179 6 Cochrane review on integrated disease management for COPD Wetering 2010). Numbers of events and total numbers are lower for Lou 2015, as we reduced the size of the study to its ‘e ective sample size’ to adjust for clustering e ects. Pooled meta-analysis showed no signi cant di erences between groups (OR 0.72, 95% CI 0.45 to 1.16). Pooled results showed considerable heterogeneity (I² = 75%) and di erences in direction of e ect. Although Lou 2015 and van Wetering 2010 showed positive e ects in favour of IDM, Kalter-Leibovici 2018 and Sridhar 2008 showed no statistically signi cant di erences. The di erent ndings could have resulted from variation in follow-up duration which ranged from 24 months in Sridhar 2008 and van Wetering 2010 to 36 months in Kalter-Leibovici 2018 to 48 months in Lou 2015. Finally, heterogeneity could be explained by the large di erences in study size ranging from 104 participants in Sridhar 2008 to 6221 participants (435 e ective sample size) in Lou 2015. Sensitivity analysis including only high-quality studies did not show a statistically signi cant e ect (OR 0.88, 95% CI 0.61 to 1.27; I² = 38%). 3.8. Hospital days per patient We were able to pool 14 studies that reported on the number of hospital days among those (3563 participants) hospitalised during the study. Pooling showed an overall reduction of 2.27 days spent in the hospital in favour of IDM; this nding was statistically signi cant (MD -2.27, 95% CI -3.98 to -0.56; I² =7 8%) (see Figure 6). Figure 6. Forest plot of comparison: 1 Integrated disease management versus control, update, outcome: 1.24 Hospital days per patient (all causes).
RkJQdWJsaXNoZXIy MTk4NDMw