Approaches to optimise review research recommendations when evidence is very limited: going beyond “high quality studies are needed”

Session: 

Oral session: Innovative solutions to challenges of evidence production (4)

Date: 

Tuesday 18 September 2018 - 14:20 to 14:30

Location: 

Carrick 1

All authors in correct order:

Corbett M1, South E1, Marshall D1, Woolacott N1
1 Centre for Reviews and Dissemination, University of York, UK
Presenting author and contact person

Presenting author:

Mark Corbett

Contact person:

Mark Corbett
Abstract text
Background: When randomised controlled trial (RCT) evidence is unavailable, systematic reviews cannot normally draw firm conclusions regarding practice. A key output from such reviews should therefore be clear recommendations for future research. Unfortunately, these are often brief, lacking the detail needed by research commissioners. We contend that in areas where very limited evidence is anticipated, reviews of effectiveness studies alone may not be sufficient to enable reviewers to make optimal research recommendations.
Methods: Across two reviews where limited evidence was expected - one of extravasation injury treatments, and one of neuromodulation for phantom limb pain (PLP) - we prospectively undertook additional research to inform and optimise review research recommendations. This included: extracting data on research ideas; surveying NHS clinicians about current practice and future research; reviewing epidemiological evidence; reviewing studies in related areas; and allocating sufficient review time to explore study designs deemed to be useful when recruitment is challenging.
Results: Both clinician surveys provided clear indications about the specific treatments where further research was thought most worthwhile. Most clinicians thought RCTs could be successfully undertaken, those who did not provided reasons that were informative when researching alternative study designs. A review of chronic PLP epidemiology identified evidence on important prognostic factors, and prevalence data, useful for informing eligibility criteria, and sample size feasibility, respectively. Consideration of studies in related clinical areas identified two examples of RCTs - one in each review - where recruitment was problematic (e.g. 37 patients recruited in 7 years). Additional studies also revealed potentially useful outcome measures, lacking in the included effectiveness studies. The alternative, less well-known study designs we identified and then used in our specific research recommendations included randomised registry trials, sequential trials and N-of-1 trials.
Conclusions: Use of complementary research such as surveys and reviews of evidence other than effectiveness can enhance review research recommendations. Better recommendations should maximise the probability of future studies reaching completion.
Patient involvement: PPI used for questionnaire content.

Relevance to patients and consumers: 

In areas of research where evidence is anticipated to be limited, use of the complementary research approaches we have outlined should ensure that systematic review recommendations for primary research studies are more informed and focussed. This will benefit patients by optimising future primary study results (and their relevance to clinical practice), thereby minimising the risks of wasting patient time and research resources (on studies with inappropriate designs). The approaches we advocate may be particularly beneficial in research areas where patient populations are small and where research funding is quite scarce.