Interventions to Reduce Rehospitalizations after Chronic Obstructive Pulmonary Disease Exacerbations. A Systematic Review

Abstract

Rationale: Approximately 20% of patients hospitalized for COPD exacerbations in the United States will be readmitted within 30 days. The Centers for Medicare and Medicaid Services has recently proposed to revise the Hospital Readmissions Reduction Program to financially penalize hospitals with high all-cause 30-day rehospitalization rates after a hospitalization for COPD exacerbation on or after October 1, 2014.

Objectives: To report the results of a systematic review of randomized clinical trials evaluating interventions to reduce the rehospitalizations after COPD exacerbations.

Methods: Multiple electronic databases were systematically searched to identify relevant studies published between January 1966 and June 2013. Titles, abstracts, and, subsequently, full-text articles were assessed for eligibility. Each study was appraised using predefined criteria.

Measurements and Main Results: Among 913 titles and abstracts screened, 5 studies (1,393 participants) met eligibility criteria. All studies had a primary outcome of rehospitalization at 6 or 12 months. No study examined 30-day rehospitalization as the primary outcome. Each study tested a different set of interventions. Two studies (one conducted in Canada and one conducted in Spain and Belgium) showed a decrease in all-cause rehospitalization over 12 months in the intervention group versus comparator group (mean number of hospitalizations per patient, 1.0 vs. 1.8; P = 0.01; percent hospitalized, 45 vs. 67%; P = 0.028; respectively). The only study conducted in the United States found a greater than twofold higher risk of mortality in the intervention group (17 vs. 7%, P = 0.003) but no significant difference in rehospitalizations. It was unclear which set of interventions was effective or harmful.

Conclusions: The evidence base is inadequate to recommend specific interventions to reduce rehospitalizations in this population and does not justify penalizing hospitals for high 30-day rehospitalization rates after COPD exacerbations.

Chronic obstructive pulmonary disease (COPD) affects approximately 24 million individuals and is the third leading cause of death in the United States (1). Each year, COPD leads to as many as 800,000 hospitalizations and nearly $50 billion in healthcare expenditures (1, 2). About 20% of patients hospitalized with COPD exacerbations are rehospitalized within 30 days, and healthcare expenditures for rehospitalizations in this population rank as the third highest among Medicare beneficiaries (3). It is therefore not surprising that provisions in the Affordable Care Act specify rehospitalizations after COPD exacerbations as a potential target for financial penalties by the Centers for Medicare and Medicaid Services (CMS) (4). In May 2013, CMS invited public comments on a proposal to add hospitalizations for COPD exacerbations to its Hospital Readmissions Reduction Program, which would in effect trigger financial penalties directed at hospitals if admissions for COPD exacerbations resulted in a higher-than-expected all-cause 30-day rehospitalization rate (i.e., even rehospitalizations not attributed to COPD) (5). The proposal would expand the CMS Hospital Readmissions Reduction Program, which is currently focused on reducing rehospitalizations after a hospital admission for heart failure, myocardial infarction, or pneumonia (6). Understandably, hospitals and provider groups are seeking evidence-based strategies to reduce rehospitalization rates for various patient populations, including for patients with COPD exacerbations. Although there are a number of interventions that have been shown to reduce rehospitalizations (7–10), none of these have been specifically tested in patients hospitalized for COPD exacerbations, and none include interventions to address the specific needs of this patient population (eg, oxygen titration and education, inhaler use teaching, noninvasive ventilatory support).

Previous systematic reviews in patients with COPD have studied self-management education (11, 12), integrated care (13–15), shared management between specialists and primary care physicians (16), and home care (17) in patients with stable COPD, or hospital at home for acutely ill patients with COPD exacerbations (18) or other conditions (19). Previous reviews on care transitions from hospital to home have focused on other patient populations (ie, not patients with COPD exacerbations) (20–22). However, no review to date has focused specifically on identifying interventions that reduce rehospitalizations in patients hospitalized for COPD exacerbations. We therefore conducted a systematic review of clinical trials to identify interventions that could reduce the risk of rehospitalization in patients initially hospitalized with COPD exacerbations. Preliminary findings were reported at the 2013 meeting of the American Thoracic Society International Conference (23).

