Abstract
Dear Editor:
I read the editorial by Rocker et al. 1 and I was surprised by its tone and by some of the inflammatory language used to characterize the observational research work that I and colleagues have undertaken on cardiorespiratory outcomes associated with opioid use among older adults with chronic obstructive pulmonary disease (COPD).2,3 The editorial's authors used language such as “myths,” 1 “biases created by authors,” 1 and “a lie” 1 to characterize our research work.2,3 I would like to bring to the attention of your readers several points in response.
First, the results of studies that I and colleagues undertook on respiratory 2 and cardiac 3 outcomes associated with new opioid use among older adults with COPD do not apply to individuals receiving opioids as part of end-of-life care, nor are they specific toward individuals receiving opioids for refractory dyspnea. We evaluated the safety of opioid drug use among the broader, nonpalliative, older adult COPD population. In both our respiratory 2 and cardiac 3 outcome studies, we intentionally excluded individuals receiving palliative care, “since goals of care and indications for opioid use may differ in this context”2 and “since opioids may be appropriately used in this group.” 3 There is evidence indicating that treatment of musculoskeletal pain, and not refractory dyspnea, is the most common reason for opioid use among individuals with COPD, even among those with advanced stage COPD. 4
Second, the editorial's authors write that “the obvious explanation [for our observed associations] is not causation but rather confounding by indication.” 1 However, the editorial's authors overlook the many steps we undertook to minimize possible confounding by indication. By excluding individuals receiving palliative care,2,3 we reduced chances for confounding by indication, as this group would be at increased risk for both opioid exposure and having poor health outcomes. We also used propensity score weighting in our studies, balancing opioid users and nonusers on a wide range of characteristics that would influence chances for opioid receipt and risk of subsequent outcomes.2,3 We specifically balanced opioid users and nonusers on multiple markers of COPD severity, including history of COPD exacerbation, duration of COPD, receipt of respiratory medications, and presence of comorbidities.2,3 Because COPD exacerbation history is the most important indicator of COPD severity, we also performed a sensitivity analysis, evaluating outcomes among the healthiest subgroup of individuals, that is, those with no exacerbation in the year before the index date.2,3 In this subgroup, which would be least likely to be influenced by confounding by indication, significantly increased adverse respiratory and cardiac outcomes in association with opioid drug use were still demonstrated.2,3 Increased rates of respiratory-related morbidity and mortality associated with opioid use were also found to persist after excluding individuals with malignancy from the analysis. 2
Observational studies provide the “best data on adverse effects,” 5 given the large number of unselected individuals they typically include, with little or no participant dropout, and the real-world doses and patterns of drug use they capture. Integrating data from observational studies, with the “best evidence on benefits” from clinical trials, 5 is needed to inform sound drug prescribing and policy.