiences among them. Before controlling for other factors, the likelihood of losing pregnancy was lower among women who attended antenatal care 4 or more times than among those who did not visit health facilities for antenatal care. However, after the inclusion of other factors in the model, interestingly, the probability of women losing their pregnancy was higher among those who attended antenatal care 1–3 times as well as 4 or more times. This shows the significant effect of other factors on antenatal care in determining pregnancy outcomes of pregnant women. This also explains that attending antenatal care is a necessary condition but may not be a sufficient one in safeguarding the health of the fetus.

Strengths and limitations of the study

One of the strengths of this study includes the use of large nationwide population data, which made the result generalizable to all women in their reproductive ages regarding their pregnancy loss experiences. However, because the study used secondary data, it could not account for other factors (cultural, economic) at the community and national levels that might have influenced pregnancy loss experiences among women in Ghana.

Conclusion

This study examined the factors influencing pregnancy loss experiences among ever pregnant women in Ghana. The results from the analyses enable us to understand the following: first, the proximate factors (number of antenatal visits) associated with pregnancy loss experience among ever pregnant women. Second, the individual factors (age of women, educational level, place of residence, religion, marital status, occupational type) predict pregnancy loss among pregnant women. Last, household factors (sex, age of household head, source of drinking water, household wealth quintile) influence pregnancy loss among ever pregnant women in Ghana. The results from this study on pregnancy loss would be useful to policy makers in the health sector once a clear distinction is made between experiences of induced or spontaneous abortion (miscarriage) and stillbirth and how these relate to subsequent pregnancies.

The study therefore recommends that there should be more reproductive health public education, especially for young women as they enter their reproductive ages, on safe practices during the pregnancy period. This recommendation is essential because the results show that older women who have gone through their reproductive ages have experienced more pregnancy loss than younger women. This therefore calls for more attention to be given to younger women to adopt safe practices and avoid risky behaviors that will prevent the loss of pregnancy among them through miscarriages, stillbirths, (spontaneous or induced) abortion. Again, health facilities must be closer to women, especially those living in rural areas, since the results also show that ever pregnant women in these areas were more likely to lose their pregnancies than those in urban areas.

Sources of funding

None.

Author contribution

D.K.: conceptualized and designed the study, obtained the data, and analyzed and interpreted the data. D.K. also drafted the entire manuscript and critically reviewed, revised, and approved the manuscript for publication.

Conflict of interest disclosure

The authors declare that they have no financial conflict of interest with regard to the content of this report.

Acknowledgments

The author would like to acknowledge the academic staff of the Institute of Health Research (IHR) and Ghana Statistical Service (GSS) for their invaluable contribution to this work.

