Respiratory Medicine (2014) 108, 426e437

Available online at www.sciencedirect.com

journal homepage: www.elsevier .com/locate /rmed

REVIEW

Complex interventions that reduce urgentcare use in COPD: A systematic review withmeta-regression

Chris Dickens a, Wayne Katon b, Amy Blakemore c,d,*,Angee Khara c, Barbara Tomenson e, Ashley Woodcock f,Anna Fryer c, Else Guthrie c

a Institute of Health Service Research, University of Exeter Medical School and Peninsula Collaborationfor Leadership in Health Research and Care (PenCLAHRC), Universities of Exeter, Veysey Building,Room 007, Salmon Pool Lane, Exeter EX2 4SG, UKbDepartment of Psychiatry & Behavioral Sciences, University of Washington, Seattle, WA 98195-6560,USAcDepartment of Psychiatry, Manchester Mental Health and Social Care Trust, Rawnsley Building,Manchester Royal Infirmary, Oxford Road, Manchester M13 9WL, UKdCentre for Primary Care, Institute of Population Health, University of Manchester, WilliamsonBuilding, Oxford Road, Manchester M13 9PL, UKeBiostatistics Unit, Institute of Population Health, University of Manchester, Jean McFarlane Building,Oxford Road, Manchester M13 9PL, UKf Institute of Inflammation and Repair, University of Manchester, 2nd Floor Education and ResearchCentre, University Hospital of South Manchester NHS Foundation Trust, Southmoor Road, ManchesterM23 9LT, UK

Received 21 March 2013; accepted 25 May 2013Available online 24 June 2013

KEYWORDSChronic obstructivepulmonary disease;Long-term conditions;Complex

Abbreviations: COPD, chronic obstruinterval.* Corresponding author. Department

Infirmary, Oxford Road, Manchester ME-mail addresses: amy.blakemore@

0954-6111/$ - see front matter ª 201http://dx.doi.org/10.1016/j.rmed.20

Summary

Context: Chronic obstructive pulmonary disease is common and accounts for considerablehealthcare expenditure. A large proportion of this healthcare expenditure is attributable tothe use of expensive urgent healthcare. The characteristics of interventions that reduce theuse of urgent healthcare remain unclear.

ctive pulmonary disease; LTCs, chronic physical health problems; OR, odds ratio; CI, confidence

of Psychiatry, Manchester Mental Health and Social Care Trust, Rawnsley Building, Manchester Royal13 9WL, UK. Tel.: þ44 0161 276 5380; fax: þ44 0161 273 2135.nhs.net, amy.blakemore@mhsc.nhs.uk (A. Blakemore).

3 Elsevier Ltd. All rights reserved.13.05.011

Interventions that reduce urgent healthcare in COPD 427

intervention;

Urgent healthcareutilisation;Systematic review;Meta-regression

Objective: To examine the characteristics of complex interventions intended to reduce the useof urgent and unscheduled healthcare among people with COPD.Data sources: Electronic searches of MEDLINE, EMBASE, PSYCINFO, CINAHL, the British NursingLibrary and the Cochrane library, from inception to 25th January 2013 were conducted. Thesewere supplemented by hand-searching bibliographies and citation tracing identified reviewsand eligible studies.Study selection: Studies were eligible for inclusion if they: i) included adults with chronicobstructive pulmonary disease, ii) assessed the efficacy of a complex intervention using rando-mised controlled trial design, and iii) included a measure of urgent healthcare utilisation atfollow-up.Data extraction: Data on the subjects recruited, trial methods used, the characteristics ofcomplex interventions and the effects of the intervention on urgent healthcare utilisationwere extracted from eligible studies.Results: 32 independent studies were identified. Pooled effects indicated that interventionswere associated with a 32% reduction in the use of urgent healthcare (OR Z 0.68, 95%CI Z 0.57, 0.80). When study effects were grouped according to the components of the inter-ventions used, significant effects were seen for interventions that included general education(OR Z 0.66, 95% CI Z 0.55, 0.81), Exercise (OR Z 0.60, 95% CI Z 0.48, 0.76) and relaxationtherapy (OR Z 0.48, 95% CI Z 0.33, 0.70).Conclusions: Use of urgent healthcare in patients with COPD was significantly reduced by com-plex interventions. Complex interventions among people with COPD may reduce the use of ur-gent care, particularly those including education, exercise and relaxation.ª 2013 Elsevier Ltd. All rights reserved.

Contents

Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428

Data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428Risk of bias in individual studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429Statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430

Details of studies included . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430Details of patient populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430Details of the intervention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430

Publication bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434

Information on access to data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Disclosure of potential conflicts of interest, activities, relationships and affiliations . . . . . . . . . . . . . . . . 434Sources of funding and support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Role of the sponsor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Information on previous presentations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Listing of supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Contributions of others compensated for their contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435

Background

COPD is an important cause of morbidity and mortalityworldwide, it was the fifth leading cause of death in 2002and is predicted to be the fourth by 2030.1 It is associated

with severe impairment in quality of life, disability andincreased healthcare use.2,3 A recent cross country, crosssectional survey has shown that, while the greatest pro-portion of healthcare use was in primary care, hospitalinpatient care accounted for 68% of total COPD healthcare

428 C. Dickens et al.

costs. The cost was the highest in the UK at £277 per pa-tient, per month.2

A considerable proportion of the healthcare cost ofpeople with COPD is attributable to the use of expensiveunscheduled and urgent healthcare (henceforth urgenthealthcare) rather than purposefully designed routine ser-vices. Two previous studies found that as many as 26% ofCOPD patients will make emergency department visitsannually and that 13% of patients will make 6 or morevisits.4,5 COPD is associated with high prevalence ofdepression (26%) and anxiety (10e55%).6,7 Among peoplewith long term conditions (LTCs), anxiety and depressionare closely associated with poorer health outcomes,8,9

greater healthcare utilisation10,11 and increased use of ac-cident and emergency departments.12,13 In COPD inparticular, depression and anxiety have been shown to besignificantly associated with increased risk of emergencydepartment visits, hospitalisations and higher urgent carecosts,14e22 though it remains unclear whether the influenceof anxiety and depression is causal or not.23

Current UK government policy recommends that un-scheduled use of care should be reduced in people withLTCs by introducing alternative pathways in primary care24;COPD is one of several LTCs which are cited as beingpotentially amenable to early prevention or interventionwithin the community as an alternative to emergencyadmission.25 Previous studies have shown that interventionscan reduce use of urgent healthcare, though these in-terventions have usually involved multiple components andtherefore can be considered complex. They also vary withregards to the intensity and setting in which interventionshave been delivered and professionals involved. As a resultthe effects of such complex interventions have beenvery mixed among people with COPD, and the componentsof interventions associated with a reduction in urgenthealthcare remain obscure.

