Pulmonary rehabilitation as a mechanism to reduce hospitalizations for acute exacerbations of chronic obstructive pulmonary disease: A systematic review and meta-analysis
Running Head: Pulmonary rehabilitation for exacerbations of COPD
Correspondence to:
Elizabeth Moore, Imperial College London.
National Heart and Lung Institute, Emmanuel Kaye Building, 1B Manresa Road, London, SW3 6LR
E: T: 0207 594 8824
Authors:
Elizabeth Moore, MSc, Department of Respiratory Epidemiology Occupational Medicine and Public Health, Imperial College London, London, UK
Thomas Palmer, MSc, Department of Primary Care and Pubic Health, Imperial College London, London, UK
Dr Roger Newson, PhD, Department of Primary Care and Pubic Health, Imperial College London, London, UK
Prof Azeem Majeed, Department of Primary Care and Pubic Health, Imperial College London, London, UK
Dr Jennifer K Quint, MRCP, PhD, Department of Respiratory Epidemiology Occupational Medicine and Public Health, Imperial College London, London, UK
Dr Michael A Soljak, PhD, Department of Primary Care and Pubic Health, Imperial College London, London, UK
Word count: abstract = 265; main text = 3567
Conflicts of Interest Statement
JKQ reports grants from the Medical Research Council (MRC), GlaxoSmithKline (GSK), British Lung Foundation (BLF), Wellcome Trust, during the conduct of the study, and personal fees from AstraZeneca and GSK outside of the submitted work. MS reports other current research grants from Arthritis Research UK, Public Health England and the EU EITHealth Programme, with no conflicts of interest. No other conflicts of interest are reported.
Abbreviations: COPD = Chronic Obstructive Pulmonary Disease; AECOPD = Acute Exacerbation of COPD; GP = General Practitioner; ER = Emergency Room; PR = pulmonary rehabilitation; UC = Usual Care; ED = Emergency Department; FEV1 = Forced Expiratory Volume in 1 second; FCV = Forced Vital Capacity; CI = Confidence Intervals; SD = Standard Deviation; RCT = randomised controlled trial; OR = odds ratio; ICD = International Classification of Diseases
Key words: COPD, exacerbations, hospitalizations, pulmonary, rehabilitation
Abstract
Background: Acute exacerbations of COPD (AECOPD) have a significant impact on healthcare utilization, including physician visits and hospitalizations. Previous studies and reviews have shown that pulmonary rehabilitation has many benefits but the effect on hospitalizations for AECOPD is inconclusive.
Methods: A literature search was carried out to find studies that might help determine, using a meta-analysis, the impact of pulmonary rehabilitation on AECOPD, defined as unscheduled or emergency hospitalizations and emergency room (ER) visits. Cohort studies and randomised controlled trials (RCTs) reporting hospitalizations for AECOPD as an outcome were included. Meta-analyses compared hospitalization rates between eligible pulmonary rehabilitation recipients and non-recipients, before and after rehabilitation.
Results: 18 studies were included in the meta-analysis. Results from ten RCTs showed that the control groups had a higher overall rate of hospitalizations than the pulmonary rehabilitation groups (0.97 hospitalizations/patient-year, 95% Confidence Intervals (CIs) 0.67, 1.40; 0.62 hospitalizations/patient-year,95% CI 0.33, 1.16respectively). Five studies compared admission numbers in the 12 months before and after rehabilitation, finding a significantly higher admission rate before compared to after (1.24 hospitalizations/patient-year, 95% CIs 0.66, 2.34; 0.47 hospitalizations/patient-year, 95% CIs 0.28, 0.79 respectively). The pooled result of three cohort studies found the reference group had a lower admission rate compared to the pulmonary rehabilitation group (0.18 hospitalizations/patient-year, 95% CI 0.11, 0.32 for reference group versus 0.28 hospitalizations/patient-year, 95% CI 0.25, 0.32 for pulmonary rehabilitation).
Conclusions: Although results from RCTs suggested that pulmonary rehabilitation reduces subsequent admissions, pooled results from the cohort studies did not, likely reflecting the heterogeneous nature of individuals included in observational research and the varying standard of pulmonary rehabilitation programmes.
