Developing criteria for Cesarean Section using the RAND appropriateness method

First we searched the literature to identify available clinical practice guidelines and evidence summaries. We identified fourteen documents with relevant scopes [25‐38]. We screened these documents and selected three guidelines and evidence summaries that provided a relatively comprehensive coverage of issues relevant to cesarean section [34‐36]. The three selected documents had been developed as part of an established guideline development program or by recognized institutions. Then one author (AR) conducted a short workshop on development and appraisal of clinical guidelines. Three panel members and three authors used the validated Farsi translation of the AGREE (Appraisal of Guidelines for Evaluation and Research) tool (Additional file 1: Table S1) [39, 40] to appraise the guidelines with comprehensive coverage and selected two clinical guidelines as a result [34, 35]. In the next step, two obstetricians & gynecologists (BHR, SH), one midwife (FH) and a health service researcher (RO) extracted potential scenarios about the cesarean section from the selected clinical guidelines. We aimed to develop as short a list of scenarios as possible which were 'comprehensive', 'mutually exclusive' and 'homogeneous' [21]. Each scenario comprised of a few words or a short sentence capturing the main clinical features that identified the patients. The scenarios were categorized within main potential indications for cesarean section for further analyses and tabulation purposes.

The panel members comprised of twelve individuals: nine obstetricians & gynecologists and three midwives. The panel members were from the Ministry of Health and Medical Education (one), four different public medical universities (nine), a non-profit medical school (one), and a private hospital (one). One participant exclusively worked in the private sector, and one exclusively in the public sector. The rest worked both in public and private hospitals.

Out of twelve invited members, ten responded to the first round. Those ten were then invited to participate in the second round, out of which nine participated.

In the first round we sent the list of scenarios along with a summary of the clinical guidelines, the scoring system and the definitions to the panel members. The panel members were asked to give their opinions about each scenario for cesarean section on a scale ranging from one (totally inappropriate) to nine (totally appropriate). They were asked to consult the scientific resources provided for them while giving their opinions. The purpose of the literature review and the provision of evidence-based clinical guidelines and evidence summaries to the panel members was to provide them with an up-to-date summary of the best available evidence about the indications for cesarean section. The panels were not limited to the guidelines as they were required to use their own professional judgments as well as the evidence presented to them.

We then collated the views of the panel members and summarized the views in a format suitable for feedback so that each member received a summary of the panel view as well as a reminder of the scores that the member had assigned to each scenario. Then the panel members were invited to a second round, (face-to-face panel meeting), to view the feedbacks, and review and discuss their opinions. Nine members attended the second round (one day meeting from 8.00 to 17.00) and all the scenarios were reviewed and discussed. The panel members decided to add a limited number of further scenarios at this stage. Final decisions were recorded in specific forms.

As recommended by the RAM, we asked the panel members to rate the appropriateness of each indication (score them) based on their own professional judgment instead of what they perceived to be the views of other respected clinicians while considering the possible outcomes resulting from their decisions. The panel members were asked to consider average pregnant women presented to average clinicians in usual settings of care relevant to each indication discussed [21].

We used the median scores for reporting the results of the first panel to the second panel. The scores were divided into three groups: appropriate (score = 7-9), equivocal (score = 4-6), and inappropriate (score = 1-3). If the median score fell into any of the above groups, it was considered as such (e.g. if a scenario's median score was 8, the scenario was considered as 'appropriate'). A further condition had to be met in order to reach agreement: if the minimum and maximum scores were ignored, all other scores must fall in the same scoring group. Additionally, we compared the rate of agreement between the two panels using weighed Kappa values and frequency charts.

Tan et al [43] have described the complexities and limitations of using RAM. Many of their criticisms equally apply to other consensus development methods, such as Delphi [45]. It should be noted that consensus methods are most useful where there are disagreements or variation in practice, and reliable evidence is limited. In these circumstances formal consensus methods are valuable and their use is inevitable.