We have conducted a systematic review of the literaturewith meta-regression to identify the characteristics ofcomplex interventions that reduce the use of urgenthealthcare among people with COPD. Identification of suchcharacteristics would facilitate the design of optimal in-terventions with potential to reduce use of urgent health-care and thus result in considerable savings in healthcareexpenditure. We did not limit the studies included on thetypes of complex interventions delivered or by the type ofurgent healthcare used, to ensure capture of maximumrelevant data.

g Since we anticipated differing thresholds for hospital admissionsbetween countries and over time, for the purposes of this reviewhospitalisations were only considered urgent if they were describedas such in the published paper, if hospitalisations were described asbeing the result of an acute exacerbation of underlying LTC or ifresearchers confirmed that hospitalisations were urgent.

Method

Studies were eligible for inclusion in the review if they metthe following criteria.

a) Included adults with COPD.b) Assessed the efficacy/effectiveness of a complex inter-

vention. For the purposes of this review, complex in-terventions involved multiple components and/ormultiple professionals, and could be delivered on an in-dividual or group basis, or using technology such astelephone or computer. Interventions could include ed-ucation, rehabilitation, psychological therapy, social

intervention (social support, support group), organisa-tional intervention (such as collaborative care or casemanagement), also drug trials targeting a psychologicalproblem, e.g. anxiety or depression. Simple in-terventions, such as the introduction of a new treatmenttargeting the underlying long term condition, comparedto treatment as usual were not included in this review.

c) Assessed urgent healthcare use as an outcome, e.g.emergency department visits, urgent hospitalisationg orunscheduled GP visits.

d) Used randomised controlled trial design.

Studies were not excluded by date or language of pub-lication, sample size or follow up period. See Onlineappendix (pages 2 to 3) for full PICO criteria. Unpublishedstudies and those published in abstract form only, were notincluded in this review.

Electronic search strategies were developed in-team, inconsultation with librarians with experience of performingsystematic reviews (RM). Search strategies were peerreviewed by experts from within the University (LG) andmodified accordingly. Search strategies included terms rele-vant to prospective studies and also COPD, with furtherlimiting to randomised controlled trials of complex in-terventions in COPD using hand searching (see pages 4e17 inOnlineappendix fordetails of searchstrategiesused).Wewerenot able to develop sensitive and reliable strategies to identifystudies investigating use of urgent healthcare specifically, sosearches were developed to identify all healthcare use, andfurther restriction to relevant papers was achieved by handsearching potentially eligible papers.

Searches were conducted in MEDLINE, EMBASE, PSY-CHINFO, CINAHL, The British Nursing Index (using the OVIDsearch interface) and the Cochrane Library, from inceptionof each database. Electronic searches were first conductedon 19th August 2008 and then updated on 1st December2011 and again on 25th January 2013. Electronic searcheswere further supplemented by hand searches of the refer-ence lists of papers meeting PICO criteria and relevant re-views identified through searching electronic databases. Alltitles and abstracts of papers identified by these meanswere screened by two researchers (AB, AK) to identifystudies potentially meeting the inclusion criteria, withdisagreements being resolved by discussion. Full text re-ports of studies that were potentially relevant to this re-view were screened by two researchers (AB, AK) todetermine eligibility. To avoid double counting studies,findings for any population presented in multiple publica-tions were included only once in this review. See Fig. 1 for asummary of the study selection process.

Data extraction

Standardised electronic data extraction sheets weredeveloped by the team and modified after piloting on the

Figure 1 PRISMA flow diagram.

Interventions that reduce urgent healthcare in COPD 429

first 5 papers. Data were extracted for characteristics ofthe participants, the characteristics of the intervention,the methodological characteristics of the study and theeffects of the intervention on the use of urgent healthcare.

The characteristics of the complex interventions werecoded according to the following 11 key characteristics,that were generated a priori26: general education, generaldiscussion, skills training, exercise, behaviour therapy,relapse prevention, problem solving, cognitive behaviouraltherapy, social support, relaxation and biofeedback (seePage 2 of Online appendix for working definitions of inter-vention components). Intervention components not fittingthe description required for the above categories ornot described in sufficient detail were recorded as“miscellaneous”.26

Primary data extractionwas performedby two researchers(AK, AB), with discussion between researchers and anothermember of the team (CD) where there was uncertainty in anyaspect of data interpretation or extraction.

Risk of bias in individual studies

The methodological quality of the individual psychologicalintervention trials was assessed using a componentapproach27 by assessing whether:

i) the allocation sequence adequately generated(e.g. random number lists, computer generation,tossing a coin etc),

ii) used adequate methods to conceal treatment alloca-tion from researcher recruiting patients,

iii) knowledge of the allocated intervention wasadequately prevented during the study,

iv) incomplete outcome data was adequately dealt with,v) reports of the study were free from suggestion of se-

lective outcome reporting,vi) the study apparently was free of other problems that

could put it at risk of bias.

For the meta-analysis, a binary measure of quality wasused, determined by whether concealment of treatmentallocation was used or not.28

Quality for each study was assessed by two researchers(AB, AK) and any uncertainties were resolved by discussionwithin the team (AB, AK, CD).

Statistical analysis

Odds ratios (ORs) and 95% confidence intervals (95% CIs)were calculated for each study where number of subjectsusing urgent healthcare and the total number of subjects ineach trial arm were presented.27 The odds ratio is the ratioof the odds of urgent healthcare in intervention group tothe odds of urgent healthcare in control group.

An OR equal to 1 would indicate that there is no dif-ference in effect between the intervention and controlgroup whereas an OR <1 indicated that the interventionreduced the use of urgent healthcare.27 Where data were

430 C. Dickens et al.

presented in alternative formats, for example where sum-mary tests results or where continuous data were pre-sented, appropriate transformations were made.29

In cases where follow-up data were collected at mul-tiple time points, ORs were calculated for the follow-updata collected nearest to 1 year to maximise consistencyacross studies. Where studies included more than onemeasure of urgent healthcare, effects for each measurewere pooled. Where studies included more than 1 inter-vention group, the data for each was entered as separaterecords and the sample size in the control group washalved. Effects of interventions were combined acrossindependent studies using random effects models,weighted using the inverse of the variance.30 Heteroge-neity among studies was assessed using Cochrane Q, wherea large Q value indicates the presence of heterogeneity,and the I2 statistic. The I2 gives the percentage of vari-ability in the effect estimate that is due to heterogeneityrather than to chance. Suggested thresholds for theinterpretation of I2 are as follows: less than 40% indicatethere is no problem with heterogeneity, 30e60% indicatesa moderate problem, 60e90% a substantial problem and75% and over considerable heterogeneity.27,31 Publicationbias was assessed using funnel plots and Eggers’ regressionmethod.27,32,33 Funnel plots are a scatter plot of the effectestimates from each individual trial included in the reviewplotted against their relative sample size. If publicationbias is present the results of smaller studies with non-significant results will not be present in the graph andthere will be a gap at the bottom of the plot. However, ifpublication bias is not present the plot will look like asymmetrical, inverted funnel.27,32

Differences in effect across the methodological charac-teristics of the trials including, i) the features of the studypopulation, ii) the methods of delivering the interventionand iii) the methods of the trial itself, were assessed usingthe analogue to Analysis of Variance for categorical vari-ables,29 and univariate meta-regression was used forcontinuous variables. Meta-regression is a type of sub-analysis which can be used where studies share the samecharacteristics, for example the same intervention com-ponents, and therefore may yield different results fromother studies with different intervention compo-nents.27,34,35 To identify which intervention componentswere independently associated with reductions in urgenthealthcare, intervention components were entered into arandom effects multivariate meta-regression.34,35

Effects for interventions are presented in text, tablesand forest plot.