Methods

Data Sources

With the assistance of a medical librarian, we conducted a systematic search of multiple electronic databases (PubMed, EMBASE, CINAHL, the Cochrane library) for studies published between January 1966 and June 2013. The PubMed search was constructed by using combinations of Medical Subject Heading (MeSH) search terms and keywords according to the following algorithm: (“COPD” [MeSH] AND “exacerbation”) AND (“self-management” OR “care facilitator” OR “coordinator” OR “case manage*” OR “facilitator” OR “patient navigator” OR “navigator” OR “integrated care”) OR (“COPD” [MeSH] AND (“patient discharge” [MeSH] OR “discharge planning” [MeSH] OR “care continuity, patient” [MeSH]) OR (“COPD” [MeSH] AND “hospital readmission” [MeSH])). The other databases were queried using similar terms (see Table E1 in the online supplement). We used Google Scholar and Web of Science for citation searches to identify additional articles. Reference lists in review articles were also examined to identify additional articles.

Study Selection

Inclusion criteria for full-text review and data abstraction were: (1) published in English, (2) randomized clinical trial design, (3) enrolled patients with COPD, (4) patients were hospitalized for COPD exacerbation within the previous 12 months, and (5) primary outcome was rehospitalization (all-cause or COPD-related) or composite endpoint that included rehospitalization. We excluded studies that focused exclusively on decreasing the hospital length of stay(e.g., early assisted discharge, hospital at home), because such interventions are not necessarily expected to also reduce the risk of rehospitalization. Studies suggest that multiple factors contribute to safe transitions from hospital to home (e.g., access/quality of ambulatory care, social support, no follow-up appointment, transportation problems, inadequate care coordination with ambulatory providers) (24). We therefore excluded studies limited to pharmacotherapy, procedures, or technology-based interventions, because these are not designed to address the multiple factors above. We also excluded studies that exclusively focused on pulmonary rehabilitation in the context of a recent hospitalization for COPD exacerbation, as this topic was addressed in a recently published systematic review (25). Two authors independently screened studies for eligibility criteria through title and abstract review (V.P. and M.A.M.). The full-text review of the preselected articles was performed independently by two authors (V.P., M.A.M., H.A.G., N.I.R., or J.A.K.) to identify the final set of eligible studies. Disagreements over study eligibility were resolved by third-party arbitration (N.I.R. or H.A.G.).

Data Abstraction and Risk of Bias Assessment

Two authors (V.P., M.A.M., M.J.J., N.I.R., H.A.G., B.P., or S.M.N.) independently abstracted data from each study’s published protocol (Table E2) and assessed risk of bias, per the Cochrane Effective Practice and Organization of Care (EPOC) Group’s risk of bias criteria (26). The EPOC risk of bias criteria includes (1) random sequence allocation, (2) concealed allocation, (3) masking of participants and personnel, (4) masking of outcome assessment, (5) incomplete outcome data, (6) selective reporting, and (7) other bias. The risk of bias for each of these domains was graded as high, low, or unclear. If necessary, disagreements were resolved through a third-party arbitrator (J.A.K.). Additionally, to minimize the risk of data abstraction errors, the corresponding authors of the reviewed studies were contacted for clarification of study procedures (27).

Data Synthesis and Analysis

Based on a previous study (22), we classified intervention components into three temporal categories: (1) predischarge interventions, (2) postdischarge interventions, or (3) bridging interventions (spanning the pre- and postdischarge periods). Because of the heterogeneity of interventions, measurements, and reporting of outcomes, we present a narrative synthesis (rather than a metaanalysis) of findings. Outcomes of interest were rehospitalizations (all-cause and COPD-related) and mortality (all-cause and COPD-related), as defined by the authors in the various studies.

