記住我
ORIGINAL ARTICLE
Year : 2022 | Volume
: 29
| Issue : 4 | Page : 325-333
Predictors and causes of in-hospital maternal deaths within 120 h of admission at a tertiary hospital in South-Western, Nigeria: A retrospective cohort study
Joseph Ayodeji Olamijulo1, Gbenga Olorunfemi2, Halimat Osman3
1 Department of Obstetrics and Gynaecology, College of Medicine, University of Lagos, Lagos, Nigeria
2 Division of Epidemiology and Biostatistics, University of Witwatersrand, Johannesburg, South Africa
3 Department of Obstetrics and Gynaecology, Lagos University Teaching Hospital, Lagos, Nigeria
Correspondence Address:
Joseph Ayodeji Olamijulo
Department of Obstetrics and Gynaecology, College of Medicine, University of Lagos, Lagos
Nigeria
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/npmj.npmj_180_22
Background: An efficient, comprehensive emergency obstetrics care (CEMOC) can considerably reduce the burden of maternal mortality (MM) in Nigeria. Information about the risk of maternal death within 120 h of admission can reflect the quality of CEMOC offered. Aim: This study aims to determine the predictors and causes of maternal death within 120 h of admission at the Lagos University Teaching Hospital, LUTH, Lagos South-Western, Nigeria. Methods: We conducted a retrospective cohort study amongst consecutive maternal deaths at a hospital in South-Western Nigeria, from 1 January 2007 to 31 December 2017, using data from patients' medical records. We compared participants that died within 120 h to participants that survived beyond 120 h. Survival life table analysis, Kaplan–Meier plots and multivariable Cox proportional hazard regression were conducted to evaluate the factors affecting survival within 120 h of admission. Stata version 16 statistical software (StatCorp USA) was used for analysis. Results: Of the 430 maternal deaths, 326 had complete records. The mean age of the deceased was 30.7± (5.9) years and median time to death was 24 (5–96) h. Two hundred and sixty-eight (82.2%) women out of 326 died within 120 h of admission. Almost all maternal deaths from uterine rupture (95.2%) and most deaths from obstetric haemorrhage (87.3%), induced miscarriage (88.9%), sepsis (82.9%) and hypertensive disorders of pregnancy (77.9%) occurred within 120 h of admission. Admission to the intensive care unit (P = 0.007), cadre of admitting doctor (P < 0.001), cause of death (P = 0.036) and mode of delivery (P = 0.012) were independent predictors of hazard of death within 120 h. Conclusion: The majority (82.2%) of maternal deaths occurred within 120 h of admission. Investment in the prevention and acute management of uterine rupture, obstetric haemorrhage, sepsis and hypertensive disorders of pregnancy can help to reduce MM within 120 h in our environment.
Keywords: In-hospital maternal death, maternal death within 120 h, maternal mortality, Nigeria, predictors of maternal death, survival experience of maternal deaths, time to death Lagos
Maternal death is defined as the death of a woman while pregnant or within 42 days of delivery or termination, irrespective of the duration and site of the pregnancy from any cause related to or aggravated by the pregnancy or its management but not from accidental causes.[1] Globally, about 809 maternal deaths occur daily, with majority of them occurring in low-and middle-income countries (LMICs) such as Nigeria.[2] According to the 2017 estimates of the World Health Organization (WHO), Nigeria has the second highest number of maternal deaths with about 67000 estimated annual maternal deaths.[2] This constituted about 23% of the global maternal deaths despite having a population that is only about 2.6% of global population.[2] The current maternal mortality ratio (MMR) in Nigeria is very high at about 917/100,000 live births.[2] Although pregnancy and childbirth involve largely physiological processes, the WHO still estimated that women in Nigeria have 1 in 22-lifetime risk of dying from pregnancy complications compared to a risk of 1 in 49000 in high-income countries (HICs).