Meta-analyses were performed using ComprehensiveMeta-analysis (version 2.2.048, Nov 7th 2008) and Stata(version 12, StataCorp LP, Texas, US).

Results

Details of studies included

Thirty-two independent studies met criteria for inclusion inthis review that included 3941 patients with COPD.36e67

Sample sizes for randomised controlled trials varied from26 to 554 subjects, with an average sample size of 123. Nine

studies provided information on 2 different measures ofurgent care, and one provided information on 3. One studyhad 2 intervention groups, resulting in a total of 43 recordsfrom which to estimate the effect. Twenty-four recordsassessed attendance at the emergency department, 12assessed urgent hospitalisation, 7 assessed other sources ofurgent care (such as emergency visits to primary or sec-ondary care, non-specified urgent doctor visits, ambulancecalls etc). Ten studies obtained unscheduled care data frompatient records, 5 from self-reports, 11 used a combinationof sources, and for 7 studies the sources of urgent care datawere unclear. Length of follow-up varied from 1 to 24months (mean Z 10.1 months) (See eTable 1 in Onlineappendix for details of the methodological characteristicsof the studies included).

Details of patient populations

Two studies recruited male only samples and the remainder(where stated) included mixed sex samples. Mean ages ofpopulations varied from 57.5 to 80.4 years. In 24 treatmentcomparisons, patients were recruited from secondary caresetting, in 5 primary care, and in 4 interventions patientswere recruited from a combination of primary and sec-ondary care settings (see eTable 1 in Online appendix forcharacteristics of study populations).

Details of the intervention

With regard to the characteristics of the interventions, themean number of treatment components included withineach intervention was 2.9 (range 1e6 components). Theaverage number of treatment sessions (stated in 22 studies)was 13.9 (range 1e60); in 9 studies the exact number ofadditional health practitioner contacts associated with theinterventions were unclear, most frequently because thenumber of contacts was flexible.

Treatment was delivered at home or in the community in12 studies, in hospital or in a doctor’s clinic in 9 cases, andin a combination of home/community and hospital/clinic in9 cases. Treatment was delivered through face to facecontact in 19 studies, by telephone in 1 and a combinationof these in 13 cases. The intervention was delivered by anon-mental health practitioner in 29 studies and a mentalhealth practitioner in 1 study. Treatment was delivered by amultidisciplinary team in 18 studies and an unidisciplinaryteam in 11 studies. Of the treatment comparisons included,24 used a structured management plan, 28 includedscheduled follow-up, 11 included enhanced inter-professional communications (see eTable 2 in Onlineappendix for details of interventions).

Reports of the risk of bias in the individual studies var-ied; 11 studies reported details of allocation concealment,of which 8 described adequate methods of allocationconcealment and were considered to be of high methodo-logical quality for the purposes of the meta-analysis (seeeTable 3 in Online appendix for details of study quality).

Effect sizes for individual studies are presented in theforest plot in Fig. 2. Overall the combined effect acrossthe 33 comparisons indicated that interventions wereassociated with a reduction in the use of urgent care

NOTE: Weights are from random effects analysis

Overall (I-squared = 37.4%, p = 0.017)

Tougaard et al

McGeoch et alRea et alRice et al

Hernandez et al

Bourbeau et al

Jarab et al

Boxall et al

Sridhar et al

Casas et al

Gallefoss et al

Seymour et al

Wong et al

Trappenburg et al

Gadoury et al

Koff et alKo et al

Bergner et al

Guell et al

Martin et al

Farrero et alEaton et al

Hermiz et al

Ries et al

Author

Man et al

Lee et al

Khdour et al

Smith et al

Wakayabayashi et al

Lobato et al

Wood-Baker et al

Bergner et al

Soler et al

1992

200620042010

2003

2003

2012

2005

2007

2006

2000

2010

2005

2011

2004

20092011

1988

2000

Year of

2004

20012009

2002

2003

publicatin

2004

2002

2009

1999

2011

2005

2006

1988

2006

0.68 (0.57, 0.80)

0.39 (0.17, 0.86)

1.24 (0.70, 2.21)0.41 (0.12, 1.37)0.68 (0.50, 0.92)

0.37 (0.17, 0.81)

0.43 (0.24, 0.77)

0.68 (0.37, 1.26)

1.00 (0.25, 4.06)

0.49 (0.24, 1.00)

0.63 (0.35, 1.12)

0.75 (0.15, 3.73)

0.34 (0.08, 1.45)

0.36 (0.14, 0.93)

1.00 (0.62, 1.60)

0.66 (0.39, 1.10)

0.30 (0.03, 3.14)1.02 (0.41, 2.56)

0.72 (0.37, 1.41)

0.42 (0.16, 1.06)

Odds

2.34 (1.09, 5.03)

0.32 (0.15, 0.67)0.79 (0.35, 1.78)

0.28 (0.06, 1.35)

1.00 (0.54, 1.84)

ratio (95% CI)

0.16 (0.03, 0.94)

0.99 (0.46, 2.11)

0.48 (0.24, 0.96)

2.76 (0.95, 7.98)

0.66 (0.30, 1.43)

0.09 (0.00, 1.78)

0.95 (0.49, 1.87)

0.78 (0.40, 1.53)

0.50 (0.12, 2.05)

100.00

3.10

4.561.687.17

3.15

4.49

4.22

1.30

3.64

4.52

1.03

1.24

2.49

5.47

5.05

0.502.56

3.85

2.50

%

3.28

3.353.04

1.05

4.29

Weight

0.86

3.33

3.74

2.05

3.23

0.32

3.85

3.81

1.29

0.68 (0.57, 0.80)

0.39 (0.17, 0.86)

1.24 (0.70, 2.21)0.41 (0.12, 1.37)0.68 (0.50, 0.92)

0.37 (0.17, 0.81)

0.43 (0.24, 0.77)

0.68 (0.37, 1.26)

1.00 (0.25, 4.06)

0.49 (0.24, 1.00)

0.63 (0.35, 1.12)

0.75 (0.15, 3.73)

0.34 (0.08, 1.45)

0.36 (0.14, 0.93)

1.00 (0.62, 1.60)

0.66 (0.39, 1.10)

0.30 (0.03, 3.14)1.02 (0.41, 2.56)

0.72 (0.37, 1.41)

0.42 (0.16, 1.06)

Odds

2.34 (1.09, 5.03)

0.32 (0.15, 0.67)0.79 (0.35, 1.78)

0.28 (0.06, 1.35)

1.00 (0.54, 1.84)

ratio (95% CI)

0.16 (0.03, 0.94)

0.99 (0.46, 2.11)

0.48 (0.24, 0.96)

2.76 (0.95, 7.98)

0.66 (0.30, 1.43)

0.09 (0.00, 1.78)

0.95 (0.49, 1.87)

0.78 (0.40, 1.53)

0.50 (0.12, 2.05)

100.00

3.10

4.561.687.17

3.15

4.49

4.22

1.30

3.64

4.52

1.03

1.24

2.49

5.47

5.05

0.502.56

3.85

2.50

%

3.28

3.353.04

1.05

4.29

Weight

0.86

3.33

3.74

2.05

3.23

0.32

3.85

3.81

1.29

Negative Positive

1.1.2.512

Figure 2 Effect sizes and 95% confidence intervals for included studies.