Introduction
COPD 1is a major cause of morbidity and is expected to be the third largest cause of death globally by 2020.2 COPD is characterized by airflow limitation that is usually progressive, and not fully reversible.3 Patients with COPD can experience exacerbations; episodes of deterioration characterized by worsening symptoms including increased dyspnoea, cough, sputum production, and airflow obstruction.4 Acute exacerbations of COPD (AECOPD) mayrequire admission to hospital which can put a heavy burden on health systems. In the United Kingdom COPD exacerbations are the second most common cause of emergency hospital admission with an annual cost to the National Health Service of over £800 million.5 In the United States the total economic burden of COPD in 2007 was calculated to be $42.6 billion and one study reported mean costs of $647 (SD $445) for Emergency Room(ER) visits.6
Pulmonary rehabilitation is an important intervention in the management of COPD. The American Thoracic Society7 and the European Respiratory Society define pulmonary rehabilitation as “a comprehensive intervention based on a thorough patient assessment followed by patient-tailored therapies, which include, but are not limited to, exercise training, education, and behaviour change.” The intention of pulmonary rehabilitation is to improve the physical and psychological condition of people and to promote long-term adherence of health-enhancing behaviours.” It is a key component of the multi-disciplinary management of COPD8 and can improve exercise capacity, dyspnoea,activities of daily living, muscle strength, and self-efficacy.9A recent systematic review has also shown that pulmonary rehabilitation can improve quality of life.10Given the evidence of these benefits, a recent Cochrane Editorial stated that no further systematic reviews are required to show that pulmonary rehabilitation improves patient-related outcomes.
However, the evidence relating to whether it reduces ERvisits and hospitalizations for AECOPD is less definitive. A previous systematic review by Puhanet al11found that pulmonary rehabilitationsignificantly reduced hospitalizations and mortality in patients who had recently suffered an exacerbation of COPD. However more evidence is needed relating to all COPD patients, as the majority of COPD patients participate in pulmonary rehabilitation courses when stable.
The main objective of this review was to identify and review RCTs and observational studies which examined whether pulmonary rehabilitation reduces hospitalizations for AECOPD, by assessing the effects of this area of physical therapy on all patients with a defined diagnosis of COPD, rather than only those who have recently exacerbated.
Materials and methods
We used methods as outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement.12A protocol was devised for the search strategy and is included in the supplementary materials.
Types of studies, participants and outcomes
Searches were conducted for RCTs and non-randomised cohort studies comparing the effects of pulmonary rehabilitation with usual care on ER visits and hospitalizations for AECOPD. Studies comparing hospitalizations within individuals before and after the pulmonary rehabilitation programme were also included. Only studies that included participants with a confirmed diagnosis of COPD of any severity were reviewed. Pulmonary rehabilitation programmes of any duration from inpatient, outpatient or community settings were assessed. The primary outcomes were unscheduled hospitalizations and ER visits for AECOPD.We also searched for studies reporting exacerbations recorded by GP visits and mortality rates as secondary outcomes.
Electronic searches
MEDLINE, MEDLINE In-Process & Other Non-Indexed Citations, EMBASE, BIOSIS & Science Citation Index were searched up until September 2015 using PRISMA guidelines.12 The search strategy was devised to search all articles relating the terms “COPD”, “pulmonary rehabilitation”, “hospitalizations” and “mortality”. Terms were combined using the AND Boolean logic operator. The search strategy used in each database is included in the supplementary materials.
Screening
Titles and abstracts were screened and studies deemed to be relevant based on the inclusion/exclusion criteria were then reviewed in full by two reviewers. The eligibility criteria were as follows:
Inclusion criteria:
·Population: COPD patients (mild, moderate, & severe)
·Intervention: pulmonary rehabilitation (home- , community-, or hospital-based). Included rehabilitation on stable COPD and after AECOPD.
·Comparison: Compared pulmonary rehabilitation versus control (or usual care), and studies that compared hospitalizations before and after rehabilitation.
·Outcomes: physician and ER visits with AECOPD as the primary reason for admission/visit according to the International Classification of Disease (ICD): ICD-9 and ICD-10 codes. Mortality rates were also considered as a secondary outcome.
·Setting/Design: Observational studies and RCTs. We included the former as the use of high quality observational and electronic health records data is becoming increasingly accepted.