References

1. Maconochie N, Doyle P, Prior S, et al. Risk factors for first trimester miscarriage—results from a UK‐population‐based case–control study. BJOG 2007;114:170–86.
2. Michels TC, Tiu AY. Second trimester pregnancy loss. Am Fam Physician 2007;76:1341–6.
3. Wilcox AJ, Weinberg CR, O’Connor JF, et al. Incidence of early loss of pregnancy. N Engl J Med 1988;319:189–94.
4. Stirrat GM. Recurrent miscarriage I: definition and epidemiology. Lancet 1990;336:673–5.
5. Lawn J, Shibuya K, Stein C. No cry at birth: global estimates of intrapartum stillbirths and intrapartum-related neonatal deaths. Bull World Health Organ 2005;83:409–17.
6. McClure EM, Nalubamba-Phiri M, Goldenberg RL. Stillbirth in developing countries. Int J Gynecol Obstet 2006;94:82–90.
7. Chigbu CO, Okezie OA, Odugu BU. Intrapartum stillbirth in a Nigerian tertiary hospital setting. Int J Gynecol Obstet 2009;104:18–21.
8. Engmann C, Matendo R, Kinoshita R, et al. Stillbirth and early neonatal mortality in rural Central Africa. Int J Gynecol Obstet 2009;105:112–7.
9. Cousens S, Blencowe H, Stanton C, et al. National, regional, and worldwide estimates of stillbirth rates in 2009 with trends since 1995: a systematic analysis. Lancet 2011;377:1319–30.
10. Rowe JF. Pregnancy loss among British Columbia mothers. Can J Public Health 1973;41:25–35.
11. Jones RK, Jerman J. Abortion incidence and service availability in the United States, 2011. Perspect Sex Reprod Health 2014;46:3–14.
12. Weinstein M, Wood JW, Stoto MA, et al. Components of age-specific fecundability. Popul Stud 1990;44:447–67.
13. Sundaram A, Juarez F, Bankole A, et al. Factors associated with abortion‐seeking and obtaining a safe abortion in Ghana. Stud Fam Plann 2012;43:273–86.
14. Lithur NO. Destigmatising abortion: expanding community awareness of abortion as a reproductive health issue in Ghana. Afr J Reprod Health 2004;31:70–4.
15. Mills S, Williams JE, Wak G, et al. Maternal mortality decline in the Kassena-Nankana district of northern Ghana. Matern Child Health J 2008;12:577–85.
16. Ahiadeke C. Incidence of induced abortion in Southern Ghana. Int Fam Plann Perspect 2001;43:96–108.
17. Schwandt HM, Creanga AA, Adanu RM, et al. Pathways to unsafe abortion in Ghana: the role of male partners, women and health care providers. Contraception 2013;88:509–17.
18. Ghana Statistical Service. Ghana maternal health survey. Ghana: Ghana Statistical Service; 2007.
19. Wiredu EK, Armah HB. Cancer mortality patterns in Ghana: a 10-year review of autopsies and hospital mortality. BMC Public Health 2006;6:1–7.
20. Angell JN, Abdul‐Mumin AR, Gold KJ. Determining the cause of stillbirth in Kumasi, Ghana. Int J Gynecol Obstet 2019;147:173–8.
21. Afulani PA. Determinants of stillbirths in Ghana: does quality of antenatal care matter? BMC Pregnancy Childbirth 2016;16:1–7.
22. Bampoe AJ. Risk factors for breast cancer among patients in Korle-Bu Teaching Hospital a case-control (doctoral dissertation, University of Ghana).
23. Strumpf E, Lang A, Austin N, et al. Prevalence and clinical, social, and health care predictors of miscarriage. BMC Pregnancy Childbirth 2021;21:1–9.
24. Quenby S, Gallos ID, Dhillon-Smith RK, et al. Miscarriage matters: the epidemiological, physical, psychological, and economic costs of early pregnancy loss. Lancet 2021;397:1658–67.
25. Rossen LM, Ahrens KA, Branum AM. Trends in risk of pregnancy loss among US women, 1990–2011. Paediatr Perinat Epidemiol 2018;32:19–29.
26. Cleary-Goldman J, Berkowitz RL. First trimester screening for Down syndrome in multiple pregnancy. Semin Perinatol 2005;29:395–400.
27. Hollier LM, Leveno KJ, Kelly MA, et al. Maternal age and malformations in singleton births. Obstet Gynecol 2000;96:701–6.
28. Jacobsson B, Ladfors L, Milsom I. Advanced maternal age and adverse perinatal outcome. Obstet Gynecol 2004;104:727–33.
29. Jolly M, Sebire N, Harris J, et al. The risks associated with pregnancy in women aged 35 years or older. Hum Reprod 2000;15:2433–7.
30. Andersen AM, Wohlfahrt J, Christens P, et al. Maternal age and fetal loss: population based register linkage study. BMJ 2000;320:1708–12.
31. Yassin KM. Incidence and socioeconomic determinants of abortion in rural upper Egypt. Public Health 2000;114:269–72.
32. Misago C, Fonseca W, Correia L, et al. Determinants of abortion among women admitted to hospitals in Fortaleza, North Eastern Brazil. Int J Epidemiol 1998;27:833–9.
33. Vangen S, Eskild A, Forsen L. Termination of pregnancy according to immigration status: a population‐based registry linkage study. BJOG 2008;115:1309–15.
34. Toulemon L. Standard population and other standardization methods: application to the measurement of recourse to abortion according to the PCS. Population 1992;47:192–204.
35. Toulemon L, Leridon H. Twenty years of contraception in France: 1968-88. Population an English Selection. 1–33 1992.
36. Shapiro D, Tambashe BO. The impact of women’s employment and education on contraceptive use and abortion in Kinshasa, Zaire. Stud Fam Plann 1994;42:96–110.
37. Kiguli J, Namusoko S, Kerber K, et al. Weeping in silence: community experiences of stillbirths in rural eastern Uganda. Global Health Action 2015;8:24011.
38. Naidoo S, London L, Burdorf A, et al. Spontaneous miscarriages and infant deaths among female farmers in rural South Africa. Scand J Work Environ Health 2011;60:227–36.
39. Jesse DE, Schoneboom C, Blanchard A. The effect of faith or spirituality in pregnancy: a content analysis. J Holist Nurs 2007;25:151–8.
40. Aziato L, Odai PN, Omenyo CN. Religious beliefs and practices in pregnancy and labour: an inductive qualitative study among post-partum women in Ghana. BMC Pregnancy Childbirth 2016;16:1–0.
41. Allahdadian M, Irajpour A. The role of religious beliefs in pregnancy loss. J Educ Health Promot 2015;4:40–54.
42. Kalu FA. Women’s experiences of utilizing religious and spiritual beliefs as coping resources after miscarriage. Religions 2019;10:185.
43. Mavalankar DV, Trivedi CR, Gray RH. Levels and risk factors for perinatal mortality in Ahmedabad, India. Bull World Health Organ 1991;69:435.
44. Henshaw SK, O’Reilly K. Characteristics of abortion patients in the United States, 1979 and 1980. Fam Plann Perspect 1983;15:5–8.
45. Williams CM, Brett KM, Abma JC. Coercive first intercourse and unintended first births. Violence Victims 2009;24:351–63.
46. Powell-Griner E, Trent K. Sociodemographic determinants of abortion in the United States. Demography 1987:553–61.
Keywords:

Pregnancy loss; Pregnant women; Experiences; Ghana

Copyright © 2022 The Authors. Published by Wolters Kluwer on behalf of the International Federation of Fertility Societies.