Interventions that reduce urgent healthcare in COPD 431

(OR Z 0.68, 95% CI Z 0.57, 0.80). A moderate degree ofheterogeneity was seen across studies (Q Z 51.1 df Z 32,p < 0.0005, I2 Z 37.4%).32

Variations in effect across methodological characteris-tics is seen in Table 1. There was a non-significant trend foreffects to be greatest among studies that had recruitedsubjects from secondary care (p Z 0.10).) The effect sizesof the interventions did not vary significantly with the meanage of patients, the gender mix, where patients wererecruited, how or where treatments were delivered, whodelivered treatment, how it was organised or study quality(Table 2).

When study effects were grouped according to thecomponents of interventions used, significant effects onthe use or urgent care were seen for interventions that

included general education (OR Z 0.66), Exercise(OR Z 0.60) and relaxation therapy (OR Z 0.48). Therewere non-significant trends for use of urgent care to bereduced by interventions that included skills training(ORZ 0.72) and relapse prevention (ORZ 0.77). No studieshad used behavioural interventions, CBT, problem solvingor had focussed on increasing social support. Among studiesof higher methodological quality (i.e. that concealedtreatment allocation) effect sizes were comparable foreducation, exercise and relaxation and skills training: ed-ucation (9 studies, pooled OR Z 0.65, p Z 0.004), exercise(6 studies, pooled OR Z 0.65, p Z 0.069), relaxation (3studies, pooled OR Z 0.52, p Z 0.012) and skills training (2studies, pooled OR Z 0.55, p Z 0.14). Effects for relapseprevention were small and non-significant (2 studies,

Table 1 Impact of methodological characteristics on effects sizes.

n OR (95% CI) p Comparison across groups

Nature of patients recruited

Mean age of subjects 33 0.01 (�0.04, 0.07) 0.587Percentage male 32 �0.005 (�0.01, 0.004) 0.28Patient recruitment

Primary care 5 0.89 (0.52, 1.52) 0.67Secondary care 24 0.59 (0.50, 0.70) <0.0005 Q Z 4.6, d.f. Z 2, p Z 0.10Combined 4 1.06 (0.55, 2.04) 0.86Characteristics of intervention

Where delivered

Hospital/clinic 9 0.61 (0.43, 0.86) 0.005Home/community 12 0.69 (0.50, 0.97) 0.031 Q Z 0.39, df Z 2, p Z 0.82Combined 9 0.70 (0.50, 0.99) 0.043Mode of delivery

Face-to-face 19 0.72 (0.54, 0.96) 0.023Telephone 1 0.36 (0.14, 0.93) 0.034 Q Z 1.92, df Z 2, p Z 0.38Combined 13 0.66 (0.53, 0.82) <0.0005Number of intervention sessions 22 0.003 (�0.011, 0.017) 0.64Delivered by

Non-mental health professional 29 0.66 (0.53, 0.82) <0.0005Mental health professional 1 0.68 (0.50, 0.92) 0.013 Q Z 0.02, df Z 1, p Z 0.89Multidisciplinary team 16 0.70 (0.52, 0.95) 0.021 Q Z 0.02, df Z 1, p Z 0.89Unidisciplinary team 11 0.68 (0.56, 0.83) <0.0005Structured management plan 24 0.66 (0.54, 0.81) <0.0005 Q Z 0.2, df Z 1,p Z 0.69No structured management plan 9 0.72 (0.50, 1.06) 0.093Scheduled follow-up 28 0.66 (0.55, 0.80) <0.0005 Q Z 0.60,df Z 1, p Z 0.44No scheduled follow-up 5 0.78 (0.53, 1.15) 0.21Enhanced inter-professional communication 11 0.84 (0.58, 1.22) 0.37 Q Z 2.4, df Z 1, p Z 0.12No enhanced inter-professional communication 22 0.61 (0.53, 0.72) <0.0005Did intervention constitute collaborative care

Yes 5 0.86 (0.45, 1.65) 0.65 Q Z 0.78,df Z 1, p Z 0.38No 26 0.63 (0.53, 0.76) <0.0005Methodological characteristic

Type of unscheduled carea

Emergency Department 24 0.70 (0.57, 0.85) <0.0005Urgent hospital admissions 12 0.60 (0.48, 0.77) <0.0005 Q Z 0.8, df Z 2, p Z 0.67Others 7 0.65 (0.37, 1.13) 0.13Source of urgent care data

Records 10 0.62 (0.469, 0.83) 0.002Self-report 5 0.67 (0.43, 1.05) 0.082 Q Z 0.1, df Z 2, p Z 0.95Combined 11 0.62 (0.48, 0.81) <0.0005Length of follow-up 33 0.005 (�0.03, 0.04) 0.78Higher study quality 8 0.71 (0.53, 0.96) 0.025 Q Z 1.0, df Z 1, p Z 0.32Lower study quality 5 0.55 (0.37, 0.83) 0.005a There are 43 comparisons for type of unscheduled care from 32 studies.

432 C. Dickens et al.

pooled OR Z 0.94, p Z 0.77). On multivariate meta-regression, no individual treatment components contrib-uted significantly to the regression model.

Publication bias

The funnel plot did appear asymmetrical, with a relativeabsence of small studies in which interventions were asso-ciated with increased use of urgent care (Fig. 3), thoughEgger’s regression method did not confirm a statisticallysignificant association between Log OR and standard errorof Log OR [Egger’s intercept Z �0.79, (95% CI �1.87 to0.29), p Z 0.15].

Discussion

We conducted a systematic review of intervention studiesin people with chronic obstructive pulmonary disease toidentify the characteristics of interventions associated witha reduction in the use of urgent care. We identified 32studies of complex interventions that had investigated useof urgent care as an outcome. On meta-analysis we foundthat use of urgent care was reduced significantly by com-plex interventions (OR Z 0.68). This shows that complexinterventions can reduce the amount of urgent care used byCOPD patients by 32%, though the effects of individualstudies were moderately heterogeneous. When grouped by

Table 2 Types of treatment.

