Risk factors for composite adverse outcomes of postpartum haemorrhage in a low-resource setting: a single-centre cross-sectional study in Mpilo Central Hospital, Bulawayo,

Primary postpartum haemorrhage continues to cause Background: considerable global maternal morbidity and mortality. The aim of this study was to determine the risk factors for composite adverse outcomes in postpartum haemorrhage using multivariable logistic regression. The findings could potentially be used to anticipate and prevent composite adverse outcomes in postpartum haemorrhage. This was a retrospective cross-sectional study carried out at Methods: Mpilo Central Hospital, a government tertiary referral centre, covering the period 1 July 2016 to 30 November 2019. Participants were included in the study if they had a diagnosis of postpartum haemorrhage. Those variables that had a p<0.2 from the univariate logistic regression analyses were considered for multivariable logistic regression. The association between independent variables and the dependent variable was assessed using odds ratio with 95% confidence intervals, to identify independent risk factors for composite adverse outcomes in PPH. The independent risk factors for composite adverse outcomes in Results: postpartum haemorrhage were place of dwelling (AOR 4.57, 95% CI 1.87-11.12,  =0.01), prior history of a Caesarean section (AOR 2.57, 95% p CI 1.10-6.00,  =0.03), APH (AOR 5.45, 95% CI 2.23-13.27,  <0.0001), p p antenatal haemoglobin level (AOR 19.64, 95% CI 1.44-268.50,  =0.03), p and delivery by Caesarean section (AOR 10.21, 95% CI 4.39-23.74, p <0.0001). Blood loss was also an independent risk factor for composite adverse outcomes in postpartum haemorrhage with the following blood loss; 1001-1500 ml (AOR 9.94, 95% CI 3.68-26.88,  <0.0001), p 500-1000 ml (AOR 41.27, 95% CI 11.32-150.54,  <0.0001), and 2001 ml p (AOR 164.77, 95% CI 31.06-874.25, p<0.0001). This study found that the independent predictors for Conclusions: composite adverse outcomes in PPH were rural dwelling, prior history of a Caesarean section, antenatal haemoglobin level, delivery by Caesarean 1


Introduction
Primary postpartum haemorrhage (PPH) is defined as a cumulative blood loss from the genital tract of ≥500 mL or more following a normal vaginal delivery or ≥1,000 mL or more following a cesarean section within 24 hours of delivery evidenced by a rise in the pulse rate, and falling blood pressure [1][2][3] .
In 2017, approximately 810 women died from causes related to pregnancy and childbirth, and 94% of all maternal deaths occurred in low and lower middle-income countries 4 . In a systematic analysis, Say et al. found that low-and middle income countries accounted for 480,000 maternal deaths (32%) compared with 1200 (8%) in the developed regions 5 . PPH is the leading cause of maternal deaths in SSA 6 .
The multi-country Survey on Maternal and Newborn Health reported the prevalence of PPH as 1.2%, with higher rates in developing countries than developed ones 7 . Other studies in Sub-Saharan Africa (SSA) reported rates of 1.6% in Zimbabwe, 16.6% in Southern Ethiopia, 9% in Uganda and 23.6% in Cameroon. 1,3,8,9 , respectively. Ford et al. reported increasing PPH rates from 6.1% in 2003 to 8.3% in 2011 (p<0.0001) in Australia 10 .
Two-thirds of women with PPH having no identifiable risk factors 11 . Recognized risk factors for PPH include previous PPH, twin gestation, large baby, induction of labour, prolonged labour, operative delivery, preeclampsia, caesarean delivery, grand multiparity, maternal age 35 or above, and postdates 1,8,12-15 .
Tort et al. used a multivariable logistic mixed model to identify factors that were significantly associated with PPH maternal death 13 . However, in this study PPH maternal death or serious morbidity were used as composite adverse outcomes.
The aim of this research was to documents risk factors for poor composite adverse outcome in PPH. This could help clinicians identify which women with PPH are at risk of composite adverse outcomes and increase further the clinical vigilance associated with the management of PPH thereby preventing deaths.

