Impact of Medical Doctors Global Health and Tropical Medicine on decision-making in caesarean section: a pre- and post-implementation study in a rural hospital in Malawi

The proportion of caesarean births is increasing worldwide, in both high- and low-income settings [1‐3]. A worldwide ecological study representing 98% of global live births found no important association with maternal and neonatal mortality at caesarean section rates higher than 10%, in line with a systematic review by the World Health Organization and their statement on caesarean sections [4‐6]. Caesarean sections may be associated with severe maternal outcomes like massive blood transfusion, admission to intensive care unit, hysterectomy, and death, and these associations are strongest in sub-Saharan Africa [7‐9]. The risk of uterine rupture in subsequent pregnancies is increased, which is particularly problematic in countries with high fertility rates such as Malawi [10‐14]. Therefore, it remains crucial to carefully select those women eligible for surgical intervention and prevent caesarean sections without medical indication, for instance by following evidence-based indications and performing audit of indications [2]. In addition, the World Health Organization advises to monitor caesarean section rates at a facility level [6].

In the Medical Doctor Global Health and Tropical Medicine (MDGHTM) programme in the Netherlands, participants are trained to work as medical doctors in low-resource settings [15, 16]. Their postgraduate training has a duration of approximately 3 years and focuses on acquiring basic surgical and obstetric skills, as well as knowledge of tropical diseases and public health [17, 18]. MDGHTMs are expected to critically assess labour progress and indications for obstetric interventions, and are prepared to provide clinical assistance and onsite training to other (associate) clinicians in hospitals in low-income countries. MDGHTMs often work in under-staffed rural areas in low-resource settings and combine clinical, managerial, and educational tasks [16, 19]. However, the impact of their presence on obstetric management in their work environment has never been systematically assessed.

The first objective of this study was to describe the caesarean section rate during the period of introduction of MDGHTMs in a rural hospital in Malawi, where hospital management was concerned about increased caesarean section rates. Second objective was to examine whether there was a decrease in the proportion of caesarean sections that were not supported by medical indications.

Medical records of 3428 births were retrieved, of which 645 were caesarean sections (caesarean section rate 18.8%). Table 2 shows clinical characteristics of women who underwent caesarean section. Figure 2 shows a temporal trend of a reduced caesarean section rate per quarter during the study period (y = 0.0086x + 0.2295, p = 0.039). Medical officers performed 82 caesareans, clinical officers 513.

The commonest indication for caesarean section was cephalopelvic disproportion (CPD; 23.5%), followed by foetal distress (12.7%) and prolonged first stage of labour (of unknown reason; 12.2%). Nine percent of caesarean sections were performed because of two previous caesarean scars. All indications are listed in Table 3.

Of 645 caesarean sections, 501 had an auditable indication: 90 (13.9%) had indications that were not auditable (not part of the criteria listed in table 1) and 54 (8.4%) did not have a documented indication (Table 3). One hundred and thirty-two (26.3%) of the audited cases had an unsupported indication: documentation in partograph and file did not meet criteria for that specific indication. In 71 (14.2%), too little information could be found in the partograph to assess the indication properly. In 298 (59.5%) of the audited cases, partograph and file provided evidence to justify the indication (see Fig. 3).

Fig. 4 shows the proportion of unsupported indications among the total auditable caesarean sections. There was a significant downward trend over time (y = − 0.0481 + 0.4759, p = 0.008). This trend was not only seen in all caesarean sections, but also in those performed by clinical officers (Fig. 5).

Looking into the different groups of indications, 93 (70.5%) of the caesarean sections with unsupported indications were due to prolonged labour. In these women, the partograph did not show signs of prolonged labour, although caesarean section was performed for that indication. Seventeen (12.9%) of the unsupported indications were foetal distress, 11 (8.3%) foetal malpresentation, and 11 (8.3%) two previous scars.

Maternal and perinatal outcome before moment of discharge, usually 3 days postpartum, remained relatively stable over time, although varying slightly per quarter. In the study period, four maternal deaths (0.6%), 15 fresh stillbirths (2.3%), five macerated stillbirths (0.8%), and six early neonatal deaths (0.9%) occurred in women undergoing caesarean section. Supplement 1 shows maternal and perinatal outcomes of caesarean section compared to vaginal birth. Stillbirth and neonatal death rates did not change significantly over time (p = 0.455 and 0.292 respectively, see Supplement 1). Plotting stillbirth and early neonatal death rates with caesarean section percentages (Fig. 6), no significant trend could be identified (p = 0.203) suggesting the decreasing caesarean section rate had no impact on these rates.