Variable OR (95%confidenceintervals)

p Value

General education (n Z 28) 0.66 (0.55, 0.81) <0.0005General discussion (n Z 3) 0.81 (0.51, 1.28) 0.35Skills (n Z 13) 0.72 (0.49, 1.04) 0.080Exercise (n Z 11) 0.60 (0.48, 0.76) <0.0005Behaviour therapy (n Z 0) e e e

Relapse prevention (n Z 11) 0.77 (0.56, 1.07) 0.12Problem solving (n Z 0) e e e

CBT (n Z 0) e e e

Increased socialsupport (n Z 0)

e e e

Relaxation therapy (n Z 4) 0.48 (0.33, 0.70) <0.0005

Interventions that reduce urgent healthcare in COPD 433

the individual characteristics of the intervention we foundsignificant effects for studies that included education(OR Z 0.66), exercise (OR Z 0.60), and relaxation(OR Z 0.48), which can be interpreted as a 34%, 40% and52% reduction in urgent care use respectively. There werealso non-significant trends for reduction in urgent care usein studies that included skills training (OR Z 0.72,p Z 0.080) and relapse prevention (OR Z 0.77, p Z 0.12).No studies used cognitive behavioural therapy/behaviouraltherapy or problems solving. There was no significant evi-dence of publication bias. Overall, our findings may havebeen inflated by inclusion of studies of low methodologicalquality, though the effects for education, exercise, relax-ation and skills training were comparable in the smallnumber of studies with high methodological quality.

Our review has a number of strengths. First, we con-ducted extensive searches of key electronic databases andalso asked experts in the area about potentially relevantstudies to identify as many relevant studies as possible.Second, rather than rely on the development of electronicsearch strategies to identify papers relevant to theconstruct of urgent healthcare, we kept our electronicsearches broad (i.e. searched all healthcare) and narrowedto the relevant papers using rigorous hand-searching. Thirdwe did not limit the types of urgent healthcare included apriori though we were very rigorous in the exclusion of

0.5

11.

52

-3 -2 -1 0 1 2 3Log Odds ratio

Funnel plot with pseudo 95% confidence limits

Figure 3 Funnel plot for included studies.

studies for which it was not absolutely clear that the use ofhealthcare was urgent. Finally, we did not limit our reviewby the date or language of publication, sample size orduration of follow-up to maximise the number of studiesincluded.

Our study has some weaknesses. First, the effects ofdifferent complex interventions were moderately hetero-geneous, so the pooled effect from all included studiesmust be interpreted with caution. The pooled effectsacross a wide range of complex interventions of varyingintensities, delivered in varying settings by different pro-fessionals tells us little about which interventions might bemost effective in reducing the use of urgent care. Ourintention had always been to explore the extent to whichmethodological characteristics influenced this heteroge-neity, to identify intervention components associated withreduced urgent care. Second, we focused entirely onreduction in use of urgent care and we did not recordmedical outcomes, such as health status, morbidity ofhealth related quality of life. As such we can draw no in-ferences about effects of these complex interventions inthese other domains and cannot determine whether thereductions in the use of urgent care were due to a reductionin the need for urgent care, due to improved health, orsimply due to the substitution of urgent care by scheduledcare, delivered as part of the study intervention.

We interpret our findings as indicating that complex in-terventions among people with COPD may reduce the use ofurgent care, particularly those including education, exer-cise and relaxation. We acknowledge that the inclusion ofstudies of low methodological quality may have inflated thesize of this effect, though the benefits to urgent care fromeducation, exercise, relaxation and skills training remainedsignificant among the minority of studies of high quality.Since no individual study component made a significantindependent contribution in the multivariable meta-regression, we conclude that, based on current evidenceamong people with COPD, reductions in the use of urgentcare cannot be attributed to any single interventioncomponent.

Collaborative care is often suggested as a way to manageindividuals with complex needs, such as LTCs and depres-sion. It is defined as interventions delivered by a multidis-ciplinary team and including structured management plans,scheduled follow up and enhanced inter-professionalcommunication.68 Collaborative care has previously beenshown to make significant improvements to both physicaland psychological health outcomes in patients with dia-betes and coronary heart disease,69e72 post coronary arterybypass graft patients73 and older adults with other comor-bid medical conditions.71,74 Outside of the context of co-morbid medical illness, there is emerging evidence thatcollaborative care for the treatment of depression is costeffective in terms of Quality Adjusted Life Years (QALY)75

and in reducing total healthcare costs.76 However, therehave been no economic evaluations of collaborative carefor treatment of depression comorbid with LTCs to date.The interventions that met the criteria for collaborativecare in this review were not associated with reduced use orcost of urgent care.

Psychological symptoms, especially depression, arecommon in patients with COPD6,7 and associated with

434 C. Dickens et al.

multiple adverse health outcomes including poor physicalfunctioning77e80 worse heath related quality of life,81

increased use of urgent healthcare23 and mortality.82,83

Yet, it is striking how few of the interventions wereviewed included any form of psychosocial component,although current NICE guidelines recommend active treat-ment of depression in patients with long term conditions.84

In fact overall, there are very few studies of either anti-depressant medication or psychosocial interventions for thetreatment of either depression or anxiety in COPD, andmost are of poor quality.85,86

Although the evidence is relatively weak, there is somesupport for psychosocial interventions and their impactupon anxiety in COPD. A study comparing the use of aphysical activity facilitator for patients with depression inprimary care found that it did not improve mood or reduceantidepressant use when compared with usual care.87 Giventhat we know that depression predicts greater use of urgenthealthcare,23 it may be that if exercise in COPD improvesdepression it could also reduce urgent healthcare use.Therefore, offering possibilities of integrating interventionsaimed at psychological and physical health outcomes.

In light of the findings of this systematic review, thepotential impact of targeted psychological/pharmacolog-ical interventions to reduce unscheduled care in patientswith COPD and anxiety/depression requires evaluation.

Conflict of interest statement

Information on access to data

Professor Dickens is independent of any commercial funderor sponsor and has had full access to all the data in thestudy and takes responsibility for the integrity of the dataand the accuracy of the data analysis.

Disclosure of potential conflicts of interest,activities, relationships and affiliations

Katon has received honoraria for lectures for Lilly, Wyeth,Forest, Phizer and acts on advisory boards for Lilly, Wyeth

None of the other authors have conflicts of interests orfinancial interests to declare.

Sources of funding and support

This study was funded by the National Institute for HealthResearch (NIHR) under its Programme Grants for AppliedResearch Programme and by Peninsula CLAHRC (Dickens).

Role of the sponsor

The sponsor has provided funding for this project butplayed no part in the design and conduct of the study;collection, management, analysis, and interpretation ofthe data; and preparation, review, or approval of themanuscript.

Disclaimer

This article presents independent research commissionedby the NIHR under its Programme Grants for AppliedResearch Programme (RP-PG-0707-10162). The viewsexpressed in this publication are those of the authors andnot necessarily those of the NHS, the NIHR or the Depart-ment of Health.

Information on previous presentations

The findings contained in this document have not previouslybeen presented or published.