Study type, setting and participants
This was a retrospective cross-sectional study carried out at Mpilo Central Hospital, a government tertiary referral centre, covering the period 1 July 2016 to 30 November 2019. Mpilo Central Hospital is situated in the township of Mzilikazi in Bulawayo. Bulawayo is the second largest city in Zimbabwe after the capital city Harare, with a population of 653,337 as of the 2012 census 16 . Participants were included in the study if they had a diagnosis of postpartum haemorrhage within 24 hours of delivery at Mpilo Central Hospital. Women that delivered outside the hospital were excluded from the study.

Independent variables
The independent variables included socio-demographic factors, mode of delivery, fetal characteristics, blood loss, laboratory tests, causes of PPH and the management of PPH.

Main outcome measure
The main outcome of interest for the study was the composite adverse outcome which included maternal death or serious morbidity (either of hypovolaemic shock or haemoglobin <4 g/dL or massive blood transfusion >4 units or hysterectomy or admission to ICU or coagulopathy or major organ dysfunction), similar to the Delphi consensus study on PPH 17 .

Sample size calculation
The Cochran sample size formula was used to calculate the sample size as follows; n 0 =z 2 pq/e 2 where n 0 =sample size z = is the selected critical value of desired confidence level p = is the estimated proportion of an attribute that is present in the population q = is 1-p and e is the desired level of precision Assuming the maximum variability, which is equal to 50% (p = 0.5) and taking 95% confidence level with ±5% precision, the calculation for the required sample size was as follows;

Data collection
Data collection was done using a paper data collection tool (see Extended data) 18 that was used to collect secondary data from the labour ward delivery registers, and mortality registers. Hospital case notes were retrieved the clinical data were extracted.

Data analysis
Data were cleaned, coded and entered into a Microsoft Excel spreadsheet, then exported to SPSS Version 20 (IBM, Armonk, NY, USA) for analysis. Descriptive statistical analyses were performed and presented as frequencies and percentages for categorical variables. Bivariate correlations of association between main independent variables and the outcome measures were performed using Pearson 2-tailed chi-square test. A p value of <0.05 was considered to be statistically significant, and these were considered for the univariate logistic regression. Those variables that had a p<0.2 from the univariate logistic regression analyses were considered for multivariable logistic regression. The association between independent variables and the dependent variable was assessed using odds ratio with 95% confidence intervals, to identify independent risk factors for composite adverse outcomes in PPH, holding other variables constant and adjusting for co-variates. The Hosmer-Lemeshow goodness-of-fit was used to check if the model fitted well. A p< 0.05 was taken as statistically significant.

Ethical approval
The Ethics Committee at Mpilo Central Hospital made a ruling for all retrospective studies to go ahead in the institution from 2016 onwards as long as the data remained anonymous; the committee waived the requirement for patient consent. No ethical issues will arise during the study as all the data will remain anonymous with no identifying personal data. Minutes of the Committee's inaugural meeting held on the 13 th October 2016 set out the requirements of all the studies at the institution.

Results
Socio-demographic characteristics of participants A total of 386 cases of PPH were recorded during the period 1 July 2016 to 30 October 2019. The summary of maternal and fetal characteristics are shown in the Supplementary Tables in the Extended data 18 . Deidentified results are available for each patient as Underlying data 18 . Table 1 and Table 2 show the results of the multivariable logistic regression. Rural women were 4.6 times more likely to be statistically significantly associated with composite adverse outcomes compared to women from urban areas (AOR 4.57, 95% CI 1.87-11.12, p=0.01).

History of Caesarean section
Women with a prior history of a Caesarean section were statistically significantly associated with composite adverse outcomes in PPH. Such women were 2.6 times more likely to be statistically significantly associated with composite adverse outcomes in PPH, compared to women without such history (AOR 2.57, 95% CI 1.10-6.00, p=0.03).
APH APH was statistically significantly associated with composite adverse outcomes in PPH. Women who presented with APH were 5.5 times more likely to be statistically significantly associated with composite adverse outcomes in PPH compared to women who had no APH (AOR 5.45, 95% CI 2.23-13.27, p<0.0001).