Exclusion criteria:
·Studies that include all respiratory diseases combined with COPD in the statistical analysis.
·Those that include asthma (ICD-9 493; ICD-10 J45) in the study cohort.
·Programmes that included interventions from other healthcare professionals such as pharmacy and dietetics so that it was unclear if pulmonary rehabilitation was the intervention being assessed. However pulmonary rehabilitation programmes that included education on these topics were still included.
·Studies which had a large proportion of missing data, performed inadequate statistical analysis, or lacked COPD admission/hospitalisation data.
·References by the same author that repeated results of the same study.
·Not in English or translated.
·Conference abstracts lacking full details of the methods, interventions and analysis.
Statistical methods
Analyses used Version 14 of the Stata statistical software.13We used Poisson regression models with Huber variances clustered by study, with combinations of study and treatment (pulmonary rehabilitation or control) as observational units, patient-years at risk as the time-at-risk variable, and incident count as the outcome variable. The regression model parameters were a mean rate per person-year for each combination of study type (RCT, cohort, or before and after) and treatment (rehabilitation or control). In an alternative parameterisation of the same model, we estimated control rate and pulmonary rehabilitation /control rate ratio. We also estimated a mean rate difference for each study type, and a combined grand mean rate difference for all study types, using the add-on Stata package scenttest,14 which is a front-end for the Stata margins command. We produced pseudo-forest plots, in which the data points were rates or rate differences, for the individual studies, for each study type and for all study type. These pseudo-forest plots were different from forest plots in that there were no confidence limits for the rates, rate ratios and rate differences for the individual studies, as the original publications did not always contain sufficient information to estimate these, although they did give total incident counts and total person-years at risk. We used these methods because the individual studies frequently gave only total incident counts and total person-years at risk, without giving any information about the distribution of incidents among individual patients. This implied that we could not estimate overdispersion orunderdispersion, and therefore implied that we could not compute confidence intervals for single-study incidence rates and incidence rate ratios that were robust to overdispersion and/or underdispersion. However, by making individual studies the primary sampling units, making study-treatment combinations the observational units, and using clustered Huber variances, we could make inference about mean population rates in the population of studies, and about their differences and ratios. Note that the confidence limits were asymmetric, because of the log link transformation used in the Poisson regression.
Assessment of risk of bias in included studies
Included studies were individually assessed for the level of risk of bias according to Cochrane guidelines15 across the main domains including random sequence generations, allocation concealment, blinding of participants, blinding of outcome, attrition bias and reporting bias. Cohort studies were assessed using the Newcastle-Ottowa scale16.
Figure 1. Flow chart showing the screening of articles
1
Table 1: Detailed description of studies involving patients who were post-AECOPD
Reference / Methods / Participants / Characteristics / Interventions / OutcomesBehnke 2003 17 / RCT: Examined the effects of an 18-month home-based training on the rate of hospitalizations in Germany. COPD patients were recruited as in-patients admitted to a specialised hospital because of an exacerbation. / 26 patients randomised into a training (n = 14) and a control group (n = 12) . / Severe COPD patients recovering post- exacerbation / 1)Training group performed a 10-day hospital-based supervised training, while those of the control group did not.
2)Home-based training: training group were instructed to practice 15 min of walking at home 3 times per day, in each session achieving 125% of the best 6-minute treadmill distance at discharge. Control group did not receive specific instructions for exercise, neither during the hospital stay nor at home. The home-based programme was performed over 18 months, extending the initial 6-month trial, where only exercise performance and quality of life had been evaluated. / Total numbers over 18 months were 14 admissions in the control group and 3 admissions in the training group, which were significantly different (p = 0.026). The 6-min treadmill distance was, on average, higher (p<0.001) in the training than in the control group. Training group = 518m (95% CI 438; 597); control = 208m (148; 270).