More than 80% of the causes of maternal deaths in Nigeria and most other LMICs are due to preventable direct causes such as hypertensive disorders of pregnancy, obstetric haemorrhage, sepsis, obstructed labour and its sequelae and unsafe abortion.[2],[3],[4],[5],[6],[7],[8] Regular antenatal visits, conduct of deliveries by skilled attendants and access to health institutions that can provide emergency obstetrics care (EmOC) are some key evidence-based interventions capable of reducing MMR in LMICs.[3],[4],[5],[6] Indeed, EmOC is targeted at managing the acute complications of the above-mentioned major causes of direct maternal deaths, thereby having the potential to considerably reduce the burden of maternal deaths by about 85%.[9],[10],[11] EmOC may be basic (covering parenteral administration of antibiotics, uterotonics, anticonvulsant; performance of assisted vaginal delivery (vacuum), removal of products of conception and manual removal of retained placenta) or comprehensive (including caesarean section, blood transfusion services and basic EmOC).
It has been argued that short hospital stay before death may suggest that there was inadequate personnel and/or infrastructure to respond to acute obstetric emergencies.[4],[12] A review of the literature on maternal mortality (MM) in Nigeria showed that only few studies reported the duration of hospital stay before maternal death, and none conducted a rigorous analysis to assess the pattern and predictors of time from admission to death. Poor documentation resulting in missing data on the duration of hospital stay may be one of the reasons. Some studies in Nigeria and the majority of LMICs reported that about 40%–50% of maternal deaths occurred within the first 24 h of admission.[4],[8],[12],[13],[14] In contrast, Mohamed et al. reported that only one-quarter of maternal deaths occurred within 24 h in a tertiary hospital in Egypt with a relatively low MMR of 186/100,000 live births.[11] This is similar to reports from HICs.[13] The reason for higher prevalence of death within 24 h amongst women in LMICs may be related to sub-optimal management of acute intrapartum and immediate post-partum conditions in the LMICs.[12],[13] In a systematic review by Dol et al., about 79.1%, 58.2% and 44.3% of deaths due to PPH, embolism preeclampsia/eclampsia, respectively, occurred within 24 h of birth. Whereas 63.1% of deaths due to sepsis occurred between 8 and 42 days.[13] It is therefore important to document the pattern of causes of maternal deaths in relation to timing in our environment to inform policy reviews.
We surmise that a reduction in the proportion of deaths within the first 120 h of arrival in the hospital will translate to a reduction in the overall annual MMR. This study aimed to evaluate for the first time, the predictors of time to maternal death within 120 h of admission at a tertiary hospital in South-West, Nigeria, using the survival analysis methodology. It is hoped that this study will further contribute to the methods of evaluating the outcome of EmOC at our institution and elsewhere.
MethodsWe conducted a retrospective cohort study of 430 consecutive maternal deaths at Lagos University Teaching Hospital, LUTH, Lagos, South-Western, Nigeria over an 11-year period from 1 January 2007 to 31 December 2017, using data from patients' medical records. We compared the characteristics of women that died within 120 h and those that survived beyond 120 h.
Study setting
The study was conducted at the department of obstetrics and gynaecology of LUTH South-Western, Nigeria. The obstetrics and gynaecology department of the hospital conducted an average of 1,500 deliveries per annum during the study. The hospital manages patients that registered at the hospital for antenatal and maternity care (referred to as booked patients) and other patients who were referred from other facilities (unbooked patients).
Study population
These were all maternal deaths that occurred at the hospital from 1st January 2007 to 31st December 2017.
Sampling and sample size
Being a retrospective study, all the available records were utilised. Thus, sample size and sampling were not conducted.
Outcome measure
Maternal death within 120 h of admission. The time-variate variable was “Time from hospital admission to death.” Early deaths were defined as death within 120 h of admission while late deaths were defined as death occurring after 120 h of admission.
