demographic and socio-economic determinants of maternal mortality -a case study of Pumwani Maternity hospital, Nairobi

Abstract

The main objective of the study was to investigate the determinants of maternal mortality among females of the reproductive ages, and to find their significant relationships. The study attempted to investigate seven selected variables, mainly demographic and social-economic factors and to find out how these operate to determine maternal mortality. Having reviewed the various studies done elsewhere, it was found to be of particular importance to study maternal mortality in one of the oldest maternity hospitals in Kenya. The assumption was that the hospital receives the largest number of patients and so it was expected that adequate data would be collected. The research was institutional based where hospital data from the ill-patient files was collected. The period covered was 5 years (1990-1994). A sample size of 108 maternal deaths and 1124 survivors was used. Method of data collection was basically the primary source where the in-patient files, statistical registers, disease index card, annual reports were all used and the necessary information extracted. Supplemental data was obtained from other relevant sources ie. 1979 and 1989 census reports. Methods of data analysis included descriptive statistics, which mainly made extensive use of ratios and percentages to demonstrate the relationship between maternal mortality and the selected demographic variables. Cross -tabulations, the chi¬ square, and the logistic regression techniques were all used for analysis. The results obtained. varied depending on the independent variable being studied. , The independent variables were further analyzed using the chi-square technique. Most of the null hypotheses had no significant relationship to the dependent variable thus were rejected. In the logistic regression, most of the independent dummy variables showed significant relationship. However, dummy variables such as AGE4 (30-34), AGES (35+) and 813 (24 months and above) were significantly related to the dependent variable. The best equation explaining the dependent variable was shown in the logistic regression model with only the significant variables. In this equation, the standard error was low (0.1162) and the constant of beta was greater than 0.5 (ie. 2.3184) from our findings the classical U-shaped relation between maternal death and maternal age was observed although it did not conform to findings of other ~studies. Further a curvilinear relationship was observed bel0eell parity and maternal death. The study recommends that maternal mortality analysis and interpretation from health facilities would be more useful if only these information sources were improved. The study recommends that nulliparous and primgravida women be encouraged to delivery in hospitals given their high maternal risks. Proper data collection, recording and classification of maternal motility should be developed.

...._

extracted. Supplemently data was obtained from other relevant sources ie. 1979 and 1989 census reports. Methods of data analysis included descriptive statistics, which mainly made extensive use of ratios and percentages to demonstrate the relationship between maternal mortality and the selected demographic variables. Crosy~tabulations, the chi­ square, and the logistic regression techniques were all used for analysis. The results obtained. varied depending on the independent variable beinq studied.

, The independent variables were further analyzed using the chi-square technique. Most of the null hypotheses had no significant relationship to the dependent variable thus were rejected. In the logistic regression, most of the independent dummy variables showed significant relationship. However, dummy variables such as AGE4 (30-34), AGES (35+) and 813 (24 months and

above) were significantly related to the dependent variable. The best equation explaining the dependent variable was shown in the logistic regression model with only the significant variables. In this equation, the standard error was low (0.1162) and the

constant of beta was greater than 0.5 (ie. 2.3184)

from our findings the classical U-shaped relation between maternal death and maternal age was observed although it did not conform to findings of other ~studies. Further a curvilinear relationship was observed bel0eell parity and maternal death. The study recommends that maternal mortality analysis and interpretation from health facilities would be more useful if only these information sources were improved. The study recommends that nulliparous and primgravida women be encouraged to delivery in hospitals given their high maternal risks. Proper data collection, recording and classification of maternal motility should be developed.