Sepsis during pregnancy and specifically in the puerperal phase is one of the leading causes of maternal morbidity and mortality worldwide, accounting for up to 11% of maternal deaths in developing countries.^1,2,3,4,5 In the 2023 South African Saving Mothers Report (SASMR), 4.9% maternal deaths were because of pregnancy-related sepsis (PRS), which is similar to our institutional mortality rate of 5%.⁶ This is similar to the 5% mortality rate of severe sepsis in developed countries.⁷ A further 17.1% maternal deaths because of non-pregnancy related infection were reported from the SASMR.⁶ The total number of maternal deaths attributed to sepsis has steadily decreased from 70 in 2017 to 44 in 2023.^6,8 The SASMR of 2017 outlined PRS as ‘often underestimated by the healthcare provider’ with inadequate management.⁸ Although there has been a steady decline, there are still many avoidable deaths because of PRS. Puerperal sepsis, as defined by the World Health Organization, is any infection occurring from the onset of labour until 42 days postpartum.⁹ Clinically, it is defined by the presence of a fever and one of the following: abnormal vaginal discharge, pelvic pain, offensive discharge or a delay in involution of the uterus.^6,9,10 A favourable outcome is dependent on early recognition, timeous and appropriate commencement of antibiotic therapy and source control.¹¹ However, hysterectomy is also indicated when sepsis is complicated by the presence of disseminated intravascular coagulopathy, or when there is dysfunction of two or more end organs.⁸ The fifth SASMR found that delayed diagnosis and a low proportion of hysterectomies (14%) were suggestive of a delay in definitive treatment of the women who died from PRS.¹² This SASMR found the final cause of death in 90% of PRS deaths to be septic shock.¹² The implementation of the recommendations of this report, which encouraged early recognition and timeous hysterectomy, has seen a steady decline in maternal mortality as a result of PRS.¹²

A study conducted between 2007 and 2009 in the Eastern Cape revealed that secondary postpartum haemorrhage and puerperal sepsis accounted for 27% of the peripartum hysterectomies.¹³ In 2025, Mmabatswa et al. found that 32.1% of postpartum hysterectomies resulted from PRS in a study of 56 women who underwent peripartum hysterectomy at the Johannesburg General Hospital between 2018 and 2020. Almost 88.8% of these PRS were histopathologically confirmed as sepsis.¹⁴

Several risk factors have been identified as contributors to maternal sepsis. The predominant ones include the human immunodeficiency virus (HIV) seropositivity, prolonged rupture of membranes, prolonged labour, prolonged surgery, the presence of chorioamnionitis, previous surgery and bowel injury.¹²

Obstetric-related infections in the antepartum period include septic abortions and chorioamnionitis, while endometritis and wound infections occur in the postpartum period.¹⁵ Non-obstetric sources of infection in the antepartum and postpartum periods have predominantly been described in the urinary and respiratory tract.^9,16 The Global Maternal Sepsis Study found urinary tract, genital tract (endometritis), chorioamnionitis and respiratory tract infections as the most common maternal infections.^1,16 Other potential risk factors for puerperal sepsis include caesarean sections (CS), postpartum haemorrhage, preterm deliveries and comorbid conditions, such as diabetes and eclampsia.³

According to the fifth SASMR, avoidable factors, such as an overburdened healthcare system with a lack of facilities, led to missed opportunities for preventing maternal deaths from PRS. Thirty-three per cent of PRS deaths developed puerperal sepsis after a CS.¹² In 2010, a longitudinal descriptive study of puerperal infection found no risk identifying factors, which is contradictory to the findings of Mmabatswa et al.^14,17

While maternal mortality because of puerperal sepsis has been reduced by antibiotics coverage in developed countries, there is limited information on the common bacterial strains involved in South Africa and the potential sensitivity patterns amid the growing incidence of antibiotic resistance. Evidence of common infectious patterns and antibiotic sensitivity patterns is required for appropriate patient management and for antibiotic stewardship.² This study aimed to describe the demographic, intra-operative findings, organisms responsible for infection, the antimicrobial susceptibility patterns and the histological findings of women who had a hysterectomy as a measure of source control in the treatment of PRS.

This was a retrospective cross-sectional review of medical records at the Obstetrics unit of the Chris Hani Baragwanath Academic Hospital (CHBAH) situated in Soweto, Johannesburg, between 01 January and 31 December 2019. This is an Obstetrics unit of a tertiary teaching hospital, where women from peripheral clinics are referred. The study involved all women who underwent a hysterectomy for puerperal sepsis and whose samples were sent for histology and suspected endometritis during this study period. Women who had a miscarriage or a pregnancy at ≤ 24 weeks’ gestation were excluded. Demographic and clinical data, including comorbidities, previous surgical history, duration of labour, mode of delivery, surgical, microbiology data and sites, dissemination of intravascular coagulation from post-mortem findings and histological reports were extracted and analysed. The duration of CS was identified through theatre notes and categorised accordingly. Frequencies and percentages were used to describe the findings, including caesarean and hysterectomy characteristics, among others. Age, parity, gravity and gestational ages were summarised with medians and interquartile ranges (IQR). Differences between proportions were assessed using Fisher’s exact test. The differences between admission, pre-hysterectomy and post-hysterectomy values were evaluated using ANOVA (analysis of variance) for repeated measures test with a p-value of < 0.05 considered statistically significant using STATA software (StataCorp version 15).