Boxall 2005 18 / RCT: to evaluate the effectiveness of a 12-week home-based PR program. / Intervention group = 23, control group = 23
Age >60. / Housebound patients with severe COPD who were 2-weeks free from worsening symptoms. / Intervention group: 12 week home-based PR programme. Patients received an individually tailored supervised walking and arm exercise program as well as individual multidisciplinary education sessions on COPD management. Included walking and arm exercises once daily. 1 weekly visits from physiotherapist for first 6 weeks and then fortnightly visits until week 12. During the control phase, patients did not receive any treatment in addition to usual medical care. These patients were offered the pulmonary PR after 12 weeks. / Both groups demonstrated improvements in 6MWT (intervention vs control mean, 39.0 vs 4.1 m [23.9% vs. 2.8%]; P = .023). Total scores in the St George’s respiratory questionnaire also improved in both groups (-5.9% vs. -1.4%; P = .020)
Analysis of hospital admission rates 3 months after randomization reveals that 5 patients in each group were admitted to the hospital due to exacerbation (intervention vs control [n], 5 vs 5; P = 1.00)
Eaton 2009 19 / RCT: PR versus usual care on health care utilisation.
Acute health-care utilization at 3 months defined as the number of COPD-related readmissions and unscheduled emergency visits. Data were obtained from hospital and primary health-care records and reconciled with patient home diary records by assessors blinded to the intervention allocation. / Randomised into early rehab = 39 patients, Usual care = 45 / COPD patients who had been admitted with an exacerbation. / Intervention group had inpatient and outpatient PR. Inpatient rehab: 30 min daily exercise including walking and upper & lower limb strengthening, commenced as early as medically appropriate.
Outpatient PR: 1 hour twice weekly for 8 weeks including supervised exercise training and education.
Usual care participants were reviewed by a dedicated study COPD nurse to ensure that they received standardized care in accordance with COPD guidelines. They also received standardized advice on the proven benefits of exercise and maintaining daily activities. / 19 (40%) patients assigned to early rehabilitation satisfied the a priori definition of adherence (attendance at 75% of rehabilitation sessions).
The rehabilitation group demonstrated a 23% (95% CI: 11–36%) risk of readmission at 3 months, compared with 32% (95% CI: 19–45%) for the usual care group; the absolute risk reduction was 9% (95% CI: -9.5–26%).
Ko 2011 20 / RCT comparing PRP with usual care 2424. Accident & Emergency Department (AED) visits and hospitalizations in the subsequent 12 months were obtained from the patients, with verification of their medical records. / 60 patients (30 in PRP group and 30 in Usual Care) / Patients who had been admitted with AECOPD were referred to examine effects of early outpatient PRP. / PRP 3 times a week for 8 weeks (2 hours per session) started in the same week as the baseline assessment. Supervised exercise training, including the use of treadmill, arm cycling, and arm and leg weight training, intensity adjusted by the physiotherapist, based on the tolerability and physiological variables (oxygen saturation and heart rate) of the patient. Patients were also advised to perform home exercises for at least 20 min a day.
UC group: subjects were seen by the nurse specialist at the baseline assessment as for the PRP group. They were not offered any training sessions by the physiotherapist but given simple instructions to have regular exercise at home, including walking every day and also performing some muscle stretching exercise. / Mean number of A&E attendances at 12 months
PR = 1.57 ± 1.57
Usual care = 1.47 ± 2.64
Mean number of re-admissions at 12 months
PR = 1.00 ± 1.20
Usual care = 1.03 ± 1.87.
There was a trend towards fewer readmissions in the PRP group than the UC group at the first 3 months, but this effect was lost over time and there was no difference between the two groups at 12 months (P = 0.67)
Man 2004 21 / RCT: Compared early rehab with usual care. / 42 COPD patients randomized into usual care (n = 16), Early rehab (n = 18) / Patients admitted with an acute exacerbation of COPD. / Early rehab: 2 classes per week for 8 weeks. Each class lasted two hours, consisting of one hour of exercise (aerobic walking and cycling, strength training for the upper and lower limb) and one hour of educational activities. Patients also received individualised home exercise programmes, which encouraged at least 20 minutes of exercise per day.
All admitted patients received standard treatment, including nebulised bronchodilators, oxygen, oral or intravenous antibiotics, non-invasive ventilation (if required), and a one to two week course of oral prednisolone (30-40 mg daily). / Hospital readmission rate at 3 months: usual care = 57.1%; early rehab = 35.0%
Mean difference between groups (95% CI): incident rate ratio = 0.66 (0.3 to 1.5) P = 0.30
A&E visit rate at 3 months: usual care = 42.9%; early rehab = 10%
Mean difference between groups (95% CI): incident rate ratio = 0.14 (0.03 to 0.65) p = 0.01
Murphy 2005 22 / RCT: compared PR with control. / 31 patients admitted to a COPD home from hospital treatment programme were randomised into an exercise group (n = 16; FEV1 0.94 ± 0.34 L) and a control group (n = 15; FEV1 1.08 ± 0.33 Litres).
