There were 663 maternal deaths recorded during the study period-January 2007 to December 2014. During the same period the total number of deliveries were 44,307 however, 38,896 were live births. The estimated maternal mortality ratio (MMR) in each year varied with a range between 1461 and 2105 per 100,000 live births. A study conducted in the same hospital in the 1990s had MMR estimated at 736 per 100,000 live births . In our study the lowest MMR was 1461 per 100,000 live births (2009) which signify a twofold increase from the previous study . Also, the study revealed, approximately a 45% rise of MMR between 2009 and 2011. Similar rise occurred between 2012 and 2013. The MMR between 2007–2010 and 2011–2014 was 1555 and 1887 per 100,000 live births respectively, which showed about 20% increase. Therefore, the trend overall of MMR at EFSTH is on the increase despite paradoxical decrease in the national MMR . The proportion of women who have antenatal care and the proportion of deliveries attended by skilled birth attendants have increased over the years in the Gambia [8,9,10]. EFSTH as the only tertiary health centre is thus getting more cases that would hitherto not have presented to any health facility. This can be deduced from the fact that the proportion of maternal deaths that were referred from other health facilities has been steadily increasing (Fig. 3). The other contributory factor to this trend is delayed referral of cases. Our study showed that the proportion of maternal deaths that occurred within 24 h of admission at EFSTH had increased from 40% in 2007–2010 to over 52% between 2011–2014 (Table 1). Increasing referrals from other health facilities and late presentation may explain the rising trend in MMR in EFSTH. However, similar studies from other tertiary hospitals also showed fluctuating trend in MMR [11,12,13,14].
The impact of demographic characteristics of maternal deaths in this 8-year review also had remarkable findings. A study by Asamoah et al.  in Ghana and another in India  showed that majority of women who died were from rural areas and did not attend antenatal clinic. However, this was not the case in our study where 51.1% of mortalities were women who had antenatal care. This scenario is not common in literatures regarding maternal mortality. The Gambia has a wide coverage of antenatal care; over 90% receive care at least once during pregnancy . Healthcare services are provided free to all pregnant women in the Gambia. This is a huge financial burden for a country with a GDP of $800 per capita . The dwindling economic fortune in recent years  will inadvertently affect the quality of healthcare available to pregnant women. Hence, although over 90% of pregnant Gambian women have some form of antenatal care, our findings reflect the quality of the maternity services and referral system in the Gambia.
In our study haemorrhage (26.5%), hypertensive disease (19.8%), sepsis (10.6%) and anaemia (8.9%) were consistently the commonest causes of maternal mortality in the period under review. Obstetrics haemorrhage remained the commonest cause (26.5%) and advancing maternal age and parity has strongly been associated with maternal deaths. Advanced maternal age and increasing parity are independent risk factors of obstetrics haemorrhage. Previous studies conducted in the Gambia on maternal mortality in rural and tertiary hospitals revealed haemorrhage as the consistent commonest cause of maternal death [6, 17,18,19]. The trend has not changed for over 3 decades. The reasons have not changed remarkably either, as late arrival to hospital due to delay in decision and transport still persists.
In this review, 46.9% of deaths occurred within 24 h of arrival. Also availability of compatible blood and donors has remained a challenge. In our study 73% of deaths occurred in the postpartum period and haemorrhage remained the consistent commonest cause. This is consistent with similar studies conducted in the sub-region [11, 20,21,22].
While our study shows similar pattern on the causes of mortality from global studies [23,24,25,26], it is pertinent to note that studies from developed countries highlight indirect causes of maternal deaths like cardiac diseases as being more common [27,28,29]. Direct causes of maternal death like haemorrhage, hypertension and sepsis are largely preventable. That they continue to feature as major causes of maternal deaths in the Gambia is a reflection of the socioeconomic development of the country. Blood, anti-hypertensive medications and antibiotics may not be available when needed as a consequence of this economic challenge.
In our study anaemia is the most common cause of indirect maternal death however, the impact of malaria in pregnancy on chronic anaemia may have reduced as malaria contributed approximately 3% of maternal deaths in the period under review. A previous study conducted in the same hospital showed that during the malaria season, there was a 168% increase in the maternal mortality ratio (MMR), a threefold increase in the proportion of deaths due to anaemia, and an eightfold increase in the anaemia MMR . In comparison, this study found a steady decline in MMR during the malaria season although not statistically significant (Table 2, Fig. 3). Figure 3 also shows that the anaemia related maternal mortality has declined steadily over the years. The proportion of women using sulphadoxine–pyrimethamine combination for intermittent preventive treatment of malaria in pregnancy has steadily improved from 21% in 2007 to more than 90% in 2010 [9, 10]. This may explain the decreased role of anaemia and malaria as causes of maternal mortality. This may also explain why the seasonal variation in maternal mortality seen in a previous study was not seen in this study . However, other studies from the sub-region also point to anaemia as a leading cause of indirect maternal deaths [21, 30].
The study has a number of limitations including the proportion of missing case notes (10.7%) and missing information from available case notes. In addition, autopsy is not routinely done in this environment and thus the causes of maternal death were determined clinically in this study.