Results

Study Selection

Our query of the electronic databases and review of reference lists and citation searches identified 913 studies (Figure 1). Based on review of titles and abstracts, 75 were selected for full-text review. Of those, five studies met all eligibility criteria and were selected for data abstraction and synthesis (Table 1).

Figure 1.

Summary of evidence search and selection. COPD = chronic obstructive pulmonary disease. *Reasons for exclusion exceed number of studies excluded due to studies being excluded for more than one reason.

Table 1.

Characteristics of clinical trials

	Author	Year	Country	No. Sites	No. Participants	Mean Age (yr)	% Women	FEV1	Intervention Setting	Assessment Period (mo)	No. EPOC Criteria Satisfied (7 Possible)
	Bourbeau et al. (31)	2003	Canada	7	191	69	45	0.99 L	3	12	6
	Kwok et al. (30)	2005	China	2	157	74	29	NR	1, 2, 3	6	5 (of 6 evaluable domains)
	Casas et al. (29)	2006	Spain, Belgium	2	155	71	17	42% pred	1, 2, 3	12
	Bucknall et al. (32)	2012	UK	6	464	69	63	41% pred	3	12	6
	Fan et al. (28)	2012	United States	20	426	66	3	48% pred	3	12	6

Definition of abbreviations: EPOC = Effective Practice and Organization of Care criteria; NR = not reported; % pred = percent of predicted spirometry value.

Summary: The five studies were performed in six countries, involving 2 to 20 sites per study with a number of participants ranging from 155 to 464 (total number of participants: 1,393). In three out of five studies the proportion of female participants was < 30%. Most studies focused on postdischarge interventions, with very few studies performing either predischarge or bridging interventions. The two largest studies were conducted in the UK (6 sites, 464 predominantly female participants) and the United States (20 sites, 426 predominantly male participants) and focused exclusively on postdischarge interventions with a duration of 12 months. Intervention setting: 1 = interventions performed in the predischarge period (implemented before hospital discharge). 2 = interventions performed as bridging interventions (same person conducted interventions in the pre- and postdischarge periods). 3 = interventions performed in the postdischarge period (implemented after hospital discharge).

Characteristics of Clinical Trials

The five studies enrolled a total of 1,393 participants (range, 155–464 per study) in six countries; only one of these studies was conducted in the United States (28). All studies included postdischarge interventions, but only two studies also included predischarge and bridging interventions (29, 30). The studies were designed to examine rehospitalizations at 6 months (one trial) (30) or 12 months (four trials) (28, 29, 31, 32), but not at 30 days. In the four studies that reported sufficient information to evaluate the risk of bias in all seven EPOC domains (Table E3), the risk of bias was considered low in all domains except masking of study participants and personnel to the intervention (28, 29, 31, 32).

Study Interventions

Interventions in each study are summarized in Tables 2 and 3, with more detailed descriptions in Tables E2, E4, and E5. More than 15 different strategies formed intervention bundles (i.e., multiple interventions implemented as part of a care strategy) across the various studies. The number of interventions in the bundle used in each study ranged from 9 to 11. All five studies provided patient education about use of respiratory inhalers, developed an action plan (instruction on steps to be taken in case of worsening symptoms), and provided participants a hotline (phone or pager number that patients could call as needed). The intervention bundle in four studies also included a different combination of education about COPD, general health counseling (e.g., living a healthy lifestyle), coordination with the patient’s primary provider, home visits, and a follow-up phone call. Less frequent interventions included smoking cessation counseling (three studies), social services referral (two studies), assessment of comorbidities (one study), discharge planning (one study), and pulmonary rehabilitation (one study). Medication reconciliation, scheduling follow-up appointments with the patient’s provider, provider continuity, patient-centered discharge instructions, or referral to a smoking cessation program were not components of the intervention bundle for any of the five studies. There was substantial heterogeneity between studies, such as the timing (eg, predischarge vs. postdischarge), frequency (e.g., number of home visits), and how each intervention was delivered (e.g., type and number of personnel conducting interventions). In two studies, the control group also received lower-intensity interventions; in the other studies, usual care was the comparator (28, 32). Two studies initiated interventions more than 28 days after hospital discharge (28, 31).