Antenatal haemoglobin count
Antenatal haemoglobin count was also statistically significantly associated with composite adverse outcomes in PPH. Women with haemoglobin counts of 0-5.99 g/dL were 19.6 times more likely to be statistically significantly associated with composite adverse outcomes in PPH compared with women with haemoglobin counts of 11 g/dL and above (AOR 19.64, 95% CI 1.44-268.50, p=0.03).

Delivery by Caesarean section
Delivery by Caesarean section was statistically significantly associated with composite adverse outcomes in PPH. Women who had a Caesarean section were 10.2 times more likely to be statistically significantly associated with composite adverse outcomes in PPH compared to women who delivered vaginally (AOR 10.21, 95% CI 4.39-23.74, p<0.0001).

Blood loss
Blood loss was statistically significantly associated with composite adverse outcomes in PPH. The odds rose significantly higher as the amount of blood loss increased. Women who lost 1001-1500 ml of blood were 9.9 times more likely to be statistically significantly associated with composite adverse outcomes, compared to women that lost 500-1000 ml (AOR 9.94, 95% CI 3.68-26.88, p<0.0001). The odds rose to 41.3 times more like to be associated with composite adverse outcomes in those women who lost 1501-2000ml compared to those women who lost 500-1000 ml (AOR 41.27, 95% CI 11.32-150.54, p<0.0001). Whereas women who lost 2001 ml and above were 164.8 times more likely to be statistically significantly associated with composite adverse outcomes in PPH compared to women who lost 500-1000 ml (AOR 164.77, 95% CI 31.06-874.25, p<0.0001).

Discussion
PPH rates have been reported to be rising in both lowincome and high-income countries 11,19 . This means that PPH will remain an important global subject. The strength of this research is that it involves a large homogenous group of patients with PPH, in SSA where PPH continues to contribute significantly to global mortality and morbidity.
Rural women were 4.6 times most likely to be statistically significantly associated with composite adverse outcomes compared to women from urban areas (AOR 4.57, 95% CI 1.87-11.12, p=0.01). National governments need to made healthcare accessible to rural women so that the Sustainable Development Goals on Maternal Mortality to reduce global maternal mortality ratio to less than 70 per 100,000 live births by 2030 4 could be achievable.
Women with a prior history of a Caesarean section were statistically significantly associated with composite adverse outcomes in PPH. These women are not only at risk of developing a PPH (OR 3.15, 95% CI 1.02-10.3) 20 , but the women were 2.6 times most likely to be statistically significantly associated with composite adverse outcomes in PPH, compared to women without such history (AOR 2.57, 95% CI 1.10-6.00, p=0.03). The means that women with a prior history of a Caesarean section should receive extra clinical vigilance.
APH was statistically significantly associated with composite adverse outcomes in PPH. Women who presented with APH were 5.5 times more likely to be statistically significantly  Women who lost 2001 ml of blood and above were 164.8 times more likely to be statistically significantly associated with composite adverse outcomes in PPH compared to women who lost 500-1000ml (AOR 164.77, 95% CI 31.06-874.25, p<0.0001). The amount of blood loss was found to be related to adverse maternal outcomes 19 . Prompt, effective management of PPH. 19 , should be the aim to reduce the amount of blood loss and prevent the development of composite adverse outcomes.

Limitations
The major limitation of this study is that it was a retrospective, single-centre study that used secondary data. This could limit the generalizability of its findings to other centres of low-resourced settings.

Conclusions
The independent predictors for composite adverse outcomes in PPH were rural dwelling, prior history of a Caesarean section, antenatal haemoglobin level, and delivery by Caesarean section. Blood loss was also an independent predictor for composite adverse outcomes in PPH. Crucially, this new information should help in increasing clinical vigilance and preventing maternal deaths especially in low-and middleincome countries were PPH mortality is of high prevalence. Regular on-site training of staff can focus on drilling on these important issues and can improve outcomes 21 .
This project contains the following underlying data: • PPH-Data-Share (XLSX). The raw de-identified data gathered from each patient examined in this study. This project contains the following extended data: • Data Collection Sheet-PPH (DOCX).