The effectiveness of interventions to achieve coordinated multidisciplinary
care and reduce hospital use for people with chronic diseases
Progress to date and methodological challenges
Outline • Background to the ‘overview study’ • Aim and objectives • Selection of outcome measures • Methods and data analysis • Preliminary findings: hospitalisation • Methodological challenges
– Double counting of primary studies – Excluding studies on the basis of quality score – Grouping studies with similar interventions
Background • People with chronic diseases often require care crossing
multiple health and social care settings • These patients typically experience fragmented and poorly
coordinated management of their conditions • There will be a significant increase in the number of patients
with chronic diseases in the coming years, and associated increase in costs of management
• Government policy increasingly emphasises integrated care and developing new models of delivering healthcare services e.g. Better Care Fund
• Yet there is a paucity of evidence about what works, for which patients, in which settings, and for which outcomes
Aim and objectives Aim: To assess the effectiveness of interventions and integrated management models for chronic disease in terms of service provider and patient outcomes Objectives: 1. Describe and review the conceptual frameworks being used
to define integrated models of care for chronic disease 2. Describe and evaluate trends in service provision 3. Assess which models or individual elements are particularly
effective in delivering improved provider and patient outcomes
4. Identify gaps in the evidence base where further targeted systematic reviews of primary research may be useful
Selection of outcome measures Service providers: • Stakeholder meeting with representatives from each of
the Theme 4 case study sites, CCGs and others • Discussion to establish which questions and outcomes
are the most important to providers Patients: • Literature review to inform a short survey • List of commonly reported patient outcomes derived and
given to patients via CLAHRCWM PPI database • Input from Theme 4 PPI representative
Outcome measures chosen: • Healthcare utilisation: Including (re)admission rates,
length of hospital stay, A&E visits, clinic and GP resource use, institutionalisation rates
• Costs: Healthcare costs or savings from interventions
• Quality of life: Generic or disease specific measures
• Patient satisfaction: Impact of interventions or service models on patient satisfaction with care
• Care coordination: Interventions or service models which assess the effectiveness of providing care coordination and ‘patient centered’ care
Search strategy Medline Cochrane Library and DARE
Medline in Process Health Economics Evaluations Database
Applied Social Sciences Index and Abstracts (ASSIA)
TRIP database (systematic reviews and evidence synthesis)
PsycINFO PROSPERO
Health Management Information Consortium Web based searches
CINAHL EPPI-Centre Library
Embase Reference lists of included papers
Inclusion and exclusion criteria INCLUSION CRITERIA EXCLUSION CRITERIA
Published during or since 2000 Reviews with unsystematic methods
Published in English language Studies focusing on caregivers
Systematic reviews, meta-analysis, narrative reviews
Editorials, opinion pieces, commentaries, letters, conference proceedings
Comparison group includes usual care, no intervention or other interventions
Studies assessing integration in less economically developed countries
Adult patients with one or more chronic diseases Palliative or end of life care
Assess care models or interventions crossing the boundary between two or more healthcare settings
Interventions solely psychosocial or related to spirituality, mindfulness, health literacy or the use of CAM
Include data related to one or both provider outcomes and/or one or both patient outcomes
Interventions relating solely to lifestyle change or treatment/medication adherence
Study selection • Removal of duplicate records • Two reviewers screen titles and abstracts for relevance • Relevant reviews proceed to full text evaluation • Two reviewers screen full texts for outcomes of interest • Data extraction and quality assessment according to
pre-defined forms • For grey literature, relaxation of strict criteria on
methodological robustness and systematic methods • All disagreements resolved by discussion or arbitration
by a third reviewer if necessary
Data extraction Reference ID Research questions Who delivers the intervention?
Citation Study designs included Who co-ordinates the intervention?
Country of publication Number of studies included Source of intervention
Databases and search years Type of review Length of follow-up
Geographical scope Definition of intervention Intervention context
Language restriction Study population Primary and secondary outcomes
Healthcare setting(s) Number of participants Data on our outcomes of interest
Chronic disease(s) General description of intervention
Review summary and conclusions
Overall aim of review Specific features of intervention Any other information
Quality assessment • Centre for Evidence Based Medicine checklist: 5
questions, total score from 0-5 (5 = high quality) 1. Is the research question clearly stated? 2. Is it unlikely that important, relevant studies were
missed? 3. Were the inclusion criteria appropriate? 4. Were included studies sufficiently valid for the type of
question asked? 5. Were results similar from study to study?