Subjects had moderate to severe COPD (FEV1<60%) and a long history of cigarette smoking. / Patients with moderate or severe COPD after discharge from hospital following an exacerbation. / PR twice a week for 6 weeks. Each subject received 12 supervised exercise sessions in total. The exercise sessions lasted 30–40 min including rests. The patient was given a home exercise diary and was instructed to exercise for at least 15 min on the other days. Aerobic exercises included stepping up and down a stair and sitting to standing from a chair. Upper limb muscle groups were exercised using a Thera-Band.
Subjects in the control group received their standard medical treatment without any form of rehabilitation exercises or lifestyle changes advice. / At 3-month follow-up, three of the control group and none of the exercise group had experienced subsequent exacerbation of COPD (P =0:06). At 6 months, 5 of the control group and 2 of the active group had exacerbations (P = 0:1).
Rasekaba 2009 23 / Cohort study of COPD patients who were eligible to participate in the chronic disease management pulmonary rehab (CDM) program. Compared CDM-PR Cohort and Opt-out Cohort between for acute hospital care utilization 12 months before and after the introduction of CDM-PR. / CDM-PR Cohort n = 29, Opt-out Cohort n = 24.
Patients with confirmed spirometry COPD diagnosis (FEV1/FVC<70%) and absence of physical limitation to participation in exercise were included. / Patients were referred for CDM by GPs, respiratory specialists or acute hospital services during an emergency admission. / The CDM-PR Program was a once-weekly, 8-week programme consisting of 1 hour of physiotherapy and 45 minutes of education.
The exercise sessions consisted of a warm-up, rotating around a seven-station circuit of aerobic and resistance training for both the upper and lower body, followed by a cool-down.
Education topics: exercise, respiratory system and COPD, diet and nutrition, psychological effects and stress management, medications, energy conservation. / Emergency department admissions
12 month before
CDM-PR (n = 29) = 17.11
Opt out group ( n= 24) = 11.04 (p = 0.415)
12 month after
CDM-PR (N = 29) = 2.03
Opt out group (n = 22) = 14.96 ( p = < 0.001)
Revitt 2013 24 / Cohort study not included in meta-analysis. Data were collected from patients who attended an outpatient PR assessment at Glenfield Hospital, Leicester, from November 2006 to October 2008. All patients had been referred following an admission for an acute exacerbation of COPD (AECOPD). / One hundred sixty (87 males) patients with COPD. Readmission data for 155 patients / Patients who were referred by the Early Discharge Service after an admission for an exacerbation. / PR 2 x / week for 4 weeks. Patients continued with a 3-week structured unsupervised home exercise programme. Each session lasted for 2 hours with 1 hour of supervised exercise training and 1 h of education. / Incidence rate ratio = 0.623 (95% CI: 0.462–0.840) (P = 0.002). There was a 37.7% reduction in incidence of admissions in year post PR compared to the year prior to PR.
Seymour 2010 25 / RCT: compared PR with usual care. / COPD patients with a ratio of FEV1 to FVC of <0.7. Usual care (n=30), PR (n=30) / Patients who had been admitted for an exacerbation of COPD and were randomized to receive either PR or usual care. / Participants were incorporated into standard PR classes consisting of twice-weekly exercise and education sessions (each lasting 2 hour) for a period of 8 weeks. Exercise was a mixture of limb strengthening and aerobic activities, tailored to individual baseline function. Patients in both the UC and PR groups were provided with general information about COPD prior to randomization and offered outpatient appointments with their general practitioner or respiratory team. / Hospital or ED attendance for exacerbation
UC (N=30) = 17 (57%) PR (N=30) = 8 (27%)
Odds Ratio (95% CI) PR vs UC* = 0.28 (0.10 to 0.82) p = 0.02
Table 2: Detailed description of studies involving stable COPD patients