169 6 Cochrane review on integrated disease management for COPD 1.1.1. Subgroup analysis based on setting Six studies reporting on SGRQ total score were performed in primary care (Boxall 2005; Fernandez 2009; Gottlieb 2011; Kruis 2014; Wood-Baker 2006; Zwar 2016), nine studies in secondary care (Bourbeau 2003; Engstrom 1999; Fan 2012; Jimenez-Reguera 2020; Kalter-Leibovici 2018; Rice 2010; Rose 2017; Titova 2017; Wakabayashi 2011), and three studies in tertiary care (Ko 2016; Vasilopoulou 2017; Wang 2017). A test for subgroup di erences showed a statistically signi cant di erence between subgroups (P = 0.001). Studies performed in primary and secondary care showed no statistically signi cant di erences between IDM and control, and pooling of tertiary care studies showed a clinically and statistically signi cant improvement in favour of IDM (MD -14.58, 95% CI -21.56 to -7.61; Analysis 1.4). However, pooling indicated considerable heterogeneity for all three subgroups. Hence, results of the subgroup analysis should be interpreted with caution. 1.1.2. Subgroup analysis based on study design We performed subgroup analysis based on study design and compared RCTs (total 2865 participants) with cluster-RCTs (total 1420 participants) (Analysis 1.5). Tests for di erences showed a statistically signi cant di erence between both groups. Heterogeneity within the RCT remained considerable (I² = 83%). 1.1.3. Subgroup analysis based on dominant component of the programme Two studies (total 294 participants) included individualised education as the dominant component (Fan 2012; Wakabayashi 2011), ve studies (total 1825 participants) included self-management as the dominant component (Bourbeau 2003; JimenezReguera 2020; Kruis 2014; Rice 2010; Wood-Baker 2006), four studies (total 175 participants) included exercise as the dominant component (Boxall 2005; Engstrom 1999; Fernandez 2009; Gottlieb 2011), and ve studies (total 1610 participants) included structural follow-up as dominant component (Kalter- Leibovici 2018; Ko 2016; Rose 2017; Titova 2017; Vasilopoulou 2017). Post hoc, we identi ed telemonitoring as an important dominant component in two studies (Vasilopoulou 2017; Wang 2017). Tests for subgroup di erences showed a statistically signi cant result (Chi² = 17.89, df = 4, P = 0.001) indicating di erences in e ect between subgroups based on the dominant component. A statistically signi cant di erence was found only in the group with telemonitoring as the dominant component (MD -18.33, 95% CI -26.72 to -9.94) (Analysis 1.6). However, the subgroup included only two studies. Also, heterogeneity remained moderate within subgroups. Hence, results should be interpreted with caution. 1.1.4. Subgroup analysis based on region of study Four studies (total 1147 participants) were performed in North America (Bourbeau 2003; Fan 2012; Rice 2010; Rose 2017), four in Northwestern Europe (total 1286 participants) (Engstrom 1999; Gottlieb 2011; Kruis 2014; Titova 2017), three in Southern Europe (total 227 participants) (Fernandez 2009; Jimenez-Reguera 2020; Vasilopoulou 2017), three in Oceania (total 380 participants) (Boxall 2005; Wood-Baker 2006; Zwar 2016), three in East Asia (total 385 participants) (Ko 2016; Wakabayashi 2011; Wang
RkJQdWJsaXNoZXIy MTk4NDMw