182 6 Chapter 6 one course of antibiotics (Littlejohns 1991; Rea 2004; Sanchez-Nieto 2016). The number of patients using at least one course of antibiotics was not statistically di erent between groups (OR 1.46, 95% CI 0.51 to 4.18; I² = 53%; Analysis 1.27). A sensitivity analysis of high-quality studies showed decreased heterogeneity (I² = 53%) and signi cantly increased risk when a course of antibiotics was received by people in the IDM group (OR 2.35, 95% CI 1.02 to 5.42). Further exploration of these studies revealed that they provided the same follow-up (12 months) but represented very di erent settings, as Rea 2004 was a cluster-randomised trial in a primary care setting, and Littlejohns 1991 and Sanchez-Nieto 2016 were RCTs conducted in a secondary care setting. Secondary outcomes 4. Dyspnoea Fifteen studies reported on modi ed MRC Dyspnoea Scale scores as an outcome for dyspnoea (Bernocchi 2017; Gottlieb 2011; Kalter-Leibovici 2018; Khan 2019; Ko 2016; Kruis 2014; Lenferink 2019; Lou 2015; Mendes 2010; Öztürk 2020; van Wetering 2010; Vasilopoulou 2017; Wakabayashi 2011; Wang 2017; Zhang 2020). Gottlieb 2011 did not publish any results, and results from Wang 2017 could not be included due to a reporting error. Outcomes were reported after 3 months, 4 months, 6 months, 12 months, and/or 24 months. The data allowed us to calculate the MRC Dyspnoea Scale score at short-, medium-, and long-term follow-up. Pooling showed signi cant improvement in favour of IDM for short-term follow-up (MD -0.33, 95% CI -0.52 to -0.15). Pooling of mMRC Dyspnoea Scale scores at medium- and long-term followup showed heterogeneity (I² = 96%) too large to be permit conclusions based on the results (Analysis 1.28). Dyspnoea as measured by Borg Scale score in three studies showed no di erences between groups (MD 0.14, 95% CI -0.70 to 0.98; I² = 39%) (Boxall 2005; Gottlieb 2011; Güell 2000). 5. Mortality Fifteen studies assessed mortality as an outcome or as part of patient safety assessment. Of these studies, two assessed mortality at 6 months’ follow-up (Aboumatar 2019; Bernocchi 2017), nine at 12 months’ follow-up (Fan 2012; Farrero 2001; Kessler 2018; Littlejohns 1991; Rice 2010; Rose 2017; Sanchez-Nieto 2016; Smith 1999; Vianello 2016), and four after more than 15 months’ follow-up (Kruis 2014; Lou 2015; Sridhar 2008; Titova 2017). The numbers for Lou 2015 are lower, taking clustering into account. Fan 2012 was temporarily stopped because all-cause mortality was higher in the intervention group than in the usual care group. A thorough investigation of the circumstances of death by an independent and blinded panel showed that death was unrelated to the intervention, and a minority of deaths were due to COPD. Pooling of death events in IDM and control groups across all studies showed a non-statistically signi cant e ect in favour of the intervention (OR 0.86, 95%CI 0.59, to 1.25). Heterogeneity was substantial and could not be explained by duration of follow-up, as outcomes were comparable after medium-term (OR 0.80, 95% CI 0.45 to 1.43) and long-term follow-up (OR 0.87, 95% CI 0.48 to 1.57) (Analysis 1.30).
RkJQdWJsaXNoZXIy MTk4NDMw