Table 2.

Interventions performed

	Interventions	Study, year
		Bourbeau et al., 2003 (31)	Kwok et al., 2005 (30)	Casas et al., 2006 (29)	Bucknall et al., 2012 (32)	Fan et al., 2012 (28)
	Discharge planning			1
	Disease education	3		1	3	3*
	Health counseling	3	1		3	3
	Inhaler use teaching	3	1	1	3*	3
	Development of action plan	3	1	1	3	3
	Medications given for action plan	3			3	3
	Smoking cessation counseling	3			3*	3
	Assessment of comorbidities			3
	Referral to pulmonary rehabilitation				3*
	Exercise program	3	3			3
	Referral to social services		3	3
	Communication with patient’s PCP	3	3	3		3*
	Transition navigator		2	2
	Home visits	3	3	3	3
	Follow-up telephone call	3	3	3		3
	Patient hotline	3	3	3	3	3*

Definition of abbreviations: PCP = primary care provider.

Intervention setting: 1 = Interventions performed in the predischarge period. 2 = Interventions performed as bridging interventions. 3 = Interventions performed in the postdischarge period.

^*Interventions that were also implemented in some form in the control group. See Table E2 in the online supplement for further details.

Table 3.

Home visit procedures

	Study, Year
	Bourbeau et al., 2003 (31)	Kwok et al., 2005 (30)	Casas et al., 2006 (29)	Bucknall et al., 2012 (32)	Fan et al., 2012 (28)
Staff performing home visit	RN, PT, RT	RN	RN, MD, SW	RN	None
Days from hospital discharge to first home visit	>28	<7	<3	29 (median)
No. of home visits	8	9	1	>10
Months from first home visit to last home visit	2	6	n/a	12

Definition of abbreviations: MD = medical doctor; n/a = not applicable; PT = physical therapist; RN = registered nurse; RT = respiratory therapist; SW = social worker.

Outcomes (Rehospitalizations and Mortality)

Measurement and reporting of outcomes varied across studies. Rehospitalizations were assessed by reviewing hospital or national health records (two studies) (30, 32), patient interviews or questionnaires (one study) (31), or a combination of both (two studies) (28, 29). Mortality was assessed through review of medical records and interviews of family members in two studies (28, 29), review of national health records (32), or was not adequately described (30, 31). None of the studies reported 30-day rehospitalization rates, although one study involving 157 patients in China examined 28-day rehospitalization rates as a secondary outcome (30).

Two of the five studies demonstrated a significant decrease in all-cause rehospitalizations in the intervention group versus comparator group at 12 months (mean number of hospitalizations per patient, 1.0 vs. 1.8; P = 0.01; percent rehospitalized, 45 vs. 67%, P = 0.028) (29, 31), with one also reporting a significant decrease in COPD-related rehospitalizations at 12 months (32 vs. 50%, P = 0.01) (Table 4) (31). None of the studies demonstrated a significant reduction in all-cause mortality in the intervention group at 6 or 12 months, although one study (the only study conducted in the United States), was terminated prematurely by an independent Data Safety Monitoring Board due to excess risk of death in the intervention group (compared with usual care; 17 vs. 7%, P = 0.003) (28). An extensive evaluation by the authors of this study, a multicenter study conducted across 20 Veterans Affairs hospitals, failed to identify a reason that mortality was higher in the intervention group.

Table 4.

