Pregnancy loss and postpartum mortality in a sub-Saharan African setting

Objective: Studies from industrialized countries report a harmful association between pregnancy loss and postpartum mortality. However, the nature of this relationship has not been established in resource limited sub-Saharan Africa. Given the potential implications of this knowledge for risk stratification and subsequent maternal health in sub-Saharan Africa, our study sought to use data from one of the continent's largest and longest running population-based cohorts to investigate the relationship between pregnancy loss and postpartum mortality. Methods: We conducted a population-based cohort study involving 25048 postpartum women from KwaZulu-Natal, South Africa. The study data was collected during biannual/triannual general household surveys, which also incorporated a pregnancy questionnaire for women who reported a pregnancy between survey waves. Pregnancy loss was defined as a pregnancy which ended in abortion, miscarriage, or stillbirth. Postpartum mortality was defined as the death of a woman, from any cause, within one year after the date that her pregnancy ended. We used a quasi-poisson regression model, adjusted for maternal age and other setting-specific predictors of postpartum mortality, to investigate the relationship between pregnancy loss and postpartum mortality. Results: The incidence of postpartum mortality was three times higher in women who experienced pregnancy loss (Adjusted Incidence Rate Ratio: 3.23, 95% Confidence Interval: 2.13-4.71), when compared with women who had live births. Conclusion: Our results reveal a clear association between pregnancy loss and increased postpartum mortality in a typical rural sub-Saharan African setting. Women who have recently experienced pregnancy loss should be targeted with a package of risk reduction interventions, including additional opportunities for medical care or psychosocial support.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ;https://doi.org/10.1101/2021 doi: medRxiv preprint 4 ABSTRACT 53 Objective: Studies from industrialized countries report a harmful association between 54 pregnancy loss and postpartum mortality. However, the nature of this relationship has not 55 been established in resource limited sub-Saharan Africa. Given the potential implications of 56 this knowledge for risk stratification and subsequent maternal health in sub-Saharan Africa, 57 our study sought to use data from one of the continent's largest and longest running 58 population-based cohorts to investigate the relationship between pregnancy loss and 59 postpartum mortality.

Methods:
We conducted a population-based cohort study involving 25048 postpartum 61 women from KwaZulu-Natal, South Africa. The study data was collected during 62 biannual/triannual general household surveys, which also incorporated a pregnancy 63 questionnaire for women who reported a pregnancy between survey waves. Pregnancy loss 64 was defined as a pregnancy which ended in abortion, miscarriage, or stillbirth. Postpartum 65 mortality was defined as the death of a woman, from any cause, within one year after the date 66 that her pregnancy ended. We used a quasi-poisson regression model, adjusted for maternal 67 age and other setting-specific predictors of postpartum mortality, to investigate the 68 relationship between pregnancy loss and postpartum mortality. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The current population of sub-Saharan Africa, which was estimated at 1.07 billion people in 80 2019, is expected to double by the year 2050. 1 On average, each sub-Saharan African woman 81 will have between 4 and 5 children during her lifetime. 1 Pregnancy and childbirth is not 82 without risk, 2, 3 and postpartum mortality will continue to be of public health relevance in this 83 resource-constrained region for many years to come. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.18.21253872 doi: medRxiv preprint characteristics, such as age, date of last menses, and prior maternal history; and pregnancy 128 characteristics, such as the date that the pregnancy ended and the outcome of the pregnancy. 13 129 Data on pregnancy outcomes are predominantly collected using a single item on the 130 pregnancy questionnaire -"Was the baby born alive, born dead (i.e. stillbirth), or did the 131 pregnancy end in a miscarriage or abortion?". Only one of the aforementioned pregnancy is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The study outcome, postpartum mortality, was defined as the death of a woman from any 155 cause, within one year after the date that her pregnancy ended. This was established from 156 subsequent waves of the general household survey. For a woman who had suffered 157 postpartum mortality, the follow-up time (in woman yearswy) was calculated as the 158 difference in time between the date of her death and the date that her pregnancy ended.

159
Women who did not suffer postpartum mortality were censored at one year after the date that 160 their pregnancy ended.