During the same period, 33 hysterectomies were performed for puerperal sepsis (Table 1), representing 0.17% (n = 33/18 458) of all deliveries at this study site. Forty-five per cent of all deliveries were CS. The median gestational age of the study population at the time of delivery was 38 weeks (range 34 to 40), with a median body mass index (BMI) of 28 kg/m² (range 34 to 40). Seven (21%) women were HIV seropositive and eight women (24%) had hypertensive-related disorders (Table 1).

The outcomes of 33 women who underwent hysterectomy for puerperal sepsis at CHBAH during the study period are described. These women constituted 0.17% of the total (18 458 women) deliveries during this study period.

The background incidence of hypertension in pregnant women in Soweto (unpublished data) is 11%. The 15% incidence of hypertension in this study may highlight a higher level of hypertension among women with hysterectomy for puerperal sepsis. Whether this is because of the disease process or an increased frequency of iatrogenic interventions requires further investigation. This confirms previous findings of the contribution of hypertensive disease in pregnancy to placental insufficiency, but also pre-eclampsia. Iatrogenic delivery for growth-restricted foetuses via CS may contribute to a higher frequency of CS in this population, putting them at risk of sepsis associated with a CS.¹⁸

Total abdominal hysterectomy is a method of source control in the management of severe uterine sepsis. However, variations in technique, surgical experience and consistency of approach are not standardised but rather individualised.^9,19 In our study, 29 women (88%) had a TAH, while 4 (12%) had a STAH. There is no existing literature supporting that a TAH is superior to an STAH in the management of puerperal sepsis. The only evidence regarding TAH and STAH was in post- and peripartum haemorrhage.¹⁹ The mortality rate in this study was 6% which may suggest that early source control and probably by TAH, has better outcomes.

It is widely known that the common organisms of puerperal sepsis are E. coli, Group B streptococcus, Klebsiella pneumoniae and Staphylococcus species.²⁰ Our study found that all but one of the ESKAPE organisms, namely Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, A. baumannii and Enterobacter spp., were the most common organisms cultured. Pseudomonas aeruginosa was not cultured from any of the women. With A. baumannii being the most frequently cultured, followed by E. coli and Klebsiella, the same organisms were mostly cultured in a study conducted in the Western Cape, South Africa.¹⁹ Given the paucity of evidence from South Africa on the types of organisms that are most often cultured, this study adds to the body of knowledge on the subject. All three, A. baumannii, E. coli and Klebsiella of these organisms were found to be resistant to Augmentin, Ampicillin and Gentamycin, which are commonly prescribed in our hospital facility to treat puerperal sepsis and chorioamnionitis. The resistance to Ampicillin is increasing, prompting a re-evaluation of antibiotic prophylaxis for premature rupture of membranes which contributes to puerperal sepsis. However, E. coli was the only ESKAPE organism with some sensitivity to Augmentin. The organisms commonly cultured in this study showed that our current prophylactic and treatment regimens are ineffective at preventing sepsis that required hysterectomy as source control measure. As most of these organisms are resistant to our current prophylactic and treatment regimens, this needs further investigation. These findings substantiate the need to use targeted antibiotic therapy in the management of puerperal sepsis and highlight the importance of antimicrobial stewardship as more organisms are becoming resistant to broad-spectrum antibiotics. E. coli and A. baumannii were cultured from one of the women who died. Whether the colonisation with extremely drug-resistant A. baumannii may have contributed to the demise of this woman because of inadequate antibiotic availability needs confirmation. The study was, however, not designed to ascertain whether these organisms contributed to the demise of this woman. Because of the increasing incidence of A. baumannii and the varying sensitivity patterns among these patients, alternative broad-spectrum antibiotic treatment options may be required.²⁰

The source of sepsis was independent of HIV seropositivity in this study. It is, however, important to note the lower seropositivity rate in the current study than the background rate of HIV seropositivity of 28.1% in Gauteng.²¹ Additionally, antiretroviral treatment of the few seropositive cases was noted. At our centre, the protocol of postpartum prophylactic antibiotic use for 5 days in HIV seropositive women who had a CS may have led to no impact of HIV infection on the frequency of puerperal sepsis and hysterectomy. There was no difference in histopathological findings between HIV seropositive and seronegative women.

The findings of bacteraemia, respiratory tract and urinary tract infection highlight the importance of concurrent investigations to improve detection, as foci of infection may occur in more than one site.

The indications for hysterectomy in this institution are based on clinical features of the women, two or more organ dysfunction, necrotic cervix, pus draining from the cervix or if at laparotomy, the uterus appeared necrotic. A unique finding of this study was that histological and microbiological diagnoses are useful additions to the surgeon’s macroscopic diagnosis of necrosis and the appearance of pus, in comparison, evidence of which may improve antibiotic prophylaxis in sepsis. The indication for the hysterectomy was not based solely on the uterine appearance. There is a paucity of evidenced literature regarding the complementation of surgeons’ diagnoses of necrosis, which highlights the need for an appropriate scoring system to objectively identify women who require a hysterectomy.

In conclusion, timeous and appropriate hysterectomies as a measure of source control are required in uterine sepsis. Puerperal sepsis may result in multiple end organs being affected, motivating the use of multi-organ assessment models. This study creates awareness about the increasing number of organisms associated with puerperal sepsis, the rising antibiotic resistance in this population, suggesting a review of our prophylactic and treatment antibiotic regimens used in the population of South Africa.