Article type
Abstract
Background: Given the diversity of nonpharmacological interventions, it is essential to characterize them to understand not only ‘what’ intervention is the most effective, but ‘how’ it is provided to achieve the greatest benefit. In a 2018 survey, only 8.1% of network meta-analyses (NMAs) provided clear rationales for characterizing nonpharmacological intervention nodes. In response to this gap, we developed the TIP (Theme, Intensity, and Provider/Platform) framework as a uniform approach for characterizing nonpharmacological interventions.
Objective: To compare NMA effect estimates of nonpharmacological interventions using the TIP framework with those based solely on the types of interventions.
Methods: We selected two published NMAs that used the TIP framework: one focused on psychosocial interventions in adults with harmful use of alcohol; and the other on music interventions in older adults. We calculated NMA effect estimates, including the mean difference (MD) of the Alcohol Use Disorder Identification Test (AUDIT) score, and the standardized mean difference (SMD) of depression measures. The treatment options were recharacterized using the lumping approach based on intervention types. Subsequently, we compared NMA effect estimates and 95% confidence intervals (CI) between these approaches, where lower values for both measures indicate higher effectiveness.
Results: The NMAs for psychosocial interventions and the music interventions comprised 19 (n=7,149) and 15 (n=1,144) randomized-controlled trials, respectively. When compared with usual care, the lowest MD of the AUDIT score derived from the TIP NMA was 1.37 lower than the lumping NMA, with 0.90 narrower 95%CI (-3.61[-5.22, -1.99] vs -4.98[ -7.04, -2.91]). A similar pattern was observed in the music intervention comparison as the SMD of the TIP NMA was 2.29 lower than that of the lumping NMA, with 0.27 narrower 95%CI (-3.00[-3.64, -2.35] vs -0.71[-1.22, -0.20]). Some TIP features exerted significant effect estimates with narrower confidence intervals, whereas their lumping counterparts showed no effects (Table 1).
Conclusions: Our findings reveal a higher magnitude of effect estimates with less variation than lumping nonpharmacological interventions based solely on intervention types, suggesting the uniform effects of the interventions characterized by the TIP framework. We recommend using the TIP framework as it appears to better characterize nonpharmacological interventions for NMA.
Objective: To compare NMA effect estimates of nonpharmacological interventions using the TIP framework with those based solely on the types of interventions.
Methods: We selected two published NMAs that used the TIP framework: one focused on psychosocial interventions in adults with harmful use of alcohol; and the other on music interventions in older adults. We calculated NMA effect estimates, including the mean difference (MD) of the Alcohol Use Disorder Identification Test (AUDIT) score, and the standardized mean difference (SMD) of depression measures. The treatment options were recharacterized using the lumping approach based on intervention types. Subsequently, we compared NMA effect estimates and 95% confidence intervals (CI) between these approaches, where lower values for both measures indicate higher effectiveness.
Results: The NMAs for psychosocial interventions and the music interventions comprised 19 (n=7,149) and 15 (n=1,144) randomized-controlled trials, respectively. When compared with usual care, the lowest MD of the AUDIT score derived from the TIP NMA was 1.37 lower than the lumping NMA, with 0.90 narrower 95%CI (-3.61[-5.22, -1.99] vs -4.98[ -7.04, -2.91]). A similar pattern was observed in the music intervention comparison as the SMD of the TIP NMA was 2.29 lower than that of the lumping NMA, with 0.27 narrower 95%CI (-3.00[-3.64, -2.35] vs -0.71[-1.22, -0.20]). Some TIP features exerted significant effect estimates with narrower confidence intervals, whereas their lumping counterparts showed no effects (Table 1).
Conclusions: Our findings reveal a higher magnitude of effect estimates with less variation than lumping nonpharmacological interventions based solely on intervention types, suggesting the uniform effects of the interventions characterized by the TIP framework. We recommend using the TIP framework as it appears to better characterize nonpharmacological interventions for NMA.