Data collection and procedure
Socio-demographic characteristics, booking status, diagnosis and cause of death, date and time of admission, date and time of death, duration of hospital stay, cadre of staff (consultant, senior registrar and junior registrar), types of intervention received by each woman were obtained from the case files of the decedents using a data extraction sheet.
The apparently dominant cause of death was taken as the cause of death. The 10th revision of the International Statistical Classification of Diseases and related problems (ICD-10) diagnostic codes by the WHO were used to classify the maternal deaths.
Ethical considerations
Ethical approval for this study was obtained from the Human Research and Ethics Committee of LUTH (Ref. No: ADM/DCST/HREC/APP/3867), on 5th October 2020, before commencement of the study. The study was an anonymous review of medical records. Confidentiality was ensured and the strict ethical principles were followed in the conduct of the study.
Statistical analysis
Data were entered into an Excel spread sheet and then imported into Stata version 17 (Statacorp, College Station, Texas, USA) Statistical Software for analysis. Data validation and data cleaning were done. Descriptive statistics was conducted, and categorical variables were presented as frequency and percentages, while continuous variables were presented as mean and standard deviation or median and interquartile range (IQR) (for skewed data). The annual prevalence of mortality within 120 h was calculated and plotted on a graph.
The time-variable outcome of interest was the time to death (in hours). Since all the study population failed (all were maternal death), the duration between admission to death was “stset” in Stata. The association between categorical and continuous variables and death within 120 h was respectively conducted using Pearson's Chi-square and Student's t-test (or Wilcoxon rank sum test in case of nonnormally distributed variable). Survival life table analysis was utilised to describe and evaluate the probabilities of death at 6 h, 12 h, 24 h, 72 h and 120 h and other time points. The analysis was stratified by selected sociodemographic characteristics and log-rank test of equality of survivors was utilised to compare survival across groups. The overall mortality rate and within 120 h of admission amongst women that died was also calculated. Kaplan–Meier plots were conducted to describe survival experience within 120 h of admission. Univariable and multivariable Cox proportional hazard regression were conducted to evaluate the association of baseline factors affecting death within 120 h of admission. Baseline variables with Wald's P < 0.2 from univariable Cox proportional hazard regression analysis were included in the multivariable analysis using a stepwise backward elimination technique. Age was chosen a priori. Crude and adjusted hazard ratio (HR) of death within 120 h of admission was obtained and reported. The Schoenfeld's global test was utilised to test the final model for any violation of the assumption of parallelism or Cox proportionality. The Harrell's C concordance statistics was also used to test if the multivariable model was a good discriminant between the hazard of death or survival within 120 h of admission. Statistically significant levels were set at 95% confidence interval (P < 0.05). Two-tailed test of hypothesis was assumed.
ResultsSocio-demographic and clinical characteristics
During the study (1st January 2007–31st December 2017), 104 (24.19%) of the 430 maternal deaths had incomplete records on the duration of their hospital admission. Thus, 326 maternal deaths were further analysed. The mean age at death was 30.7± (5.9) years and about half (n = 173, 53.7%) were aged 30–39 years and majority (n = 192, 85.3%) of the women were married [Table 1].
Table 1: Comparison of the sociodemographic characteristics of decedents who survived beyond 120 h and those that died within 120 h in LUTH, Lagos (2007-2017)[Table 2] shows some clinical characteristics of the deceased. The majority of the women that died were unbooked (n = 280, 87.2%) and physically reviewed by the consultant before death (n = 233, 77.7%). Furthermore, nearly half (n = 132, 46.6%) of the women that died had caesarean section and about one-third (n = 106, 33.9%) had intensive care unit (ICU) admission.
Table 2: Comparison of the clinical characteristics of decedents who survived beyond 120 h and those that died within 120 h LUTH, Lagos, (2007-2017)Duration of admission before death
The median time to death was 24 (IQR: 5–96) h. Furthermore, nearly half (n = 141/326, 43.3%, 95% confidence interval [CI], 38.0%–48.7%) of the maternal deaths occurred within 24 h of admission and about 82.2%, 95% CI: 77.7%-86.0%, n = 268/326) of the decedents died within 120 h of admission [Figure 1].
