Factors That Predict Demand For Perinatal Care Among Egyptian Women: A Multivariate Logistic Regression Analysis!
K Yassin, G Saida
birth attendance, egypt, institutional delivery, maternal mortality, perinatal services
K Yassin, G Saida. Factors That Predict Demand For Perinatal Care Among Egyptian Women: A Multivariate Logistic Regression Analysis!. The Internet Journal of Public Health. 2012 Volume 2 Number 1.
Maternal mortality continues to be a serious public health problem in Egypt. Data from community based surveys indicated that though maternal mortality had declined from 170 per 100,000 live birth in 1993 to 84 in 2000, underreporting, misclassification and sampling effects might have underestimated the true rate and a national MMR was calculated to be 94/100,000 live births.1 However, a recent joint report of the World Health Organization, UNICEF, United Nation Population Fund and World Bank estimated that the true MMR might be as high as 130.2 These rates are worse than expected for low middle income countries and represent a serious challenge for Egypt towards achieving millennium development goals.
The vast majority of maternal deaths occur during or immediately after childbirth, largely due to bleeding, dystocia, sepsis, eclampsia, ruptured uterus, amniotic fluid embolism and complications of underlying medical conditions.1,3–6 Many of these deaths could be avoided, if childbirth takes place in medical institutions or attended by a skilled health care worker who can appropriately manage complications of labor and make timely referral. There is ample evidence from literature that women attended by trained medical professionals are less likely to die or develop serious complications during childbirth.7-10
In Egypt there exists a widespread network of primary health care facilities providing perinatal medical care. The overwhelming majority of the population lives within five kilometres distance from one of these facilities. Furthermore, an extensive network of public and private hospitals providing emergency and specialized obstetric care covers all districts in the country. As a result, supply of perinatal care in Egypt is high compared with other developing countries with similar national income .11
The paradox of relatively high maternal mortality rates and high availability of perinatal medical services could be attributed either to low levels of utilization of these services by pregnant women or to substandard quality of care of these services.
This study aimed at examining levels of utilization of perinatal care and identifying socioeconomic determinants of such utilization. Results could help devising specific policies for improving utilization of these services that hopefully could accelerate pace of decline in maternal mortality rate in Egypt. The study placed a special emphasis on maternal choice of place of delivery, attendance of delivery by a trained health care worker and postnatal care.
Data for this study was driven from the Egypt Demographic and Health Survey in 2000. The survey was undertaken by the Ministry of Health and Population to provide national level information on fertility, family planning and child and maternal health. In the survey a national representative sample of 8999 ever married woman (15 – 49 years old) were taken from 26 governorates in Egypt. Ever married women include women who have been married at least once, irrespective of their current marital status.
The sample was obtained from randomly chosen 228 towns and 272 villages. Each selected town or village was divided into equal parts of approximately 5,000 population and one part was selected randomly for inclusion in the sample. The selected parts were then divided into standard segments of about 200 households and two segments were selected randomly from each part. Within each selected segment a systematic random sample was obtained after listing of all households of the segment.
In the survey women were asked to provide information about pregnancy, child birth and postnatal period of all pregnancies that took place in the five-year period prior to the survey. Further, data were collected about the demographic and socioeconomic background factors.
We included in the analysis only the last birth of each woman. This restriction was made to adjust for the hierarchical nature of data i.e. data about several births for the same woman. If this hierarchical nature was ignored, the results might have been biased since many of the statistical modeling techniques assume independence of observations. Another reason for this restriction is to shorten the recall period of mothers and ensure quality of maternal reporting.
Three dependent variables were selected to reflect quality of perinatal care; 1) institutional delivery; 2) attendance of childbirth by qualified medical staff and 3) having a postnatal check up within 24 hours from delivery. Each of these dependent variables was redefined as a dichotomous variable, coded 1 for giving birth in a medical institution, attended by qualified medical staff and received postnatal checkup within 24 hours after delivery, respectively and 0 for otherwise.
