Retracted and Republished COPD Study Shows Harm From Intensive Hospital-to-Home Intervention

  • earn free cme credit

    Earn CME credit by reading this article and completing the posttest.

Study Authors: Hanan Aboumatar, Mohammad Naqibuddin, et al.; Seppo T. Rinne, Peter K. Lindenauer, David H. Au; Howard Bauchner, Robert M. Golub

Target Audience and Goal Statement: Pulmonologists, infectious disease specialists, hospitalists, emergency department physicians, primary care physicians

The goal of this study was to evaluate whether a hospital-initiated program that combined transition and long-term self-management support for patients hospitalized with chronic obstructive pulmonary disease (COPD) can improve outcomes.

Question Addressed:

  • Can a hospital-initiated program lead to reduced acute care use and better quality of life for patients hospitalized with COPD?

Study Synopsis and Perspective:

A research paper’s retraction does not always signify misbehavior by researchers. In an era in which the arc of the scientific record continues to bend towards accuracy, researchers took the proactive stance of alerting JAMA editors to statistical programming errors that led to initial erroneous findings of a positive association between a hospital-initiated COPD care intervention and reduced acute care use.

Action Points

  • A corrected version of a retracted study based on a single-site randomized clinical trial of patients with chronic obstructive pulmonary disease (COPD) found that a program combining transition and long-term self-management support resulted in a greater number of COPD-related hospitalizations and emergency department visits compared with usual care.
  • Realize that this type of interventional program may result in more acute care use among patients with COPD, but this requires further research to determine the reason for this unexpected finding.

Subsequent complete re-analysis by the authors led by Hanan Aboumatar, MD, MPH, of the Johns Hopkins School of Medicine in Baltimore, which was republished in JAMA (following retraction of the 2018 paper), showed that a 3-month program combining care transition and long-term self-management support resulted in significantly greater COPD-related hospitalizations and emergency department visits, without improvements in quality of life.

In an editor’s note, JAMA editors Howard Bauchner, MD, and Robert M. Golub, MD, noted that a detailed explanation by the authors of the coding error, corrected findings, and additional analyses to elucidate potential sources of bias that could explain the unexpected results factored into why the original paper was retracted and the revised data were published as a new article. Generally speaking, they urged authors to “continue to report errors in their own work, so that along with editors, they can jointly decide whether a correction, retraction, retraction with replacement, or retraction with republication is required.”

COPD is a leading cause of death in the U.S. and a major cause of hospitalizations and emergency department visits. How to prevent acute care use remains a vexing public health issue, compounded by the fact that the Hospital Readmission Reduction Program of the Centers for Medicare and Medicaid Services financially penalizes hospitals with high 30-day risk-adjusted readmission rates for select conditions, including COPD.

Co-developed by Aboumatar’s group, BREATHE (Better Respiratory Education and Treatment Help Empower) was a single-blinded randomized clinical trial comprised of two groups of COPD patients (intervention and usual care). The goal of the family-centered transitional program was to help prepare hospitalized COPD patients and their family caregivers for the transition from hospital to home and to reduce their need for future acute care. Services were individualized to suit patient needs and priorities at the hospital and for 3 months after discharge. The nurse-implemented program delivered care to 240 patients (mean age 64.9 years, 61.7% women) randomized to receive the program (120 patients) or usual care (120 patients). Mean baseline St. George’s Respiratory Questionnaire (SGRQ) score was 62.3 in the intervention group and 63.6 in the usual care group. Most of the patients completed the study (203 patients [85%]).

Co-primary outcomes were the number of COPD-related acute care events (hospitalizations and emergency department visits) per participant at 6 months and the change in participants’ health-related quality of life measured by SGRQ at 6 months after discharge (score, 0 [best] to 100 [worst]: 4-point difference is meaningful).

The total number of COPD-related acute care events in this study was 238 (196 COPD-related hospitalizations and 42 COPD-related emergency department visits).

At 6 months, the mean number of COPD-related acute care events per participant was 1.40 (95% CI 1.01-1.79) in the intervention group versus 0.72 (95% CI 0.45-0.97) in the usual care group (difference 0.68, 95% CI 0.22-1.15, P=0.004).

Mean change in total SGRQ scores were 2.81 and -2.69 for the intervention and usual care groups — a difference that was similar after adjustment for hospital unit and baseline SGRQ score (adjusted difference 5.18, 95% CI -2.15 to 12.51, P=0.11). Because the differences in SGRQ scores had wide confidence intervals, there was increasing uncertainty in estimating intervention effects on health-related quality of life.

