WORK PACKAGE 5

Comparative analysis: Network meta-analysis

Main objectives:

  • To develop statistical models to disentangle the effect of components of self-management interventions (SMIs) as complex interventions
  • To compare SMIs using network meta-analysis (NMA), calculate the relative treatment effect for all pairwise comparisons and rank SMIs for each outcome (from ‘most-effective’ to ‘least effective’)
  • To evaluate the NMA and determine the confidence in the output of the analysis both overall and separately for each comparison
  • To identify the most effective SMIs under various circumstances (overall, by age group, gender, etc.) for each outcome and combinations of outcomes

Approach:

As self-management interventions (SMIs) are considered as complex interventions, we first develop statistical models that disentangle the effects of the interacting components of a SMI. We identify the hierarchy of components and the mechanism under which a SMI operates on the outcome through expert opinion. In NMA protocols (one for each of the four chronic conditions), we assess eligibility of interventions and studies for including them in the network of comparison. We liaise with clinicians and experts on SMIs about possible lumping of interventions as it is expected that SMIs have multiple interacting components and differ substantially across studies. After programming the statistical model, we identify and define a ranking of interventions according to their level of effectiveness. Subgroup and meta-regression analyses are conducted where data are available. Finally, we prepare the integration of the results into the decision-making tools, so that users can see which intervention works best, but also how much better it works than another intervention.

Leading partner: University of Ioannina (UOI), Greece

Other partners involved:

  • Fundación Avedis Donabedian (FAD), Spain
  • Netherlands Institute for Health Services Research (Nivel), The Netherlands
  • Fundació Privada Institut de Recerca de l’Hospital de la Santa Creu i Sant Pau (IR-HSCSP), Spain