Records from databases n=10,682
Records from other sources
n=11
After duplicate removal n=7,747
Titles and abstracts screened
n=7,747
Full text articles assessed
n=353
Included reviews n=81
Removal of duplicates n=2,946
Excluded after title and abstract screen
n=7,394
Full text articles
excluded n=272
Review characteristics • 81 reviews eligible for data extraction
• All published between 2004 to 2014
• 45 covered 3+ settings; 2 primary and secondary care, 8 primary and community/social care/home, 24 secondary care and community/social care/home
• 22 general chronic disease; 59 covered individual or multiple specific conditions
• Number of primary studies ranged from 4 to 560
• 45 narrative reviews, 32 systematic reviews with meta-analysis, 4 reviews of reviews
• Interventions highly heterogeneous in design, intensity, duration and outcomes reported
Methodological challenges
•Double counting of primary studies •Excluding papers on the basis of quality score •Grouping of interventions
Double counting • 81 reviews include a total of 1293 primary studies
• The same primary studies frequently occur in more than one included review
• Potentially problematic when synthesising results from multiple reviews on the same intervention
• Artificial inflation of effect size and over-estimation of effectiveness of intervention or service model
586
103
41 22 13 3 2 5 2 8 5 0
100
200
300
400
500
600
700
1 2 3 4 5 6 7 8 9 10 11
Number of times a single reference is cited
0
2
4
6
8
10
12
14
16
0 to 9 10 to 19 20 to 29 30 to 39 40 to 49 50 to 59 60 to 69 70 to 79 80 to 89 90 to 100
Num
ber o
f rev
iew
s
Percentage of references duplicated
Proportion of references duplicated, by review
Potential solutions • In a given intervention category, select only the most
recent high quality review and exclude previous reviews (after checking conclusion agreement)
• Only include reviews that had RCTs as their primary study design
• Include only studies with meta-analyses • Subdivide reviews on a specific intervention by time-
band (i.e. 2000-2005, 2005-2010, 2010-present) and analyse change in outcome over time for ‘drift’ in results
Excluding on QA score
QA SCORE NUMBER OF REVIEWS
1 3
1.5 7
2 6
2.5 5
3 14
3.5 8
4 20
4.5 5
5 13
• 21 of 81 reviews have QA score less than 3/5
• Real difference in quality at this cut-off point
• Very little, if any outcomes data can be extracted from the low scoring reviews
• Restrict inclusion to reviews with higher quality evidence?
Grouping interventions • Interventions could be categorised in an almost infinite
number of ways and most interventions overlap Professional
interventions
Financial or regulatory
interventions Organisational interventions
Provider orientated
Professional role change; MD teams;
service integration; continuity of
care
Patient orientated
Consumer participation in care governance; self-management
e.g. professional education
e.g. provider incentives or
penalties
Structural Changes to setting/site of service delivery; changes in medical record systems
0 studies 0 studies Nearly all included studies have elements of each
Top down… Category Case
manage-ment
Discharge planning
Shared care Care planning
Disease manage-ment
MD care
Integrated care
Revision of professional roles
✓ ✓ ✓ ✓ ✓ ✓
MDTs ✓ ✓ ✓ ✓ ✓ ✓ ✓
Formal service integration ✓ ✓ ✓ ✓ ✓ ✓ ✓
Continuity of care ✓ ✓ ✓ ✓ ✓ ✓
Self-management ✓ ✓ ✓ ✓ ✓
Changes to setting of service delivery
✓ ✓ ✓ ✓ ✓
Bottom up…
Recipient Facilitator(s) Key strategies
Method of contact Intensity
• Patient • Caregiver • Physician
• Nurse • MDT • OT/PT • Dietician • Social
worker • GP • Specialist
• Education • Advice • Goal
setting • Managing
risk factors
• Care plans
• In person home visit
• Phone • In person
clinic visit • Remotely
• Frequency • Duration • Complexity
(Re)admissions summary findings POSITIVE MIXED +ve / -ve NO DIFFERENCE
Prof. roles • Nurse led care (2)
MDTs • Pharmacist-led (1) • MDT (1)
Continuity of care
• Shared care (1) • Discharge man. (3)
• Transitional care hospital to home (4)
• Case man. (4) • Shared care (1) • Discharge man. (2)
Self-management
• Self-management (2) • Self-management (1)
Change to setting of service delivery
• Hospital at home (1) • Specialist HF clinics (1)
• Early supported discharge (1)
• Primary care follow up (1)
• E-health (1) • Early supported
discharge (1) • Primary c. f-up (1)
Formal service integration
• Comm. Based care (1) • Disease management
(5) • Integrated care (1)
• Disease management (6) • Integrated care (2)