There were heated discussions in our panel on whether it is indicated to conduct cesarean sections for repeat cesarean patients. Based on their personal experiences, members believed that the evidence on the benefits of trial of labor over repeat cesarean is dependent on a context with good quality pre-hospital and hospital care and may not apply to Iran's conditions. This was in contrast with the results of a systematic review of relevant evidence [46]. It concluded that repeat cesarean was not indicated as a routine practice, mostly based on non-randomized trials originating from high resource countries [46]. In our opinion, while high quality evidence on the issue is lacking, the panel members may have been somehow justified to doubt the benefits of trial of labor over previous cesarean section. A qualitative study in Canada reported that clinicians had similar concerns about trial of labour over previous caesarean section [16]. Further research originating from low and middle income countries is required.

We spent time on familiarizing the panel members with the method and attracting their valued cooperation. Membership of the panel involved open discussions of personal views and practices, and that certain practices might not be supported by evidence or by other panel members. We also selected the members from different backgrounds and settings to improve comprehensiveness of the views [47, 48]. As an advantage, we used the AGREE tool for selection of the evidence sources [39]. The AGREE tool has been used extensively for appraising clinical practice guidelines for different conditions including obstetrics care [49]. Using the AGREE tool provided a chance for the panel to reach a shared understanding of the evidence before embarking towards consensus building.

RAM usually results in a long list of scenarios [48]. To ease the use of its results it may be possible to develop user-friendly software, or to categorize the scenarios into indications and packages in the format of clinical guidelines. The RAM guidelines suggest that on average 150-200 scenarios can be rated in an hour once the panel members are used to the process [21]. In our experience a dedicated team of panel members reviewed and rated 294 scenarios in one long-day second-round meeting.

We used the weighted kappa values to compare the results of the two rounds of the study. It should be noted that the interpretation of kappa values here differs from other agreement studies. In a consensus development study, the investigators seek to improve agreement via changes in the views of the panel members. Hence high values of kappa are not sought. Still its measurement is useful as it quantifies how the views have changed during the study.

Our results will help decision makers in identifying misconceptions on the benefits of cesarean section and focusing their efforts on changing the views of the clinicians. For example the participants agreed that cesarean section was appropriate for many repeat cesarean and cephalic-pelvic disproportion scenarios, despite recent evidence and the recommendations of some evidence-based guidelines against use of section in these scenarios [34, 35, 46].

While our study focused on developing criteria for a wider use in the country, our approach may also benefit obstetricians and midwives working in hospitals to develop corporate strategies. It will require methodological support and group work, while it will help in generating shared views and understanding. In a way, the process will be similar to 'participatory guideline development' that has been shown to be effective in changing professional practice [50].

Other variables such as payment method, medico-legal issues and patient preferences may affect provider practice and views on conducting cesarean section [16]. According to previous studies different factors cause high cesarean rates in Iran. They include factors that affect women preference (e.g. increasing women level of education, employment and age at marriage and decreasing intended number of deliveries) [18, 51‐53], provider behavior and clinical factors (e.g. repeat cesarean, dystocia, CPD and physician preference) [51‐53] and health system factors (e.g. health insurance coverage, delivery at private hospital) [51].

In Iran, as in many low and middle income countries, health system regulatory mechanisms are insufficient, and the fee-for-service payment provides further income if cesarean section is performed. It is also generally perceived that women prefer cesarean section over vaginal birth. In such context it is hardly surprising to see the current high rates of cesarean section. Also there is a growing culture of suing doctors because of malpractice claims, and this may fuel cesarean section rates as a form of 'defensive' medical practice. Hence our panel members may have been more lenient towards section (e.g. advocating repeat cesarean) than it might be observed in other countries.

The results of this study can contribute to the development of national guidelines for use in the country. Obviously the impact of implementing such guidelines will depend on many factors, including using effective implementation strategies [44]. The guidelines more likely affect provider and health system factors. They may also help clinicians in effective communication with pregnant women when they request cesarean section.

For certain scenarios, however, agreements on appropriateness may not result in reducing variation in practice. For example it may be easy to agree with cephalic-pelvic disproportion scenarios, but these are difficult to measure and implement in practice.