cropped-NIHR-Applied-research-logo_S_london_CMYK_editable.png

6th UK Implementation Science Research Conference

  • Programme
  • Plenary Lectures
  • Poster Presentations
  • Oral Presentations
  • Meet the Experts
  • Panelists
  • Organisation Team
Menu
  • Programme
  • Plenary Lectures
  • Poster Presentations
  • Oral Presentations
  • Meet the Experts
  • Panelists
  • Organisation Team

Maximizing knowledge from systematic reviews of complex interventions

O3

PRESENter

Kristin J Konnyu
IMG_9782

authors

Kristin J Konnyu, Jeremy M Grimshaw, Noah M Ivers, Thomas Trikalinos

Biography

Dr. Konnyu is an epidemiologist and implementation scientist. She is an Assistant Professor in the Center for Evidence Synthesis in Health in the Department of Health Services, Policy and Practice in Brown University’s School of Public Health. Dr. Konnyu’s diverse training across a range of disciplines (biological sciences, rehabilitation sciences, education, epidemiology) provides her with a multidisciplinary perspective that she brings to her public health work. Areas of recent interest have included understanding the barriers and facilitators to citizen’s willingness to receive a COVID-19 vaccine and designing interventions to address these factors to promote vaccine uptake, as well as novel methods to synthesize complex interventions evaluated in diverse clinical settings (e.g., diabetes quality improvement, rehabilitation following major joint replacement, and substance use disorder in adolescents).

background

Well-conducted randomized controlled trials (RCTs) are the gold standard for estimating intervention effects, and systematic reviews (SRs) of trial evidence are cornerstone to informing evidence-based practice, policy, and research. However, understanding the effects of complex interventions using standard SR approaches is challenging given the diversity of intervention content, delivered in diverse ways, evaluated in diverse designs using diverse outcomes. We describe methodological adaptations to standard review processes to enhance the informativeness of complex interventions SRs.

MEthod

The adaptations described are drawn from experiences in conducting 3 large SRs over the past 10 years.

results

Question formulation – we adopted a modest and multivariable approach to inference. We assume true causal inference is not viable nor appropriate, but principled learning about associations between factors of interest and outcomes may be feasible. Data collection – contacting authors for additional details about interventions is feasible to supplement trial reports and authors are twice as likely to respond to requests if contacted by telephone vs email (1). Constructing a posterior distribution of intracluster correlation coefficients (ICC) is feasible and offers a principled approach to imputing missing ICCs among cluster RCTs that fail to account for unit of analysis errors (2). Data extraction – we have operationalized standardized taxonomies to code intervention content (3) to ensure robust (i.e., clinically or theoretically meaningful) coding. Data synthesis – we have found multivariable meta-regression models offer a feasible and informative approach to estimating the association between factors of interest and outcomes (4). Data reporting – we have adopted transparent reporting of our methods of data collection, manipulation, imputation, and analysis to complement the interpretation of our findings. We suggest complex interventions reviews are optimally suited to a living review framework (5).

Conclusion

Methodological adaptations of standard approaches may help enhance the informativeness of complex intervention SRs.