The relationship between attendance at birth and maternal mortality rates:an exploration of United Nations' data sets including the ratios of physicians and nurses to population, GNP per capita and female literacy

Background. This is the third and final paper drawing on data taken from Un ited Nations (UN) data sets. The first paper examined the global distributi on of health professionals (as measured by ratios of physicians and nurses to population), and its relationship to gross national product per capita ( GNP) (Wharrad & Robinson 1999). The second paper explored the relationships between the global distribution of physicians and nurses, GNP, female lite racy and the health outcome indicators of infant and under five mortality r ates (IMR and u5MR) (Robinson & Wharrad 2000). In the present paper, the gl obal distribution of health professionals is explored in relation to matern al mortality rates (MMRs). The proportion of births attended by medical and nonmedical staff defined as 'attendance at birth by trained personnel' (ph ysicians, nurses, midwives or primary health care workers trained in midwif ery skills), is included as an additional independent variable in the regre ssion analyses, together with the ratio of physicians and nurses to populat ion, female literacy and GNP. Aim. To extend our earlier analyses by considering the relationships betwee n the global distribution of health professionals (ratios of physicians and nurses to population, and the proportion of births attended by trained hea lth personnel), GNP, female literacy and MMR. Design. Using a database on 155 countries, regression analyses were perform ed using numbers of physicians, and numbers of nurses, per 1000 population, the proportion of births attended by trained health personnel, GNP per cap ita and female literacy as independent variables and MMRs as the dependent variable. Results. Linear regression analyses show positive associations for MMRs and the ratios of physicians to population (73%, n=136), ratios of nurses to p opulation (56%, n=137), and the proportion of births attended by trained he alth personnel (83%, n=118). Multiple regression analyses reveal a more com plex picture, with nurses disappearing altogether when regressed with physi cians, GNP, female literacy and MMR. The three variables, attendance at bir th by trained personnel, GNP and physicians per 1000 population explained 8 7% of the variation in MMR (n=112) when included in the multiple regression analysis. Conclusions. As in the previous papers, caution is required regarding the v alidity and reliability of the UN data sources used in these analyses. Mate rnal mortality rates are particularly susceptible to inaccuracies. Neverthe less, the strength of the positive correlations suggests that real relation ships are identified between the independent variables and the dependent va riable of MMR. The strength of the linear and multiple correlations between births attended by trained personnel and lower MMRs indicates that materna l deaths are substantially reduced when a high proportion of births are att ended by health professionals, including primary health care workers traine d in midwifery skills, with the maintenance of an aseptic environment, the identification of maternal and foetal complications, and the opportunity wh en necessary to transfer parturient mothers to centres with higher level sk ills and facilities.