Listing of supplemental material

Additional material is included in the material for Onlineonly appendix, including full PICO criteria, details of thesearch strategies used and databases searched, tablescontaining details of the methodological characteristics ofincluded studies and study quality and a list of referencesfor papers included in this review.

Contributions of others compensated for theircontributions

No individual were compensated for their contribution tothis article.

Acknowledgements

Authors contributionsDickens: planned and designed this review and meta-

analysis, drew up the inclusion and exclusion criteria, su-pervised the development of search strategies, supervisedthe identification of eligible papers, supervised dataextraction and meta-analysis, interpreted findings, wrotethe first draft of the paper for submission, co-ordinatedwith co-authors to collate comments, wrote the finaldraft of the paper.

Katon: contributed to the planning and design of thereview and meta-analysis, assisted in the identification ofeligible papers, made important intellectual contributionto the interpretation of the findings, the drafting of thepaper and approved the final version of the paper.

Khara: screened titles and abstract to identify eligiblepapers, assisted in data extraction, the drafting of thepaper and approved the final version of the paper

Blakemore: screened titles and abstract to identifyeligible papers, assisted in data extraction, the drafting ofthe paper and approved the final version of the paper

Tomenson: assisted in the design of the study, con-ducted meta-analysis, assisted in the drafting of the paperand approved the final version of the paper

Guthrie: assisted in the planning and design of this re-view and meta-analysis, assisted in the development of in-clusion and exclusion criteria, assisted in the developmentof search strategies, assisted in the interpretation of find-ings, assisted in the drafting of the paper and approved thefinal version of the paper.

Interventions that reduce urgent healthcare in COPD 435

Other acknowledgementsRos McNally, Librarian advised on the electronic search

strategies.Linda McGowan, peer-reviewed electronic search

strategiesClaire Blakley, assisted with identifying eligible papers

from rerun of search in January 2013

Appendix A. Supplementary data

Supplementary data related to this article can be foundonline at http://dx.doi.org/10.1016/j.rmed.2013.05.011.

References

1. Mathers C, Loncar D. Projections of global mortality andburden of disease from 2002 to 2030. PLoS Medicine 2006;3(11):2011e30.

2. Fletcher M, Upton J, Taylor-Fishwick J, Buist S, Jenkins C,Hutton J, et al. COPD uncovered: an international survey onthe impact of chronic obstructive pulmonary disease (COPD)on a working age population. BMC Public Health 2011;11:612.

3. Jones P, Bruselle G, Dal Negro R, Ferrer M, Kardos P, Levy M,et al. Health-related quality of life in patients with COPDseverity within primary care in Europe. Respiratory Medicine2011;105:57e66.

4. Tsai C, Grsiwold S, Clark S, Carmargo C. Factors associatedwith frequency of emergency department visits for chronicobstructive pulmonary disease exacerbation. Journal ofGeneral Internal Medicine 2007;22:799e804.

5. Menzin J, Boulanger L, Marton J, Guadagno L, Dastani H,Dirani R, et al. The economic burden of chronic obstructivepulmonary disease (COPD) in a U.S. medicare population.Respiratory Medicine 2008;102(9):1248e56.

6. Zhang M, Ho R, Cheung M, Fu E, Mak A. Prevalence ofdepressive symptoms in patients with chronic obstructivepulmonary disease: a systematic review, meta-analysis andmeta-regression. General Hospital Psychiatry 2011;33:217e23.

7. Willgoss T, Yohannes A. Anxiety disorders in patients withchronic obstructive pulmonary disease: a systematic review.Respiratory Care 2012;58(5):858e66.

8. Moussavi S, Chatterji S, Verdes E, Tandon A, Patel V, Uston B.Depression, chronic diseases, and decrements in health: re-sults from the World Health Surveys. The Lancet 2007;370:851e8.

9. Stein M, Cox B, Afifi T, Belik S, Sareen J. Does co-morbiddepressive illness magnify the impact of chronic physicalillness? A population-based perspective. Psyhcological Medi-cine 2006;36:587e96.

10. Creed F, Morgan R, Fiddler M, Marsham S, Guthire E, House A.Depression and anxiety impair health-related quality of lifeand are associated with increased costs in general medicalinpatients. Psyhcosomatics 2002;43:302e9.

11. Gilmer T, O’Connor P, Rush W. Predictors of health care costsin adults with diabetes. Diabetes Care 2005;28:59e64.

12. Keene J, Rodriguez J. Are mental health problems associatedwith use of accident and emergency and health-related harm?European Journal of Public Health 2007;17(387):393.

13. Williams E, Guthrie E, Mackway-Jones K. Psychiatric status,somatisation and health care utilization of frequent attendersat the emergency department: a comparison with routine at-tenders. Journal of Psychosomatic Research2001;50(161):167.

14. Coventry P, Gemmell I, Todd C. Psychosocial risk factorsfor hospital readmission in COPD patients on early

discharge services: a cohort study. BMC Pulmonary Medi-cine 2011;11.

15. Fan V, Ramsey S, Giardino N, Make B, Emery C, Diaz P, et al.Sex, depression, and risk of hospitalization and mortality inchronic obstructive pulmonary disease. Archives of InternalMedicine 2007;167(21):2345e53.

16. Ng T, Nitti M, Tan W, Cao Z, Ong K. Depressive symptoms andchronic obstructive pulmonary disease: effect on mortality,hospital readmission, symptom burden, functional status andquality of life. Archives of Internal Medicine 2007;167:60e7.

17. Almagro P, Barierio B, Ochoa de Echaguen A, Quintana S,Rodriguez Carballeira M, Heredia J, et al. Risk factors forhospital readmission in patients with chronic obstructivepulmonary disease. Respiration 2006;73:311e7.

18. Farkas J, Kosnik M, Flezar M, Suskovic S, Lainscak M. Self-rated health predicts acute exacerbations in patients withCOPD. Chest 2010;138(2):323e30.

19. Ghanei M, Aslani J, AzizAbadi-Farahani M, Assari S, Saadat S.Logistic regression model to predict chronic obstructive pul-monary disease exacerbation. Archives of Medical Science2007;3(4):360e6.

20. Carneiro R, Sousa C, Pinto A, Amedia F, Olivera J, Rocha N.Risk factors for readmission after hospital discharge in chronicobstructive pulmonary disease. Revista Portuguesa de Pneu-mologia 2010;16:759e77.

21. Xu W, Collet J, Shapiro S, Lin Y, Yang T, Platt R, et al. Inde-pendant effect of depression and anxiety on chronicobstructive pulmonary disease exacerbations and hospital-isations. American Journal of Respiratory and Critical CareMedicine 2008;178:913e20.

22. Gudmundsson G, Gislason T, Lindberg E, Hallin R, Ulrik C,Brondum E, et al. Mortality in COPD patients discharged fromhospital: the role of treatment and co-morbidity. RespiratoryResearch 2006;7:109.