Figure 1: Distribution of the duration from admission to death of maternal mortalities at LUTH, Lagos southwestern (2007 – 2017)Trends in prevalence of deaths within 120 h from 2007 to 2017
The trends in the prevalence of maternal death within 120 h fluctuated from 2007 to 2017. The minimum and maximum prevalence of maternal deaths within 120 h occurred in 2012 (63.3%, 95% CI: 44.3%–78.9%) and 2014 (100.0%), respectively. The prevalence of early maternal death appeared to increase in the last 3 years of the study (20015–2017) [Figure 2].
Figure 2: Trends in prevalence of maternal deaths within 120 hours of admission at LUTH, Lagos southwestern (2007 – 2017)Comparison of socio-demographic and biological characteristics amongst early (<120 h) and late deaths maternal (≥120 h)
There was no statistically significant difference in the age, parity, marital and booking status of the patients that had early death within 120 h as compared to late death beyond 120 h. All (100%, 8/8) women that died due to a first-trimester pregnancy complication died within 120 h of admission while just above half of the women that died during the second trimester of pregnancy died within 120 h of admission (53.3%, n = 7/8). Furthermore, 80.2% of those that died in the third trimester died within 120 h of admission. About 74.1% (n = 180/268) of the women that died within 120 h were reviewed by the consultant while about 93% (n = 53/58) of women who survived beyond 120 h were reviewed by the consultant before death.
There was a higher prevalence of ICU admission amongst women that survived beyond 120 h as compared to women that died within 120 h (49.1% vs. 30.5%, P = 0.007). Almost all the patients that died from uterine rupture (n = 21/22, 95.2%), obstetric haemorrhage (n = 48/55, 87.3%) induced miscarriage (n = 16/18, 88.9%) and sepsis (n = 58/70, 82.9%) died within 120 h of admission. Furthermore, 77.9% (n = 74/95) of women that died from hypertensive disorders of pregnancy died within 120 h of admission [Table 2].
Overall survival experience/probabilities amongst decedents of maternal deaths at a LUTH, Lagos
The overall follow-up time for all the participants was 31,812 h and the mortality rate was 10.2 per thousand women hours. The overall median time to death was 24 (5–96) hours and the minimum and maximum duration of admission before death was 0.25 and 3192 h, respectively. When stratified based on early (within 120 h of admission) and late (≥120 h of admission), the overall follow-up time for early and late deaths was 7,622 h and 24,191 h, respectively, while the mortality rate was 35.2 and 2.4 per thousand women-hours for early and late periods of deaths respective. The risk of surviving beyond 6 h of admission was 73.0% (95% CI: 68.0%–77.0%) and the risk of surviving beyond 120 h dramatically reduced to 17.5% (95% CI: 14.0%–22.0%) [Supplementary Table 1].
[Supplementary Table 2] shows the differences in overall survival across sociodemographic characteristics. There was no statistically significant difference in survival experience based on age, marital status, parity, gestational age and booking status (P > 0.05). However, overall survival experience was statistically different across category of year of diagnosis (P = 0.0378), admission to ICU (P = 0.0030), clinical diagnosis (P = 0.0238), mode of delivery (P = 0.0329) and the most senior cadre of doctor that reviewed before death (P < 0.0001). The median time to death amongst women admitted or not admitted to ICU was 50 h and 24 h, respectively.
Kaplan–Meir Plot for survival experience within 120 h of admission
[Figure 3] shows the Kaplan–Meir Plot of the survival experience amongst women that died within 120 h. The median time to death of women that died within 120 h was 20 h [Figure 3]. Furthermore, 25% of women that died within 120 h of admission died within 3 h of admission, while 75% of those that died within 120 h died within 37.4 h of admission.
