The Uganda safe mother hood initiative was introduced in the early 1990s with the aim of halving maternal mortality by the year 2000. This objective was not achieved since maternal mortality in Uganda has remained high. The most frequent cause of death was post- partum haemorrhage. This is similar to the finding in other developing countries [8, 10, 11, 18]. Majority of fatal cases (85%) and non fatal (90%) were below the age of thirty years with a similar mean age. Severe maternal morbidity and maternal deaths afflict young women and this is similar to other reports from Sub Saharan Africa [10, 12].
The further away the woman lived from Mulago hospital, the higher the risk of death with those living more than ten kilometres having a risk of three times to die. This is because they developed severe maternal morbidity conditions and when referred but arrived in a very critical condition and sometimes it was difficult to salvage them. Other studies have also reported that the further the distance the mother stayed from hospital the higher the risk of death [18, 20]. A distance of 10 to 15 kilometres was not too far from a referral hospital, but due to lack of an effective referral system, women developed complications in a health unit and when referred, arrived late in hospital due to lack of ambulance system.
The delay in reaching the hospital has been attributed to women seeking permission from their spouses or their in-laws before they could access the health facility.
The women who sought permission before attending a health unit were associated with three times the risk of dying compared to those who didn't. Similar findings were reported from the study in Tanzania . It is likely that seeking for permission before accessing a health facility at the time complications occurred may be a manifestation of lack of empowerment of the women as they could no make their own decisions.
Women who had some form of employment were associated with less risk of dying compared to the unemployed. The employed women had better education, were more likely to attend antenatal clinic and may be empowered and therefore carried less risks for maternal mortality. Low educational status, poverty and lack of empowerment have been reported to be associated with maternal mortality, [8, 12, 20–24].
The women who were admitted to hospital during the present pregnancy for any medical conditions were four times more likely to die compared to those who were not admitted. These women could have had medical conditions like malaria, hypertension or anaemia which predispose to mortality.
Birth interval of greater or equal to thirty six months was associated with a reduced risk of dying relative to those whose birth interval was thirty six or less months. The reasons for this may be that prolonged birth interval promotes recovery of a woman from a previous pregnancy. This confirms that spacing most likely due to use of family planning was associated with reduction of maternal deaths.
Non-attendance of antenatal care was found to be associated with an increased risk of maternal death compared to those who had attended antenatal care. Antenatal care is essential to screen women for risk factors for obstetric complications and those that do not benefit from this have an increased risk of maternal death. For antenatal care to be effective the patients should book in the first trimester and attend regularly as prescribed by the health provider.
When women were asked what they would do when they noticed vaginal bleeding during pregnancy, those who said that they would go to hospital were more likely to survive. Those who did not know what to do and where to go were associated with four times risk of dying compared to those who knew what to do. Bleeding during pregnancy is called ante- partum haemorrhage which needs to be evaluated and managed because of the un predictable outcome. Having complication during pregnancy has been reported as a predictor of maternal death .
The women who delivered male babies were associated with four times risk of dying compared to those who delivered female babies. It could be that male babies were bigger were associated with obstructed labour or ruptured uterus. In addition male babies were associated with pre- eclampsia and abruptio placenta and these complications predisposed the mothers to death [8–10, 12, 20].
Patients who did not get oxytocics soon after delivery were associated with four times risk of progressing to death compared to those who had it. The reason for this is that some patients delivered outside Mulago hospital and did not get oxytocics soon after delivery and presented with severe post partum haemorrhage. The other possible reason is that because of poor record keeping by staff and recall bias of patients could have caused some bias and this could not be a true reflection of the practice. However, the management of the third stage of labour if not accompanied by use of oxytocics is associated with increased chance of developing post- partum haemorrhage which is a major cause of maternal mortality.
The patients who needed blood but blood was not available for transfusion were associated with an increased risk of fifty four times of progression to maternal death compared to those who didn't require blood transfusion. Shortage of blood has been reported as one of the factors that have increased the maternal mortality in Mulago hospital . The shortage has been made worse by the HIV prevalence in Uganda as blood donated have a big proportion of high HIV positivity. Shortage of blood has also been reported to increase maternal deaths in Nakuru provisional general hospital in Kenya . Review of other studies show shortage of blood as a major cause of maternal deaths .
Puerperal sepsis contributed to a bout 6% of the maternal deaths in this study. This is lower than that reported from Nigeria . Puerperal sepsis can be reduced by having skilled delivery attendants and by carrying out delivery in a clean and aseptic environment. Patients with puerperal sepsis who needed specific antibiotics for their condition and were not available were associated with four times the risk of progressing to maternal mortality compared to those who received antibiotics in the hospital. This accounts for high proportion of deaths among mothers with puerperal sepsis. The shortages increased the patients delay in access of the right treatment which also predisposed them to maternal death. The most affected group of patients were those of the low income group who couldn't afford the cost of treatment.
The study found that HIV positive women were associate d with five times the risk of dying compared to those who were HIV negative. These results are similar to what has been reported in Uganda on the contribution of HIV to maternal mortality with odds ratio of 3.2  and 5.4 . This is similar to the results of Meta analysis of studies from developing countries . However, this is in contrast to the study from Ireland which reported no effect of HIV on severe maternal morbidity cases that progressed to maternal death . It is possible that patients who are HIV positive in Ireland are on antiretroviral drugs and their immunity is high and are not likely to experience severe maternal morbidity and mortality.
Limitation and strength of study
The number of cases that died were few and this limited the study in some degree. However the predictors identified and recommendation made can be generalised in all tertiary hospitals but was not substantial enough to make a policy reform in prevention of maternal mortality.