Chronic diseases like chronic obstructive pulmonary disease (COPD) account for at least 2/3 of medical care spending in the U.S. Policy makers, payers, and many physicians recognize that the outpatient clinic-based model is poorly suited to provide support in between physician visits, when most complications or exacerbations occur. Many hospitalizations and decline in function could be prevented, the thinking goes, with effective "disease management programs" using telephone support by non-physicians, mobile/remote technology, and patient-triggered treatment algorithms. Such programs have been modestly successful for diabetes, depression, congestive heart failure, and arthritis, and insurers are rolling many such programs out nationally to their members in an attempt to rein in costs.
Vincent Fan, Michael Gaziano, Dennis Niewoehner et al report the puzzling and messy results of implementing just such an outpatient disease management program for COPD at Veterans Affairs hospitals across the U.S. in the May 15 2012 Annals of Internal Medicine. For unclear reasons, more patients died while receiving the intervention, and the randomized trial was stopped midway for safety.
What They Did
At 20 VA hospitals across the U.S., investigators randomized 426 outpatients with COPD, who had been hospitalized within the last year for COPD, to receive either usual care or a "comprehensive care management program (CCMP)" that included:
- COPD education in 4 individual sessions and one group session (with breathing exercises, anxiety reduction, exercise, smoking cessation, etc.);
- Scheduled telephone calls by a case manager (monthly for 3 months, then once every 3 months) to check up on them, answer questions, and offer support;
- Prescriptions for prednisone and antibiotics, with a written action plan on when to begin taking them. Authors don't say which antibiotics were prescribed.
Patients were followed for an average of 250 days. Primary outcome was time to first hospitalization for COPD.
What They Found
The initial plan in 2007 was to enroll 960 patients (the number expected to be needed to detect an effect), but long before that -- at 426 patients, in 2009 -- the trial was stopped when an excess of deaths was observed in the intervention group:
- 28 deaths from all causes in the COPD self-management group vs. 10 in the usual-care group (hazard ratio of 3.0, p=0.003).
- 71% of the deaths were explainable; most of the explainable excess deaths were due to COPD (10 in the intervention group, 3 in the usual care group, p=0.053).
No reason could be discovered from the data as to why the excess deaths occurred -- neither those due to COPD, nor the deaths from unknown causes:
- Patients were hospitalized roughly equally frequently (27% in the intervention group, 24% in the usual-care group, non-significant difference).
- Patients in both groups self-reported the same total number of COPD exacerbations (600 vs. 610, or ~4.3 per year in each).
- No single study site was responsible for skewing the data.
- No patient-level differences (e.g., depression, severity of COPD) could be detected.
Notably, the patients in the COPD self-management group didn't treat themselves very "aggressively." They used prednisone ever-so-slightly more than the usual care patients, but not antibiotics. Yet they didn't have significant delays in care that were noted in the data, either.
What It Means
As authors say, it's impossible to conclude from the data what went wrong, and why. Was it just chance? Differences between patients (drug use, undiagnosed heart disease) not detectable from the data? Not enough case management phone calls (only 1 every 3 months)? Not enough "empowerment" or training of patients to know when to start their action plan of prednisone and antibiotics? Undetected delays in care in the intervention group -- for example, a false sense of security induced by having the scripts on-hand, causing intervention patients to minimize symptoms -- resulting in an "equal" time from onset-to-hospitalization, that was actually longer?
Previous trials testing similar COPD self-care interventions did show a marked reduction (~40% relative) in either hospitalizations or emergency department visits (1, 2), with no observed safety problems. One of these was in 5 VAs, and made a nice splash in the press when its positive findings were announced 2 years ago, possibly prompting some physicians to start similar programs informally (e.g., giving COPD patients standing scripts for prednisone and/or antibiotics to take when needed).
Is that still a good idea (or was it ever)? I don't know; what do you think?
Fan VS et al. A Comprehensive Care Management Program to Prevent Chronic Obstructive Pulmonary Disease Hospitalizations. A Randomized, Controlled Trial. Ann Intern Med 2012;156:673-683.
Rice KL et al. Disease Management Program for Chronic Obstructive Pulmonary Disease. A Randomized Controlled Trial. AJRCCM 2010;182:890-896.
Bourbeau J et al. Reduction of hospital utilization in patients with chronic obstructive pulmonary disease: a disease-specific self-management intervention. Arch Intern Med 2003 Mar 10;163(5):585-91.