Figure 3: Overall Kaplan-Meir Survival experience within the first 120 hour of hospital admission in LUTH, Lagos, Nigeria (2007-2017)[Supplementary Figure 1]a, [Supplementary Figure 1]b, [Supplementary Figure 1]c, [Supplementary Figure 1]d shows that there was no statistically significant difference in the survival experience of the dead women based on their age, marital status, gestational age and booking status. About 25% of booked and unbooked women, respectively, died within 4 h and 3 h of presentation while half of booked and unbooked women died within 24 h and 19 h, respectively. Subsequently, 25% of booked women survived beyond 25 h while 25% of unbooked patients survived beyond 40 h.
[Supplementary Figure 2]a shows that the women died earlier during the period 2007–2012 as compared to maternal deaths during the period 2013–2017 (logrank P = 0.0029). Indeed, about one-quarter of the women died within 2 h of arrival in the hospital during 2007–2012 while one-quarter of women died within a longer period of 12 h during the second year period of 2013–2017. The median time to death was also shorter during 2007–2012 as compared to the later period of 2013–2017 (11 h vs. 24 h). Furthermore, 25% of women that died during 2007–2012 survived beyond 28 h while 25% of women that died during the 2013–2017 period survived beyond 50 h [Supplementary Table 2].
The women who were not physically reviewed by the consultant (logrank P < 0.001) [Supplementary Figure 2]d, not admitted to ICU (logrank P = 0.004) [Supplementary Figure 2]b or died undelivered (Log-rank P = 0.0156) [Supplementary Figure 2]c died earlier than others or had a worse survival experience than others. [Supplementary Figure 2]e shows women who had obstetric haemorrhage generally died earlier while those that had sepsis died later. About 75% of those that died of uterine rupture, obstetric haemorrhage, hypertensive disorder of pregnancy and sepsis died within 22, 25, 30 and 72 h of admission, respectively.
[Table 3] shows the multivariable regression model of the hazard of death within 120 h. At univariable analysis, the hazard of death within 120 h of admission was about 30% less amongst women that died during the 2013–2017 as compared to women that died between 2007 and 2012, but this association was lost after controlling for other confounders. Although the hazard of death within 120 h was 1.2 folds amongst unbooked as compared to booked patients, the relationship did not reach statistical significance (P = 0.433). After correcting for confounders, the hazard of death within 120 h was associated with cadre of the most senior attending physician, diagnosis, mode of delivery and admission to ICU. The hazard of deaths within 120 h increases with decreasing level of seniority of the highest-ranked attending medical personnel. Furthermore, the risk of death within 120 h was highest amongst women who died from obstetric haemorrhage (adjusted HR: 3.30, 95% CI: 1.08–10.05, P = 0.036) followed by women who died from hypertensive disorders of pregnancy (adjusted HR: 3.18, 95% CI: 1.06–9.50, P = 0.038). Women who did not have ICU admission had about 1.6 folds hazards of dying within 120 h as compared to women who had ICU admission (Adjusted HR: 1.58, 95% CI: 1.13–2.22, P = 0.007). Women who had caesarean section or spontaneous vaginal delivery had about 58% and 50% lesser hazard of death within 120 h as compared to women who were undelivered. When compared with caesarean section, women who delivered through all other mode of deliveries had a higher hazard of death within 120 h. The Harrell's C concordance statistics was 0.6854, which showed that the multivariable model well discriminated the hazard of death within 120 h of admission. There was no violation of the proportionality assumptions since the Schoenfeld's global P > 0.05 (P = 0.2693).
Table 3: Univariable and multivariable Cox proportional hazard regression analysis of the risk of maternal death within 120 hours of admission at a tertiary hospital, in South-Western, Nigeria Discussion This study applied survival analysis methodology to evaluate the causes and predictors of time from admission to maternal death at a tertiary hospital, in south-western Nigeria from 2007 to 2017. We found that women who had obstetric haemorrhage, hypertensive disorder of pregnancy, were managed by only registrars (without physical presence of the consultant), died undelivered, died between 2007 and 2012 or were not admitted to ICU had a higher hazard of death within 120 h. The mode of delivery was also associated with early mortality.