The independent variables included 21 factors, covering maternal, paternal and socioeconomic characteristics as well as mother's position in the family. These variables included maternal age at survey, maternal age at marriage, maternal age at first birth, maternal education, maternal occupation, fertility, whether the mother has ever had a terminated pregnancy or under 5 death, utilization of antenatal care, current use of modern contraception, desire to have more children, preceding birth interval, paternal age, paternal education, residence, type of toilet facility, whether they have a refrigerator and whether the mother has a say on own health care, on large household purchases and on visits to family or relatives. Near birth problems were included as a variable.
Before applying the regression model, we had to examine data for multicollinearity, which is the statistical phenomenon of having two or more highly correlated independent variables (predictors) in a multiple regression model. Multicollinearity is known to weaken the reliability and predictive power of regression models. Therefore, all independent variables were tested for multicollinearity and when the correlation coefficient was above 0.65 for any two variables, one of the variables was excluded. For example, maternal age at marriage was correlated with maternal age at first birth. Likewise, fertility was correlated with pregnancy order. A forward conditional stepwise regression method was adopted and the predicted variables with highest R² (at the 0.05 level of significance) were retained. These were maternal age at first birth and fertility.
A logistic regression model was used then to ascertain the association between each dependent variable and independent variables. If P is the probability of occurrence of dependent variable, then
where β is the vector of the unknown coefficient to be estimated and X is a vector of independent variables that influence the dependent variable. The general logistic regression model can be further stated as:
which indicates the log odds of the dependent variable as a linear function of independent variables.
In the analysis a total of 7916 women from five regions in Egypt were included; 1379 from urban governorates, 3003 from Lower Egypt, 3065 from Upper Egypt and 541 from Frontier governorates.
Table 1 shows that 51% of women delivered in health institutions and 49% at home. With respect to spatial distribution, the rate of institutional delivery was highest in the urban governorates (80%), followed by urban Lower Egypt (74%) and urban Upper Egypt (62%). The frontier governorates showed a slightly lower rate than rural Lower Egypt. The corresponding rates of institutional delivery in these two regions were 43% and 47%, respectively. Rural Upper Egypt, which is the poorest and most deprived part of Egypt, showed the lowest rate of institutional delivery (27%).
National and regional rates of birth attendance were slightly higher than institutional delivery. The aggregate national rate of birth attendance was 64%. The spatial distribution was similar to that of institutional delivery, with the highest rate reported in Urban Governorates (87%) and the lowest in rural Upper Egypt (40%).
With respect to postnatal care, the results were alarming in all regions. One in every five women received a postnatal checkup within the first 24 hours after delivery. Paradoxically, urban Upper Egypt had the highest rate of postnatal checkup (30% of women) and rural Upper Egypt the lowest rate (15%). The rates in other regions were 23% in urban governorates, 26% in urban Lower Egypt, 17% in rural Lower Egypt and 17% in frontier governorates.
Table 2 shows the results of the multivariate logistic regression analysis of determinants of institutional delivery in Egypt. The strongest determinant for institutional delivery was shown to be the utilization of antenatal care. Women who utilized antenatal care during pregnancy had higher probability of institutional delivery than women who did not utilize antenatal care (OR = 3.2 and P-value <0.01). The difference was significant and independent of the effect of other variables.
Place of residence was the second strongest determinant of institutional delivery. Women residing in urban areas were more likely to deliver in health institutions than women in rural areas (OR = 2.4 and P-value <0.01). Likewise, pregnancy order appeared to be a significant predictor of institutional delivery, with first and second order pregnancies were more likely to deliver in health institutions than higher order pregnancies (OR = 2.4 and P-value <0.01). This association between pregnancy order and institutional delivery was shown to be independent of woman’s age. The odds ratio of Institutional delivery among women younger than 30 was 0.8 (P-value <0.05) compared with those who were 30 or older.
Women who reported near birth problems, namely difficult or prolonged labor, severe vaginal bleeding or convulsions were more likely to deliver in a health facility than other women (OR = 1.6 and P-value <0.01). As expected, maternal illiteracy was associated with higher rates of institutional delivery, with the probability of institutional delivery among literate women was significantly higher than illiterate women (OR = 1.5 and P-value <0.01).
