Induction Of Labour: The production of an integrated series of systematic reviews

Article type
Authors
Kelly A, Kavanagh J, Thomas J, Alfirevic Z, Hofmeyr J, Neilson J
Abstract
Background: During the early development of an evidence-based guideline on induction of labour by the RCOG, a collaboration with the Cochrane Pregnancy and Childbirth Review Group was formed to allow the present evidence base relating to induction of labour to be updated. Clinical guidelines have been defined as: 'systematically developed statements which assist clinicians and patients in making decisions about appropriate treatment for specific conditions'. Suggested criteria for reviewing guideline methodology include; systematic review, multidisciplinary development groups, and explicit links between the strength of evidence and grade of recommendation. Complex interventions present a challenge for the production of systematic reviews and for guideline development. We outline here the model we developed for an integrated system of systematic reviews regarding all agents used for a complex intervention, induction of labour.

Methods: A core working group was assembled to decide on the population to be studied, outcome measures to be used, agents to be considered and the methodology needed to allow integration of the reviews. An homogenous population was selected and a core set of primary and secondary outcomes were agreed derived from a set of outcomes employed by two of the authors (ZA & JH) in recent published reviews on induction of labour. The commonality of these factors in all the reviews will allow data from individual reviews to be compared. A generic protocol was developed for the entire series of reviews and has been published in the Cochrane library. A two tier system of reviews was planned. The first tier will include a series of 23 reviews focusing on induction using a specific intervention or agent. These have been arranged into a hierarchy. The order within the hierarchy was decided by the group according to clinical importance and presumed volume of trials relating to that agent or intervention. To prevent duplication of work, agents were only compared to those interventions above in the hierarchy. E.g. intracervical prostaglandins' were third on our hierarchy and hence would only be compared to other intracervical prostaglandins as well as vaginal prostaglandins and placebo. These primary reviews will inform a series of six secondary reviews which will be clinical question-led. To allow data from the primary reviews to be incorporated into the secondary reviews without the need to return to the original trial data, outcome data will be reported in a number of structured comparison groups. These comparison groups forming the basis of the structure of the secondary reviews. To allow the process to progress within the timetable for the RCOG guideline, the data extraction process was centralised. An eligibility exercise was undertaken by two of the authors (AJK & JK), and agreement was assessed by a third party (JT). A standardised data extraction form was developed and piloted. Trial quality will be assessed in relation to selection, performance and attrition bias. The standard quality assessment suggested in the current Cochrane handbook has been expanded to allow an in-depth sensitivity analysis to be conducted. All outcomes will be presented in dichotomous format using risk ratios and a fixed effects model as the common pooling method. Data from 700 clinical trials have been extracted and forwarded to a group of experienced reviewers. Agreement on inclusion/exclusions and data inconsistencies will be reached by discussion. The secondary protocols are in development.

Conclusions: We aim to produce a system of structured and integrated systematic reviews, which will allow clinicians to make an informed clinical judgement regarding the method of induction for women requiring induction of labour. The model may be useful to other Review Groups with similar challenges.