Rehospitalizations and mortality

Intervention Components	Study, Year
	Bourbeau etal., 2003 (31)	Kwok etal., 2005 (30)	Casas etal., 2006 (29)	Bucknall etal., 2012 (32)	Fan etal., 2012 (28)
	(n = 191)	(n = 157)	(n = 155)	(n = 464)	(n = 426)
Assessment period, mo	12	6	12	12	12
Rehospitalizations, intervention vs. control
All-cause	1.0 vs. 1.8 (mean)*	76 vs. 62 (at 28 d: 47 vs. 37)	45 vs. 67*	69 vs. 72	37 vs. 36
COPD-related	32 vs. 50*			44 vs. 46	27 vs. 24
Mortality, intervention vs. control
All-cause	5 vs. 9	4 vs. 8	18 vs. 16	13 vs. 9	17 vs. 7*
COPD-related				10 vs. 7	7 vs. 3

Data are given as percent unless otherwise noted.

^*P < 0.05.

Discussion

To our knowledge, this report represents the first systematic review of literature to evaluate interventions to reduce rehospitalizations among patients hospitalized with COPD exacerbations. Results indicate that (1) there are no published clinical trials examining the effects of interventions to reduce 30-day rehospitalization rates in this population as the primary outcome, with studies largely designed to examine outcomes at 6 months or 12 months; (2) there is substantial heterogeneity in the design, measurement, and reporting of study results, precluding a formal metaanalysis; and (3) interventions tested to date in this population may result in benefit, harm, or have no discernible effect. Overall, it appears that there is inadequate evidence to recommend disease-specific interventions to reduce 30-day rehospitalization rates in patients initially admitted for COPD exacerbations.

Although 30-day rehospitalization rates among patients hospitalized for COPD exacerbations are of particular interest in the United States, given the potential financial penalties imposed by CMS, we did not identify any clinical trial designed specifically to address this outcome. Only one of the five studies we identified was conducted in the United States (28), and neither this study nor the other four were designed specifically to examine 30-day rehospitalization rates. Also, two studies initiated interventions 28 days or more after hospital discharge, limiting the usefulness of such studies to identify interventions to decrease 30-day rehospitalization rates (28, 31).

We were unable to combine the results from the different studies and carry out a metaanalysis due to several sources of heterogeneity. First, even among studies that implemented similar interventions (e.g., home visits), there was significant variability in both content (e.g., topics addressed during home visits) and context (e.g., number and timing of home visits, personnel conducting the home visits). Second, there was significant heterogeneity in the measurement (e.g., self-report, review of medical records) and reporting (e.g., percent rehospitalized, mean number of rehospitalizations per patient) of outcomes. Third, there was inconsistent reporting of patient characteristics such as socioeconomic status, level of social support, and comorbidities, all of which have been linked with rehospitalization rates (33–35). Establishing standards for measuring and reporting patient characteristics and outcomes will help readers determine whether study results are applicable to their clinical populations and can therefore be used to develop local quality-improvement programs. For example, although the average individual with COPD has six or more comorbid conditions and more than 95% have at least one condition that may complicate the treatment of COPD (36), only one study reported addressing participants’ comorbidities as part of their intervention (29). Even in this report, there was insufficient information about the evaluation or treatment of comorbidities (e.g., which comorbidities were assessed, how did this evaluation modify the postdischarge treatment plan) in the context of the interventions to reduce rehospitalizations.

Interventions had highly variable effects on rehospitalizations and mortality (from benefit to harm). Most of the studies focused on postdischarge interventions, with only two studies also including predischarge and bridging interventions (29, 30). It was unclear, however, which intervention or groups of interventions were most likely to cause benefit or harm. For example, three studies all used COPD education, health counseling, inhaler use teaching, action plans, smoking cessation counseling, and a patient hotline (28, 31, 32), yet observed vastly different outcomes. Although one study demonstrated a significant reduction in rehospitalizations (31), others showed either no significant effect or even an increase in mortality in the intervention group (28, 30). Differences in interventions or study populations likely contributed to heterogeneous findings.