Other variables which were significantly associated with institutional delivery included maternal age at first birth (OR = 1.4, P < 0.01), current use of modern contraception (OR = 1.3, P < 0.01), woman has a say on large household purchases (OR = 1.3, P < 0.01) and having a refrigerator (OR = 1.3, P < 0.01).
Table 3 shows the results of the logistic regression analysis of the determinants of birth attendance in Egypt. The table shows that the same variables which were significantly associated with institutional delivery appeared to be significantly associated with birth attendance. However, the magnitude of association differed to some extent.
For example, women who sought antenatal care during pregnancy were more likely to be attended at birth by a trained health worker than women who did not seek antenatal care (OR = 2.9 and P-value < 0.01). Urban residence appeared to be the second strongest determinant of birth attendance, associated with higher rates of birth attendance (OR = 2.7 and P-value < 0.01). Likewise, birth attendance was significantly higher for first and second pregnancies than higher order pregnancies (OR = 2.1 and P-value < 0.01).
Literate women were shown to be more likely to be assisted at delivery than illiterate women (OR = 1.8, P < 0.01). Further, birth attendance was shown to be higher among women whose age at first birth was 20 or higher (OR = 1.4, P < 0.01), among current users of modern contraception (OR = 1.3, P< 0.01), among women who reported having a near birth problem (OR = 1.4, P< 0.001) and among women having a refrigerator (OR = 1.4, P< 0.001).
All the indicators for woman’s position, a woman having a say on her own health care, on large household purchases or on visits to relatives, were not significantly associated with birth attendance.
In addition, four variables, which were not significantly associated with institutional delivery, appeared to be significantly associated with birth attendance. These were maternal occupational status, preceding birth interval and paternal education and type of toilet facility. Working women (OR = 1.3, P< 0.05), women of literate husbands (OR = 1.4, P< 0.01), a preceding birth interval of 24 months or more (OR = 1.3, P< 0.05) and having a flush toilet (OR = 1.3, P < 0.05) were all significantly associated with birth attendance.
Table 4 shows the results of the logistic regression analysis of determinants of postnatal care in Egypt. Only five variables were shown to be significantly associated with postnatal care. In terms of magnitude of association, these variables were utilization of antenatal care (OR = 2.1, P < 0.01), near birth problems (OR = 2.0, P < 0.01), maternal literacy (OR = 1.3, P < 0.5), maternal occupation (OR = 1.3, P < 0.5), and use of modern contraception (OR = 1.2, P < 0.05). All other variables including place of residence did not show a statistically significant association with postnatal care.
The results of this study could be summarized in three basic findings. First, perinatal services though available in Egypt are not fully exploited by pregnant women. Second, women from poorer areas and with disadvantaged socioeconomic background make less use of perinatal services. Third, multiple socioeconomic factors seems to determine utilization patterns of perinatal services in Egypt.
Though maternal and child birth services enjoy remarkable availability and geographic coverage with the vast majority of pregnant women residing within 3 – 5 kilometres from a health facility, a considerable proportion of these women are not utilizing these facilities for childbirth and postnatal care. This study indicated that one-half of pregnant women in Egypt gave birth at home, the majority of them without supervision by qualified medical staff. Further, postnatal care was shown to be alarmingly very low in all regions and warrants, therefore, a meticulous investigation.
Because of this gross underutilization of available maternal health services, there exist large unrealized health gains in Egypt. Improving utilization patterns by pregnant women would result in substantial decline of avoidable maternal deaths.
The finding that utilization of perinatal care is least among women in poorer and less developed areas or women with disadvantaged socioeconomic background is troublesome. This study showed that in rural Upper Egypt, where maternal mortality is highest in the country, only 27% of women gave birth in a health institution, 40% of births were attended by qualified medical staff and only 15% received postnatal care. The corresponding figures for rural Lower Egypt are unacceptably low as well. Likewise, illiterate women who constitute two thirds of women in the child-bearing period showed very low utilization rates estimated at 33% for institutional delivery and 13% for postnatal care.
These figures are alarming not only because it reflects geographic and social disparities that are unethical, but because it jeopardizes the national effort to reduce overall maternal mortality as well.