It was difficult to compare our findings with previous literature as the use of survival methods to evaluate time to maternal death is not common. This study showed that nearly half (43.3%) of the maternal deaths occurred within 24 h of admission while about four-fifths occurred within 120 h of admission. The proportion of women that survived beyond 24 h at our centre (56.7%) was similar to the report from other institutions in Nigeria (51.9%)[15] and Gambia (53.1%),[3] but lower than the reports from North central Nigeria (63.6%);[12] Accra, Ghana (60.4%)[8] and Egypt (59%).[11] However, the survival rate beyond 24 h was higher at our centre as compared to some facilities in Nigeria and elsewhere (29.7%–46%).[16],[17],[18]
We found that the median time from admission to death was shorter during 2007–2012 as compared to the later period of 2013–2017 (17 h vs. 24 h) and the hazard of death within 120 h of admission was about 30% less likely amongst women that died between 2013 and 2017 as compared to women that died between 2007 and 2012. In addition, the proportion of maternal deaths within 24 h was lower than a previous report from the same institution (59% in 2002–2006 vs. 46% in 2007–2017).[4] A similar reduction in the proportion of maternal deaths within 24 h of admission have been reported from other tertiary hospitals, in Nigeria.[12] The decrease in the proportion of deaths within 24 h of admission at our centre and other hospitals in Nigeria may suggest an improvement in the provision of EmOC at the hospitals over time. Prompt access to acute resuscitative measures, blood banking services and emergency operative procedures might have improved during the later period. In addition, reduction in the number of women dying within 24 h of hospital stay may be because peripheral centres refer patients earlier than they used to do.[4],[12],[19] It has also been observed that tertiary centres with lower proportion of deaths within 24 h (<50%) generally reported lower MMR (<800 per 100,000)[8],[11],[12] and vice-versa.[16],[17],[18] Indeed, our study showed that about 82% of women with recorded length of hospital stay died within 120 h of admission which further suggests that improvement in resuscitative measures and upgrade of acute maternal care can considerably reduce MM at our centre.
Majority of patients that died from the major causes of MM such as uterine rupture (n = 21/22, 95.2%), obstetric haemorrhage (n = 48/55, 87.3%), sepsis (n = 58/70, 82.9%) and hypertensive disorder of pregnancy (n = 74/95, 77.9%) died within 5 days of admission. Although obstetric haemorrhage is the second or third leading cause of death at most centres in Nigeria and other LMICs,[4],[5],[16],[18],[20] we found that the hazard of death within 120 h was highest amongst patients with haemorrhage and hypertensive disorders of pregnancy. Thus, it is important to ensure that personnel, facilities and drugs are mobilised for the acute care of the aforementioned major causes of MM.[4],[9],[10],[11] Acute response to obstetric haemorrhage, uterine rupture and hypertensive disorder of pregnancy should be prioritised to reduce acute deaths and overall maternal deaths in the hospital. To reduce death from hypertensive conditions such as eclampsia (with pulmonary edema and other life-threatening complications), access to high-dependency or ICU s should be prioritised.[21]
A high proportion of maternal deaths within 24 h of admission at tertiary hospitals had been largely attributed to late referral and or sub-optimal care from lower-tier hospitals.[4],[8],[12],[18],[19] Surprisingly, this study found that a quarter of the booked and unbooked women, respectively, died within a similar period of about 3–4 h of presentation and the median time to death of booked and unbooked patients were 24 and 19 h, respectively. This pattern would suggest that acute care within the first 3 h of admission produced similar outcome amongst both booked and unbooked patients at our centre. It is possible that causes of maternal death within the first 3 h of presentation were of such catastrophic nature that extraordinary interventions were required to make any difference. Although the unbooked patients contributed a disproportionately higher ratio to MM,[22] our study showed that there was no statistically significant difference in the hazard of death within 120 h of admission between booked and unbooked patients. This is probably the first study to document and compare the length of hospital stay before death between booked and unbooked patients. Further studies with larger sample size (probably multi-centre) are required to make valid conclusions on this.