There were 15 deaths (eight in the intervention group and seven in the usual care group) and 339 hospitalizations (202 and 137, respectively) during the study period. None of the reviewed adverse events were attributed to the study intervention.

Supplementary post-hoc analyses showed that — when compared with the usual care group — the incidence rate ratio of COPD-related events was 1.96 before adjustment and 1.59 after adjustment for age, home oxygen use, discharge unit, and hospitalization in the prior year.

More than half of the patients (52%) experienced at least one COPD-related acute care event in the intervention group versus 38% in the usual care group. Of these participants, 22%, 10%, 5%, and 14% had one, two, three, and four or more COPD-related acute care events in the intervention group compared with 20%, 12%, 3%, and 3% in the usual care group, respectively.

Researchers acknowledged a high proportion of low-income and less-educated participants, who might have greater challenges in accessing the healthcare system for urgent visits outside of coming to the emergency department. Small differences in the study groups at baseline (e.g., higher percentages of current smokers and continuous home oxygen therapy use) might have led to greater healthcare utilization in the intervention group, they added. Other study limitations included limited generalizability and possible misdiagnosis of some patients, since spirometry evidence of airflow obstruction was not required for enrollment into the study. According to the Global Initiative for Chronic Obstructive Lung Disease (GOLD), assessments of the disease should include measurement of post-bronchodilator spirometry.

Source References: JAMA 2019; DOI: 10.1001/jama.2019.11982

Editorial: JAMA 2019; DOI: 10.1001/jama.2019.12976

Editor’s Note: JAMA 2019; DOI: 10.1001/jama.2019.14503

Study Highlights and Explanation of Findings:

No significant difference in health-related quality of life was seen between the intervention and usual care groups in a single-site randomized clinical trial of patients hospitalized due to COPD. The 3-month program showed harm, rather than benefit, to patients.

The JAMA editors emphasized that the various analyses conducted to explain the results should be considered exploratory. Nevertheless, post-hoc subgroup analyses enabled researchers to speculate about the unexpected findings. Patients with high disease activation status at baseline and more men than women experienced an increase in acute care events. Perhaps these patients were more vigilant about detecting early signs of COPD exacerbations and accessed medical attention within 24 hours, as recommended by their action plan. It was possible that their regular physician might not have been promptly available, leading them to seek acute care services.

Some of the study participants might have lacked established connections with the healthcare system for routine care and opted to seek acute care. Alternatively, increased communication with physicians about exacerbation signs might have led to increased referrals to the emergency department (and subsequent hospitalizations). Taken together, there was a definite need for future research to explain the unexpected finding.

Revised findings were consistent with evidence from the literature showing that strategies to prevent COPD hospitalizations have produced conflicting results.

However, the revised findings also raise questions about possible unintended consequences of highly intensive patient education and management programs like the one examined in the study, said pulmonologist Seppo T. Rinne, MD, PhD, of the Veterans Affairs Center for Healthcare Organization & Implementation Research in Boston.

Rinne was not involved with the research, but he co-authored editorials published with the initial study findings and the revised findings.

He told MedPage Today that the new findings have larger implications and highlight the limitations of policies targeting hospital readmission as a metric for treatment success.

“The study findings were published in JAMA because there have been few interventions that have been shown to reduce COPD hospitalizations and readmissions, and the [initial analysis] showed really dramatic reductions in both,” he said.

The revised analysis confirms that “there is still limited evidence on how to prevent COPD hospitalizations and readmissions,” Rinne added.

“We are left with a lack of guidance, from a healthcare system perspective, on how to define high-quality care for these patients,” he said. “The policies say we have to improve this metric, but we don’t know how to do that.”

As a result, Rinne said, health systems are implementing initiatives designed to reduce readmissions that are not evidence based. “These initiatives may not be helping patients, and some may even be harming them,” he said.

In an editorial accompanying the revised article, Rinne and colleagues suggested that physicians rely on evidence-based practices to improve patient-centered outcomes, e.g., smoking cessation; treatment with guideline-concordant medications; initiating oxygen for patients with severe resting hypoxia; referral to pulmonary rehabilitation; and addressing other medical, psychiatric, and socioeconomic conditions affecting their health.

Take Posttest