177 6 Cochrane review on integrated disease management for COPD 3.2 Respiratory-related admissions - short-term We pooled data from three studies with 377 patients measuring respiratory-related admissions until 6 months’ follow- up (Bernocchi 2017; Ko 2009; Trappenburg 2011). There were no statistically signi cant di erences in the risk of respiratory-related hospital admissions in the short term (OR 0.60; 95% CI 0.30 to 1.22). Studies were homogeneous, but the number of events was too small (ranging from 1 to 11) to allow rm conclusions based on the data. 3.3. Respiratory-related admissions - medium-term Nine studies with a total of 2449 participants reported on the number of patients with at least one respiratory-related admission at 6 to 15 months’ follow-up (Bourbeau 2003; Fan 2012; Lenferink 2019; Rea 2004; Rice 2010; Sanchez-Nieto 2016; Silver 2017; Smith 1999; Vasilopoulou 2017). Pooled estimates showed a statistically signi cant reduction in admissions in favour of IDM (OR 0.60, 95%CI 0.44 to 0.81). Data showed considerable heterogeneity (I² = 57%) (Analysis 1.18). Sensitivity analysis of only highquality studies showed similar results, with only a small reduction in heterogeneity (I² = 48%) (see Table 7). To further explore the reasons for heterogeneity, we performed three subgroup analyses. 3.3.1. Subgroup analysis based on setting Heterogeneity remained substantial or considerable when we pooled all studies in which the intervention was delivered in a primary care setting (I² = 84%) and secondary or tertiary care settings combined (I² = 48%). A test for subgroup di erences showed no di erences between groups (Chi² = 0.38, df = 1, P = 0.54). In other words, there seems to be no convincing di erence between primary care and secondary or tertiary care that can explain the observed heterogeneity (Analysis 1.19). 3.3.2. Subgroup analysis based on dominant component of the programme In ve studies with a total of 1353 participants, the dominant component was selfmanagement (Bourbeau 2003; Lenferink 2019; Rea 2004; Rice 2010; Sanchez-Nieto 2016). Two studies included education (Fan 2012; Silver 2017), two studies structural follow-up (Smith 1999; Vasilopoulou 2017), and one study telemonitoring as the dominant intervention component (Vasilopoulou 2017). A test for subgroup di erence showed no di erences between groups (Chi² = 3.65, df = 3, P = 0.30). However, these results should be interpreted carefully, as only the self- management subgroup pooled more than two studies, while the other subgroups pooled two or fewer studies. Among studies with self-management as the dominant component, the e ect on respiratoryrelated admissions favoured IDM ((OR 0.55, 95% CI 0.43 to 0.71; I² = 0%) (Analysis 1.20). 3.3.3. Subgroup analysis based on region Four of the nine studies, with a total of 1788 participants, originated in North America (Bourbeau 2003; Fan 2012; Rice 2010; Silver 2017), two studies in Southern Europe (Sanchez-Nieto 2016; Vasilopoulou 2017), one study in Northwestern Europe (Lenferink 2019), and one study in Oceania (Smith 1999). The e ect estimate di ered
RkJQdWJsaXNoZXIy MTk4NDMw