We also found that women who did not have ICU admission had about 1.6-fold hazards of dying within 120 h as compared to those who had ICU admission. Our finding supports the report of a WHO multi-country study that availability and access to ICU amongst acutely ill women can reduce the prevalence of MM.[23] Failure to admit deserving patients to the ICU may be due to a lack of ICU facility, inability of patient's relatives to pay or failure of personnel to prescribe ICU admission. Indeed, obstetric ICU and high-dependency care should be prioritised alongside other EmOC components. Further studies on risk factors of mortality amongst obstetric patients admitted to ICU are also useful in designing obstetric ICU interventions in LMICs.[24],[25] Although caesarean section is associated with higher morbidity and mortality as compared to spontaneous vaginal delivery,[26] however our study revealed that caesarean section appears to be protective with regards to maternal death within 120 h of admission. Thus, we suggest that caesarean section should be performed promptly when indicated to reduce early maternal death.[27]
The development and use of maternal health indices and indicators other than MMR such as maternal near-miss ratio, severe maternal outcome and mortality index to monitor progress in maternity care are steps in the right direction.[6] We however propose that the evaluation of the progress in the improvement in maternal health and maternity care should not only be evaluated based on central rates such as MMR or other rates.[28],[29] They are too gross to capture the details of when and why women die and may not sufficiently illuminate the path for prioritising intervention strategies especially when resources are limited.[28],[29] A limitation of the widely used maternal health rates is that they are expressed per live birth instead of per women-hours, women-weeks or women-months of pregnancy or post-childbirth.[29] Furthermore, survival analysis was recently utilised to produce evidence for extending MM surveillance within and beyond 42 days and to provide evidence on the survival risk of pregnant women during and after pregnancy.[13],[28],[29] We therefore hope that the application of survival analysis to describe and evaluate predictors of time from admission to death will contribute additional evidence to monitoring and reducing MM and improving maternal health, especially EmOC.
Strength and limitations of the study
The conclusion of this study should be interpreted within its limitations. Being a retrospective cohort study, there was missing data on the outcome of interest (duration of hospital stay) and other variables. Our study findings may not be generalisable to all the health facilities in or outside Nigeria since the study is a tertiary hospital review. Furthermore, this study only utilised available variables for the modelling. The sample size may not be adequate to avoid type 2 error. Nevertheless, this study is probably the first to utilise survival analysis methodology to evaluate the probabilities and predictors of duration of admission before death amongst maternal deaths in our environment. Thus, we have contributed evidence towards unravelling interventions to reduce maternal deaths, especially during the early hours of admission. A nationwide multi-centre study is currently being designed.
ConclusionAbout 82% of maternal deaths occurred within 120 h of admission at our tertiary hospital in Lagos, South Western Nigeria. Furthermore, the majority of patients that died from the major causes of MM such as uterine rupture, obstetric haemorrhage, sepsis and hypertensive disorders of pregnancy died within 120 h of admission. Women who were not admitted to ICU, not physically reviewed by consultants, and died between 2007 and 2012 had a higher hazard of maternal death within 120 h of admission. Thus, an improvement in the acute management of emergencies at our centre, coupled with the provision of adequate obstetric ICU spaces can reduce early maternal death within 120 h and hence reduce the overall institutional MMR.
Financial support and sponsorship
GO is funded by GSK Africa Non-Communicable Disease Open Lab through the DELTAS Africa Sub-Saharan African Consortium for Advanced Biostatistics training programme. The views expressed in this publication are those of the author(s) and not necessarily those of GSK. GSK grant number D1702270-01.
Conflicts of interest
There are no conflicts of interest.
References
留言 (0)