162
Eligibility criteria: 163 We included records from AHRI population-based surveys for all women aged ≥16 years old Briefly, there were 83720 women aged ≥16 years old residing in the AHRI study area 168 between 2000 and 2017. Of these 83720 women, 82907 agreed to participate in the general 169 household survey and provide pregnancy information during subsequent waves of the survey 170 (equating to a response rate of 99%). The study sample was comprised of 25048 women after 171 excluding women who did not report pregnancies during the study period, women who had 172 twins, and subsequent pregnancies for the same woman (i.e. only the first recorded pregnancy 173 during the study period was used for each woman). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.18.21253872 doi: medRxiv preprint 9 Data analysis: 177 We used R version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria) to 178 perform the data analysis. We used descriptive statistical methods to summarize the 179 characteristics of our study sample and present these summary statistics as frequencies and   196 197 The characteristics of the study sample, comprised of 25048 women, are shown in Table 1. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The total follow-up time for the cohort was 24944.94 wy. We were able to establish 205 postpartum mortality status for 100% of women in the study sample (i.e. there were no 206 women in our study with missing postpartum mortality information). A total of 178 women 207 (0.7% of the study sample) died, and the crude incidence of postpartum mortality was 208 estimated at 7.14 (95% CI: 6.15-8.28) per 1000wy.  adjusted for established predictors of maternal mortality in our setting (maternal age, level of 215 education, parity, and HIV serostatus), 6 found that postpartum mortality was three-times 216 higher in women who experienced pregnancy loss (Adjusted IRR: 3.23, 95% CI: 2.13-4.71). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Our analysis of data from over 25000 postpartum women who participated in one of Africa's 229 longest running and largest population-based cohorts suggests a three-fold higher incidence 230 of postpartum mortality in women who experienced pregnancy loss when compared with 231 women who had live births. A recent systematic review and meta-analysis by Reardon and 232 Thorp reported that the risk of postpartum mortality was two-fold higher (pooled odds ratio:  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

RESULTS
The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.18.21253872 doi: medRxiv preprint 12 subsequent maternal mental illness, self-destructive behavior, and suicide. 9, 14, 15 Therefore, Our analysis also found that increasing age and an HIV-positive serostatus were associated 263 with increased postpartum mortality, while a prior maternal history was associated with 264 decreased postpartum mortality. Increasing age is a common predictor of maternal mortality 265 throughout the world. 17 Although our study found a significant increase in postpartum 266 mortality amongst women in the 26-35 year age group, we were only able to observe a 267 statistical trend toward increased postpartum mortality in women aged >35 years old. It is 268 likely that there would also have been a statistically significant increase in postpartum 269 mortality in women aged >35 years old had the sample of women in this specific age group 270 been larger (only 6.2% of our sample was comprised of postpartum women >35 years old). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 13 nulliparous women and women with >7 prior pregnancies at highest risk for this outcome. 19

277
Our study confirms this widely accepted relationship between nulliparity and increased 278 maternal mortality. Thus, increasing age, HIV infection, and nulliparity are additional 279 markers of increased postpartum mortality risk in our setting, which need to be considered 280 during the risk stratification process.

282
The main strength of our study was that it was population-based and does not suffer from 283 many of the biases typically associated with facility-based studies. This renders our findings 284 more generalizable to pregnant women in sub-Saharan African settings. Another important 285 strength of our study was that the data collection involved a "full pregnancy history" (FPH) 286 questionnaire (which collects information on all pregnancies during a woman's life, 287 irrespective if these ended in live births or not) rather than the "full birth history" (FBH) 288 questionnaire (which has traditionally collected information only on live births, with updates 289 of the instrument in 2013 allowing for limited information on pregnancy loss to be 290 collected). 20 As such, the pregnancy outcomes data in this study is far more robust than if the 291 data had been collected using a FBH questionnaire. The large study sample size of 25048 292 postpartum women, with complete follow-up data, also facilitated an appropriate statistical 293 analysis. This was particularly important considering that only 3.8% of our study sample 294 experienced pregnancy loss.

296
The main limitation of our study is the potential for selection bias among reported pregnancy 297 outcomes. Information on pregnancy outcomes was self-reported by women who participated 298 in the AHRI general household surveys, and the information regarding pregnancy losses is 299 sensitive and potentially vulnerable to omission. 21, 22 As a result, it is possible that pregnancy 300 losses are underreported during the AHRI general household surveys. While is difficult to 301 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.18.21253872 doi: medRxiv preprint measure how this affects our results, we hypothesize that the association between pregnancy 302 loss and postpartum mortality would be even stronger in the absence of selection. This 303 proposition is supported by the previously established relationship between women who 304 experienced pregnancy loss and postpartum mortality, as well as the shared risk factors for 305 such events.

307
In conclusion, our research from a typical rural sub-Saharan African setting demonstrates a 308 three-fold higher incidence of postpartum mortality in women who had recently experienced 309 pregnancy loss, when compared with women who had live births. Our results indicate that is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted March 24, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021