23. Dickens C, Katon W, Blakemore A, Khara A, McGowan L,Tomenson B, et al. Does depression predict use of unscheduledcare by people with long term conditions: a systematic reviewwith meta-analysis. Psychosomatic Research 2012;73:334e42.

24. NHS Employers. Quality and outcomes framework guidancefor GMS contract 2011/2012. London: NHS Confederation(Employers) Company; 2011.

25. NHS Institute for Innovation and Improvement. Directory ofambulatory emergency care for adults. Coventry: NHS Insti-tute for Innovation and Improvement; 2010.

26. HarknessE,MacDonalW,ValderasJ,CoventryP,GaskL,BowerP.Identifying psychosocial interventions that improve both phys-ical and mental health in patients with diabetes: a systematicreview and meta-analysis. Diabetes Care 2010;33:926e30.

27. Higgins J, Green S. Cochrane handbook for systematic reviewsof interventions version 5.1.0 [updated March 2011]. TheCochrane Collaboration; 2011.

28. Hewitt C, Hahn S, Torgerson D, Watson J, Bland J. Adequacyand reporting of allocation concealment: review of recenttrials published in four general medical journals. BritishMedical Journal 2005;330:557e60.

29. Lipsey M, Wilson D. Practical meta-analysis. London: SagePublications; 2001.

30. DerSimonian R, Laird N. Meta-analysis in clinical trials.Controlled Clinical Trials 1986;7:177e88.

31. Higgins J, Thompson S, Deeks J, Altman D. Measuring incon-sistency in meta-analyses. British Medical Journal 2003;327:557e60.

32. Egger M, Davey-Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. British MedicalJournal 1997;315:629e34.

33. Egger M, Davey-Smith G, Altman D. Systematic reviews inhealth care: meta-analysis in context. London: BMJ Publish-ing; 2010.

436 C. Dickens et al.

34. Thompson S, Higgins J. How should meta-regression analysesbe undertaken and interpreted? Statistics in Medicine 2002;21:1559e73.

35. Berkey C, Hoaglin D, Mosteller F, Colditz G. A random effectsregression model for meta-analysis. Statistics in Medicine1995;14(4):395e411.

36. Bourbeau J, Julien M, Maltais F, Rouleau M, Beaupre A,Begin R, et al. Reduction of hospital utilization in patientswith chronic obstructive pulmonary disease: a disease- spe-cific self-management intervention. Archives of InternalMedicine 2003;163:585e91.

37. Bergner M, Hudson L, Conrad D, Patmont C, McDonald G,Perrin E, et al. The cost and efficacy of home care for pa-tients with chronic lung disease. Medical Care 1988;26(6):566e79.

38. Boxall A, Barclay L, Sayers A, Caplan G. Managing chronicobstructive pulmonary disease in the community: a random-ized controlled trial of home-based pulmonary rehabilitationfor elderly housebound patients. Journal of CardiopulmonaryRehabilitation 2005;25:378e85.

39. Casas A, Troosters T, Garcia-Aymerich J, Roca J, Hernandez C,Alonso A, et al. Integrated care prevents hospitalisations forexacerbations in COPD patients. European Respiratory Jour-nal 2006;28:123e30.

40. Eaton T, Young P, Fergusson W, Moodie L, Zeng I, O’Kane F,et al. Does early pulmonary rehabilitation reduce acutehealth-care utilization in COPD patients admitted with anexacerbation? A randomized controlled study. Respirology2009;14:230e8.

41. Farrero E, Escarrabil J, Prats E, Maderal M, Manresa F. Impactof hospital-based home-care program on the management ofCOPD patients living on long term oxygen therapy. Chest2001;119:364e9.

42. Gadoury M. A disease specific self-management interventionreduces hospital utilization in patients with COPD: the effectremains at 2 years. 2004.

43. Gallefoss F, Bakke P. Impact of patient education and self-management on morbidity in asthmatics and patients withchronic obstructive pulmonary disease. Respiratory Medicine2000;94:279e87.

44. Guell R, Casan P, Belda J, Sangenis M, Morante F, Guyatt G,et al. Long-term effects of outpatient rehabilitation of COPD.Chest 2000;117:976e83.

45. Hermiz O, Comino E, Marks G, Daffurn K, Wilson S, Harris M.Randomised controlled trial of home based care of patientswith chronic obstructive pulmonary disease. British MedicalJournal 2002:928e32.

46. Hernandez C, Casas A, Escarrabil J, Alonso J, Puig-Junoy J,Farrero E, et al. Home hospitalisation of exacerbated chronicobstructive pulmonary disease patients. European Respira-tory Journal 2003;21:58e67.

47. Jarab A, AlQudah S, Khdour M, Shamssaim M, Mukattash T.Impact of pharmaceutical care on health outcomes in patientswith COPD. International Journal of Clinical Pharmacy 2012;34:53e62.

48. Khdour M, Kidney J, Smyth B, McElnay J. Clinical pharmacy-led disease and medicine management programme for pa-tients with COPD. British Journal of Clinical Pharmacology2009;68(4):588e98.

49. Ko F, Dai D, Ngali J, Tung A, Ng S, Lai K, et al. Effect of earlypulmonary rehabilitation on health care utilization and healthstatus in patients hospitalised with acute exacerbations ofCOPD. Respirology 2011;16:617e24.

50. Koff P, Jones R, Cashman J, Voelkel N, Vandivier R. Proactiveintegrated care improves quality of life in patients with COPD.European Respiratory Journal 2009;33:1031e6.

51. Lee D, Lee I, Mackenzie A, Ho R. Effects of a care protocol oncare outcomes in older nursing home patients with chronic

obstructive pulmonary disease. American Geriatrcis and So-cial Care 2002;50:870e6.

52. Lobato S, Lorenzo F, Meddieta M, Arechabala I, Fernandez-Montes V. Evaluation of a home hospitalization program inpatients with exacerbations of chronic obstructive pulmonarydisease. Archivos de Bronconeumologia 2005;41(1):5e10.

53. Man W, Polkey M, Donaldson N, Gray B, Moxham J. Communitypulmonary rehabilitation after hospitalisation for acute ex-acerbations of chronic obstructive pulmonary disease: rand-omised controlled study. British Medical Journal 2004;329:1209.

54. Martin I, McNamara D, Suthreland F, Tilyard M, Taylor D. Careplans for acutely deteriorating COPD: a randomizedcontrolled trial. Chronic Respiratory Disease 2004;1:191e5.

55. McGeoch G, Willsman K, Dowson C, Town G, Frampton C,McCartin F, et al. Self-management plans in the primary careof patients with chronic obstructive pulmonary disease.Respirology 2006;11(611):618.

56. ReaH,McAuleyS, StewartA, LamontC,RosemanP,DidsburyP.Achronic disease management programme can reduce days inhospital for patients with chronic obstructive pulmonary dis-ease. Internal Medicine Journal 2004;34(608):614.

57. Rice K, Dean N, Bllomfield H, Grill J, Schult T, Nelson D, et al.Disease management program for chronic obstructive pul-monary disease. American Journal of Respiratory and CriticalCare Medicine 2010;182(7):890e6.