Our data suggest that the introduction of an MDGHTM could have had an effect on the caesarean section rate in the facility and on the proportion of caesarean sections not supported by documentation in medical records. There was a declining trend in both these rates over the 2-year study period. Introduction of an MDGHTM took place in the first quarter of 2015, and from the first quarter of 2016, there were two MDGHTMs working regularly in maternity care in the hospital. Additionally, from halfway the third quarter of 2016, there was one MDGHTM covering the maternity department daily. No other major changes occurred in the study period. Both caesarean section rate and proportion of unsupported indications decreased gradually. While medical officers took up some of the caesarean sections and thereby might have had an influence on the indications, remarkably the number of unsupported indications by associate clinicians decreased significantly as well. This is most likely attributed to supervision, training, and audit taking place throughout the study period, which were all part of the job description of the MDGHTMs. As part of hospital management, they fulfil a role as supervisor and head of the clinical team. In morning handovers, caesarean section indications were discussed and labour management was also part of maternal and perinatal morbidity and mortality audits. The effect on the overall caesarean rate could be partly attributed to this leadership role, but since the rate mostly decreased in the last two quarters, it could also be contributed to having an MDGHTM continuously in the maternity department to aid and supervise the midwifery team and thereby having a significant role in preventing unnecessary caesarean sections. Which element contributed most to the decline warrants further assessment.

A significant number of auditable cases (71, 14.3%) could not be audited due to lack of information. Indications for these cases might as well be considered unsupported, since no supporting information was found. Additionally, in 8.4% of all cases, caesarean section was performed with unknown indication. Therefore, our numbers of unsupported indication could be underestimated and stress the importance of evidence-based clinical decision-making to an even greater extent. Also, some criteria in national protocols give room to perform caesarean section at a rather low threshold. For example, a one-time foetal tachycardia can be interpreted as an opportunity to decide for caesarean section.

Reducing the number of unsupported, and possibly unnecessary, caesarean sections is important. It may save women from unnecessary risks of abdominal surgery. Furthermore, considering that a great number of procedures was done on young nulliparous women, preventing unnecessary procedures could prevent serious complications in subsequent pregnancies, especially considering the high fertility rate in Malawi. It seems a role exists for MDGHTMs, together with associate clinicians, to discuss indications for caesarean section and work towards optimal case selection. While this is a single-centre study, results are comparable with other caesarean section audits in similar settings, for example in Tanzania [24, 25]. However, the effect of MDGHTM was not studied previously. More research could provide better insights in this role, for example a qualitative study of health workers’ perceptions in settings where MDGHTMs are employed, preferably accompanied by quantitative and qualitative data on decision-making in caesarean sections.

A surprising result was the high number of caesarean sections performed in preterm labour, especially in the range of 35 to 37 weeks (17.1%). Since there is no policy for determination of gestational age by ultrasound in Malawi, the first day of the last menstrual period is used to calculate gestational age. This might not always be reliable and can for example be influenced by depot contraceptive usage or long-term breastfeeding after previous pregnancies. In a significant majority, gestational age was completely unknown, but the majority of birthweights (76.5%) was above 2500 g, suggesting that the number of very preterm births by caesarean is limited.

A major limitation of the study was lack of documentation. Inclusion was done based on complete partographs. While case notes with complete unfilled partographs were not included, it sometimes still remains unclear whether a partograph contained all information supporting the decision for caesarean section. For example, a last foetal heart rate or vaginal examination prior to the procedure on which the decision of foetal distress was made, may have been forgotten to be documented, leading to cases being labelled as unknown. Furthermore, there may be underreporting of cases due to missing case files or partographs, which we believe would be on a random basis, but still may influence the calculated caesarean section rates. Poor documentation during labour not only provides a challenge for data analysis, it also burdens the clinician making the decision for caesarean section. If little is known about progress or foetal condition, one may decide more quickly to proceed to caesarean delivery. Proper and regular documentation assists clinicians in making correct decisions and thereby also attributes to prevent unnecessary caesareans. This is also the case with documentation of complications, which seem quite low based on our collected data. Complications can be missed due to lack of a structured patient file administration, whereby readmissions were not connected to previous admissions.

We cannot be certain that other variables in the hospital stayed the same during the study period, although our analysis was based on that assumption. Staff or policy changes may influence the decision-making process. Over the study period, there were, however, no major changes in management, protocols, or resource availability. Since the study setting is a single rural facility, small numbers or individuals could have influence on certain changes over time. The increase in caesarean section rates seen in the second and third quarters of 2015 remains unexplained, but might be influenced by staff changes. It is therefore also challenging to predict the sustainability of the observed effect. Supervision and training may have long-term effects, especially with the enrolment of surgical and obstetric training for associate clinicians [22, 23, 26]. Effectivity of task-sharing with associate clinicians has been studied previously, but not on the matter of indications and caesarean section rate [22, 27].

Ideally, interventions require to have a sustainable impact on quality of care. While most likely the tradition of Dutch MDGHTM in St. Luke’s Hospital will continue for the foreseeable future, we hope that there is also a sustainable effect on clinical officers. We observed that senior clinical officers took the discussion of caesarean section indications very serious and this became a standard aspect of the morning handover meeting. With the start of a bachelor training programme for clinical officers in obstetrics and gynaecology, there will be more knowledge and experience on the obstetric level available, hopefully contributing further towards open discussions and thorough decision-making in labour. Alternatively, this could also be an opportunity for a Malawian trained medical officer.

Our study demonstrates the effect of MDGHTMs on caesarean section rate, as well as on the quality of caesarean section indications. Supervision, stimulating discussion, and audit are major parts of the work of MDGHTMs. Their background training in obstetrics is of benefit. We believe that continuation of training MDGHTMs in order to support health systems in low-income settings is important and more research into their effectivity is helpful.