In a previous systematic review (not focused on patients with COPD exacerbations), the authors concluded that that no specific intervention or intervention bundle was consistently associated with a reduction in 30-day rehospitalization rates (22). Only one of the five clinical trials in the current report was included in this previous systematic review (30). Our findings, however, are consistent with this previous report, and we were unable to identify intervention to reduce 30-day rehospitalizations in patients with COPD exacerbations; our findings also indicate that there do not appear to be interventions that consistently reduce rehospitalizations at subsequent time points (e.g., 6 or 12 months post discharge). Other randomized controlled trials that enrolled ambulatory patients or patients presenting to the emergency department with COPD exacerbations (but not hospitalized) also found inconsistent benefits of interventions designed to reduce hospitalizations (37–40). Although there appear to be a large number of studies examining strategies to reduce rehospitalizations, the lack of a standard approach to defining patient populations, study interventions and comparators, outcomes, timeframe for assessing outcomes, and settings in which the study was conducted are barriers to cross-study syntheses of evidence. The National Institute of Health recently convened a highly successful multidisciplinary conference on standardizing outcomes in clinical research for patients with asthma (41). A similar effort is needed to facilitate the conduct of systematic reviews of studies designed to reduce rehospitalizations.

This systematic review has potential limitations. We may have missed identifying some effective interventions because we restricted our review to randomized clinical trials and did not include evidence from other study designs; however, randomized clinical trials are generally considered as providing the strongest quality of evidence. Also, this review did not include studies that focused exclusively on pharmacological interventions, and we may have missed identifying pharmacological agents effective in reducing rehospitalizations. Another potential limitation involves availability of data. Results of this systematic review are largely based on descriptions of interventions and outcomes published in journal articles and supplemented with queries to study authors. The availability of detailed study protocols may have helped us understand the reasons for the heterogeneous effect of interventions across studies.

The findings of this report have two implications. First, it is not possible to make recommendations for COPD-specific interventions to reduce rehospitalizations after COPD exacerbations. In fact, our findings indicate that providers must be careful when implementing programs to reduce rehospitalizations in this population, given the twofold excess risk of death identified in one of the largest studies to date. Second, although nearly all patients with COPD have clinically relevant comorbidities, only one study in this report devoted attention to these conditions (29). We suspect that developing and implementing disease-specific strategies (e.g., separate strategies for patients with COPD and for patients with heart failure) to reduce rehospitalizations supplemented by interventions focusing on specific comorbid conditions will be burdensome and inefficient. A more general patient-centered approach that includes interventions that are likely to be effective for multiple populations (eg, coordination of care with ambulatory providers, addressing financial and social barriers to care) supplemented with disease-specific interventions as needed (e.g., inhaler use and supplemental oxygen teaching in patients with COPD exacerbations) may be more efficient for health systems and providers to implement but needs further testing (42). The Society of Hospital Medicine’s Project BOOST, for example, provides the opportunity to tailor the hospital-to-home transition process to the patient’s needs but would benefit from the addition of evidence-based modules that address the needs of patients with COPD exacerbations (10).

In conclusion, we found significant heterogeneity in the design and outcome of clinical trials to reduce rehospitalizations in patients with COPD exacerbations. Clinical trials identified in this review focused on reducing rehospitalizations at 6 and 12 months; no study focused on decreasing 30-day rehospitalizations as a primary endpoint. No specific intervention or bundle of interventions could be identified as effective in reducing the rate of rehospitalizations in this population. Moreover, the only clinical trial to date conducted in the United States documented a twofold excess risk of death in patients assigned to the intervention group and no significant effect on the risk of rehospitalization. The currently available evidence provides very limited guidance to practitioners and health systems seeking to reduce rehospitalizations in this population and does not justify penalizing hospitals for high 30-day rehospitalization rates after COPD exacerbations. Quality improvement initiatives should carefully monitor the effect of their intervention on mortality and rehospitalizations. An effort to promote the use of standardized descriptions of patient characteristics, interventions, and outcomes is needed.