How hard is to do an overview of Cochrane Reviews? Could CORE outcomes help?

Article type
Authors
Bofill Rodriguez M1, Jordan V1, Lethaby A1, Wise M1, Farquhar C1
1Department of Obstetrics and Gynaecology, University of Auckland
Abstract
Over seven thousand Cochrane Reviews have been published since 1995. The first Cochrane overview was published in 1998. There are currently approximately fifty overviews, (fewer than 10 include a network meta-analysis) and 23 overview protocols.

What is the value of an overview? In terms of interventions, a broad systematic review covering all treatment options is a valuable tool for knowledge translation for both healthcare providers and consumers.

We are currently writing an overview for interventions for heavy menstrual bleeding, including a network meta-analysis. The overview protocol was published in November 2018. Our search found nine Cochrane Reviews addressing this topic. Six were outdated, one was a stable review and the other five were last updated between seven and 12 years ago. We updated five reviews, adding 20 trials. Overall there are 102 studies with 23 comparisons. Surprisingly, despite the abundance of evidence we will only be able to reach very limited conclusions. This is due to the variation in the way outcomes were reported in the trials and in the different cut-off periods they reported.

The following examples show how this issue makes it difficult to combine results:
- menstrual bleeding was reported in over 10 different ways and reported with many different time frames. (See Tables 1 and 2 for further details);
- quality of life was also reported in diverse time periods, but more importantly it was reported using nine different scales;
- 90 different adverse effects were reported. Headache, the most common medical adverse effect reported, was present in only 23 trials. Here again, perforation, the most commonly reported surgical adverse effect, was present in only 12 of more than 40 surgical trials;
- the requirement of further surgery, in over 40 surgical trials, was reported either as any gynaecological surgery or as any gynaecological surgery for treatment failure, not always specifying if it was a hysterectomy or an ablation procedure.

This shows how outcomes and measurement tools are used in diverse ways, thus hindering the ability to combine results, and make conclusions. Any network meta-analysis produced within overviews is limited by the heterogeneous way outcomes are reported. This highlights the urgent need for CORE outcomes for heavy menstrual bleeding. CORE outcomes is an agreed standardized minimum set of outcomes or outcome measures for a specific research area. This is likely to be similar to other research areas where having CORE outcomes will improve the quality of overviews.

Patient or healthcare consumer involvement: overviews of systematic reviews should be encouraged, as they are more useful and friendly for patients and consumers as they answer broader questions.