These results are congruent with many other studies from developing countries which reported socioeconomic disparities in utilization of perinatal health services.12-18 For example, a review of utilization of maternal services in 55 developing countries indicated that on average women in the richest quintile are 5.2 times more likely to give birth under the supervision of a doctor, nurse, or midwife than the poorest fifth of women.19
Multiple factors are responsible for underutilization of maternal services. On the demand side, cultural and educational factors may obscure the potential benefits from utilizing these services, while economic constraints may suppress utilization, even if benefits are recognized.
On the supply side, several studies indicated that though maternal services are physically available in Egypt, their quality is often severely deficient, leaving its effectiveness well short of potential efficacy. Poor quality of services would over time affect negatively demand for these services through undermining perception of potential benefits on the part of pregnant women.20
On the demand side, there is ample evidence that a strong direct relationship between socioeconomic indicators, particularly income, and utilization of maternal services. This relation was shown to be independent and stable after controlling for other determinants of health care demand. 21-27
This association is not surprising in settings where perinatal services are burdensome with direct and indirect costs even in public institutions which usually deliver lower quality of care. Several studies indicated that in developing countries where there is heavy reliance on out-of-pocket payment for maternal services, the poor and disadvantaged women have less opportunity to benefit maternal health services.28-38
In addition to direct health care cost, indirect costs such as travel costs and foregone earnings were shown to be a barrier to utilization of health care by the poor, especially in rural areas, where time, effort, and cost required to arrive at hospital can be substantial.39-42
The socio-cultural context is another important determinant of utilization of maternal care. A qualitative anthropological study in indicated that Egyptian women generally view giving birth as a natural process and therefore many women were not inclined to go to a hospital child birth. Adherence to these norms is influenced by the socioeconomic environment whereby gender attitudes and roles play a decisive role in determining care seeking behaviour.43
This study has shown that all three women position indicators were significantly and directly associated with utilization of perinatal care, with less empowered women are less likely to benefit from perinatal services. This finding is in line with other studies in developing countries which showed that women autonomy, independence and empowerment foster utilization of maternal care services.44-51
Further, a study from Egypt examined the relation between women's position and fertility control and concluded that the reproductive aspect of women's position had a strong connection with the non-reproductive dimensions in determining contraceptive use.52
Maternal education was shown to be a strong and independent determinant of utilization of perinatal services in Egypt. The finding is congruent with many studies from developing countries which reported constant direct relation between level of education and utilization of health care in general and maternal services specifically.14,53-58
The pathways between maternal education and improved care seeking patterns have been suggested either to be linked to better recognition of illness and perception of potential benefits of treatment or to change in the traditional balance of family relationships that shifts the focus of power away from the kin group and allows educated mothers assume greater responsibility of their own health.
The study points out the need to supplement the current effort of the Ministry of Health to improve the supply and quality of maternal health services with a simultaneous effort to improve demand of pregnant women for institutional delivery and birth attendance. If current utilization patterns continue, maternal mortality and morbidity will continue to be unacceptably high.
If demand is to be effectively created or enhanced, strategies must be consonant with the perspectives of the community. Therefore, careful qualitative research will be needed as a basis for designing messages to convey knowledge and modify behavior. Such messages need to be conveyed to the significant others in the family such as the husband, the mother and the mother in law. Mass and interpersonal media should be used to improve knowledge and practices related to pregnancy and child birth.
Furthermore, an ongoing dialogue with pregnant women should be undertaken to involve them in planning and to address their concerns about pregnancy, child birth and maternal health care. In addition, men should be encouraged to share responsibility for making informed decisions about place of delivery and birth attendance. Efforts in these and related areas must be undertaken locally. Local basic beliefs, customs, and social circumstances must be identified so that communications efforts take them into account.
A final important point at the supply side is the coverage and quality of postnatal care. A concerted effort is needed from different actor to educate the maternal care providers in the public and private settings as well as the pregnant women in the community about the importance of postnatal care. The introduction of practical guidelines fro physician may be helping and encouraging for physicians.