A valid, robust, and seemingly low-biased definition
AIM: We hypothesized that medically treated exacerbations in COPD defined as treatments with oral corticosteroids alone or in combination with antibiotics by register linkage with a nationwide prescription registry is a valid, robust and low-biased measure of exacerbations. METHODS: A total of 13,591 individuals with COPD in the Copenhagen General Population Study (2003-2013) were linked to the Danish prescription registry. Exacerbations were defined as dispensing of oral corticosteroids alone or in combination with antibiotics, dispensed less than four weeks apart during three years of follow-up. Construct validity of this definition of medically treated exacerbations was assessed by studying baseline determinants as well as by studying the association between GOLD 1 through 4 grades and time to first exacerbation during follow-up. RESULTS: Among individuals with COPD, 964 individuals (7.1%) had at least one exacerbation during follow-up. At baseline, comparing those with versus without exacerbations during follow-up, FEV1, 72% of predicted vs. 85% (p < 0.001), previous exacerbations, 43% vs. 11% (p < 0.001), breathlessness, 33% vs. 14% (p < 0.001), and use of inhaled medications, 54% vs. 14% (p < 0.001) were associated with exacerbations. Compared to individuals with GOLD 1, the multivariable hazard ratio (HR) for exacerbations was HR = 17.4 (12.3-24.5, p < 0.001) for GOLD 4, HR = 4.8 (3.9-5.9, p < 0.001) for GOLD 3, and HR = 2.0 (1.7-2.3, p < 0.001) for GOLD 2. In sensitivity analyses, our definition of exacerbations was robust and without major biases. CONCLUSIONS: Compared to individuals with GOLD 1, the risk of exacerbations was 17-fold for GOLD 4, 5-fold for GOLD 3, and 2-fold for GOLD 2. Medically treated exacerbations defined by register linkage seem a valid, robust, and low-biased measure of exacerbations in COPD.
Respiratory Medicine, 2015, Vol 109, Issue 12, p. 1562-1568