58. Ries A, Kaplan R, Myers R, Prewitt L. Maintenance after pul-monary rehabilitation in chronic obstructive pulmonary dis-ease. American Journal of Respiratory and Critical CareMedicine 2003;167:880e8.

59. Seymour J, Moore L, Jolley C, Ward K, Creasey J, Steier J,et al. Outpatient pulmonary rehabilitation following acuteexacerbations of COPD. Thorax 2010;65:423e8.

60. Smith B, Adams R, Appleton S, Trott C, Bennet P, Allan D,et al. The effect of a respiratory home nurse intervention inpatients with chronic obstructive pulmonary disease (COPD).Australia and New Zealand Medical Journal 1999;29:718e25.

61. Soler J, Martinez-Gracia M, Roman P, Orero R, Terrazas S,Martinez-Pechuan A. Effectiveness of a specific program forpatients with chronic obstructive pulmonary disease andfrequent exacerbations. Archivos de Bronconeumologia 2006;42(10):501e8.

62. Sridhar M, Taylor R, Dowson S, Roberts N, Partridge M. A nurseled intermediate care package in patients who have beenhospitalised with an acute exacerbation of chronic obstruc-tive pulmonary disease. Thorax 2007;63:194e200.

63. Tougaard L, Krone T, Sorknaes A, Ellegaard H. Economicbenefits of teaching patients with chronic obstructive pul-monary disease about their illness. The Lancet 1992;339:1517e20.

64. Trappenburg J, Monninkhof E, Bourneau J, Trrosters T,Schrijvers A, Verheij T, et al. Effect of an action plan withongoing support by a case manager on exacerbation-relatedoutcome in patients with COPD: a multicentre randomisedcontrolled trial. Thorax 2011;66:977e84.

65. Wakabayashi R, Motegi T, Yamada K, Ishii T, Jones R,Hyland M, et al. Efficient integrated education for older pa-tients with chronic obstructive pulmonary disease using theLung Information Needs Questionnaire. Geriatrics andGerontology International 2011;11:422e30.

66. Wong K, Wong F, Chan M. Effects of nurse-inititated telephonefollow-up on self-efficacy among patients with chronicobstructive pulmonary disease. Journal of Advanced Nursing2005;49(2):210e22.

67. Wood-Baker R, McGlone S, Venn A, Walters E. Written actionplans in chronic obstructive pulmonary disease increaseappropriate treatment for acute exacerbations. Respirology2006;11:619e26.

Interventions that reduce urgent healthcare in COPD 437

68. Katon W, Robinson P, Von Korff M, Lin E, Bush T, Ludman E,et al. A multifaceted intervention to improve treatment ofdepression in primary care. Archives of General Psychiatry1996;53(10):924e32.

69. Katon W, Lin E, Von Korff M, Ciechanowski P, Ludman E,Young B, et al. Collaborative care for patients with depressionand chronic illness. The New England Journal of Medicine2010;363(27):2611e20.

70. Huffman J, Mastromauro C, Sowden G, Wittman C, Rodman R,Januzzi J. A collaborative care depression management pro-gram for cardiac inpatients: depression characteristics and in-hospital outcomes. Psychosomatics 2011;52:26e33.

71. Harpole L, Williams J, Olsen M, Stechuchak K, Oddone E,Callahan C, et al. Improving depression outcomes in olderadults with comorbid medical illness. General Hospital Psy-chiatry 2005;27:4e12.

72. Katon W, Von Korff M, Lin E, Simon G, Ludman E, Russo J,et al. The Pathways Study: a randomized trial of collaborativecare in patients with diabetes and depression. Archives ofGeneral Psychiatry 2004;61:1042e9.

73. Rollman B, Belnap B, LeMenager M, Houck P, Counihan P,Kapoor W, et al. Telephone-delivered collaborative care fortreating post-CABG depression. The Journal of the AmericanMedical Association 2009;302(19):2095e103.

74. Baumeister H, Hutter N. Collaborative care for depression inmedically ill patients. Current Opinion in Psychiatry 2012;25(5):405e14.

75. van Steenbergen-Weijenburg K, van der Feltz-Cornelis C,Horn E, van Marwijk H, Beekman A, Rutten F, et al. Cost-effectiveness of collaborative care for the treatment of majordepressive disorder in primary care. A systematic review. BMCHealth Services Research 2010;10(19).

76. Unutzer J, KatonW, FanM, SchoenbaumM,LinE, PennaR. Long-term cost effects of collaborative care for late-life depression.American Journal of Managed Care 2008;14:95e100.

77. Ambrosino N, Simmonds A. The clinical management of COPD.Respiratory Medicine 2007;101(8):1613e24.

78. Cully J, Graham D, Stanley M, Ferguson C, Sharafkhaneh A,Souchek J, et al. Quality of life in patients with chronic

obstructive pulmonary disease and co morbid anxiety anddepression. Psychosomatics 2006;47:312e9.

79. Felker B, Katon W, Hedrick S, Rasmussen J, McKnight K,McDonnell M, et al. The association between depressivesymptoms and health status in patients with chronic pul-monary disease. General Hospital Psychiatry 2001;23(2):56e61.

80. Kim H, Kunik M, Molinari V, Hillman S, Lalani S, Orengo C,et al. Functional impairment in COPD patients: the impact ofanxiety and depression. Psychosomatics 2000;41:465e71.

81. Tsiligianni I, Kocks J, Tzanakis N, Siafakas N, van der Molan T.Factors that influence disease-specific quality of life or healthstatus in patients with COPD: a review and meta-analysisor Pearson correlations. Primary Care Respiratory Journal2011;20.

82. Papaioannou A, Bartziokas K, Tsikirka S, Karakontaki F,Kastanakis E, Banya W, et al. The impact of depressive symp-toms on recovery and outcome of hospitalized COPD exacer-bations. European Respiratory Journal 2013;41(4):815e23.

83. Abrams T, Vaughan-Sarrazin M, Vander Weg M. Acute exac-erbations of chronic obstructive pulmonary disease and effectof existing psychiatric comorbidity on subsequent mortality.Psyhcosomatics 2011;52:441e9.

84. National Institute for Health and Clinical Excellence.Depression with a chronic physical health problem [CG91].London: National Institute for Health and Clinical Excellence;2009.

85. Rayner L, Price A, Evans A, Valsraj K, Higginson I, Hotopf M.Antidepressants for depression in physically ill people.Cochrane Database of Systematic Reviews 2010;(3). Art NoCD007053.

86. Baraniak A, Sheffield D. The efficacy of psychologically basedinterventions to improve anxiety, depression and quality oflife in COPD: a systematic review and meta-analysis. PatientEducation and Counselling 2011;83:29e69.

87. Chalder M, Wiles N, Campbell J, Hollinghurst S, Hasse A,Taylor A, et al. Facilitated physical activity as a treatment fordepressed adults: randomised controlled trial. British MedicalJournal 2012;344:e2758.