Health State Utility Value in Chronic Obstructive Pulmonary Disease (COPD); The Challenge of Heterogeneity: A Systematic Review and Meta-Analysis

ABSTRACT

Chronic obstructive pulmonary disease (COPD) has a considerable impact on quality of life and well-being of patients. Health state utility value (HSUV) is a recognized measure for health economic appraisals and is extensively used as an indicator for decision-making studies. This study is a systematic review of literature aimed to estimate mean utility value in COPD using meta-analysis and explore degree of heterogeneity in the utility values across a variety of clinical and study characteristic. The literature review covers studies that used EQ-5D to estimate utility value for patient level research in COPD. Studies that reported utility values elicited by EQ-5D in COPD patients were selected for random-effect meta-analysis addressing inter-study heterogeneity and subgroup analyses. Thirty-two studies were included in the general utility meta-analysis. The estimated general utility value was 0.673 (95% CI 0.653 to 0.693). Meta-analyses of COPD stages utility values showed influence of airway obstruction on utility value. The utility values ranged from 0.820 (95% CI 0.767 to 0.872) for stage I to 0.624 (95% CI 0.571 to 0.677) for stage IV. There was substantial heterogeneity in utility values: I² = 97.7%. A more accurate measurement of utility values in COPD is needed to refine valid and generalizable scores of HSUV. Given the limited success of the factors studied to reduce heterogeneity, an approach needs to be developed how best to use mean utility values for COPD in health economic evaluation.

KEYWORDS:

• chronic disease
• Chronic Obstructive Pulmonary Disease (COPD)
• EQ-5D
• heterogeneity
• Quality of Life (QoL)
• utility

Introduction

Quality of life can be defined as an individual's perception of their position in life or life satisfaction. It is a complex entity incorporating physical health, psychological condition, independent living, social relationships and personal judgement (1). Health status, functional status, well-being, quality of life (QoL), health related quality of life (HR-QoL) and health state utility value (HSUV) are used interchangeably, but despite some differences in meaning, all these concepts are classified as patient-reported outcomes (PROs) (2). In clinical practice, HSUV instruments are used to design clinical management guidelines, prioritizing patient complaints, screening possible problems and making decisions about treatment modalities.

Presently, Quality Adjusted Life Years (QALYs) are commonly applied as a measure of health in economic appraisals and are extensively used as outcomes for resource allocation decisions. Cost effectiveness of medical intervention in Chronic Obstructive Pulmonary Disease (COPD) utilizes generic (such as EQ-5D, SF-36) (3, 4) or diseases-specific measures of QoL [such as St. George Respiratory Questionnaire (SGRQ) and Clinical COPD Questionnaire (CCQ)] (5, 6).

Generic instruments such as EQ-5D have the advantage of having value-sets which facilitate the quantification of patient rated health status into measures of utility. These health-state utility reflects not only the presence, frequency or intensity of symptoms, abilities, or feeling as measured by psychometric instruments (7) but also represents a social or individual's preferred value or judgment for specific health states relative to full health (8, 9). The EQ-5D is the most widely used generic measure across all diseases. To convert patient responses to the health descriptors used in the scale to a single index of HSUV, a preference-based set of weights is applied. These descriptors comprise five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression). In EQ-5D-5L (version 2005), each dimension has five levels: no problems, slight problems, moderate problems, severe problems, and extreme problems. In addition to the descriptive system, the EQ-5D contains a 25-cm vertical visual analogue scale (EQ VAS) that records the respondent's self-rated health, and can be used as a quantitative measure of health outcome. Based on societal preferences for health states, country-specific algorithms or tariffs have been generated (10, 11). The minimally important clinical difference for the EQ-5D Index has been estimated to be: ±0.074 (12).

Overviews and meta-analyses of utility-based quality of life have been undertaken in a variety of diseases including diabetes (13), various types of cancer (7, 14, 15), HIV/AIDS (16), chronic kidney disease (17), neuropathic pain (18) and orthopaedic diseases (19). The main purposes of these reviews were to examine the applicability of these utility measures in patients with the diseases and to attempt to summarize mean utility scores according to the disease states.

Utility-based health-related quality of life in patients with COPD (necessarily together with their common co-morbidities) has been measured using surveys of COPD patients, but values differ significantly across studies. For instance, the reported average utility values for stage II COPD range from 0.579 (20) to 0.929 (21). Different methods of utility elicitation measures explain part of this variability. A recent study (9) examining the role for meta-analysis for utility values has noted that combining reported utilities can be problematic, due to for example valuation methods and have recommended only combining studies reporting utility values that are derived in a similar fashion (e.g., using the same generic quality of life instrument). For this reason we confine our review to studies that employ the EQ-5D to measure utility values for COPD patients. While this may reduce some variation, the diversity in COPD patient population characteristics may also have other imposed effects on the value of utility measured in different studies.^{Figure 1}

The first aim of this study is to conduct a meta-analysis using EQ-5D, the most widely used instrument to determine mean utility scores for COPD. The second aim of this study is to explore degree of heterogeneity in the mean utility values across a variety of clinical and study characteristics.

Methods

Study selection

The literature review of HSUV studies in COPD comprises studies that use EQ-5D to estimate utility values for patient level research in COPD; simulation-based studies were not included.

Studies with the following criteria were included:

• studies on health utility that were published prior to July 2015;

• studies in which their sample population was specifically categorized as COPD as defined by standard criteria for COPD diagnosis and spirometric confirmation (should clearly be addressed in methodology of included studies);

• English language studies and non-English language studies with English abstracts;

• abstracts (e.g., seminar abstracts) and reports if adequate data for analysis were provided.

• studies with more than 10 participants

Exclusion was applied for the following criteria:

• editorials /opinion pieces, letters, systematic reviews and meta-analyses;

• studies that reported utilities from proxies, not individual participant data (e.g., reported by family member or a health professional);

• studies that obtained utility estimates from the literature, if there was not enough information on the derivation of utility;

• studies that did not distinguish COPD from other types of obstructive pulmonary disease such as asthma or cystic fibrosis;

• papers using utility values mapped from other reported Quality of Life studies;

• Studies that reported utility values from non-stable and exacerbation state COPD patients.

Studies with different epidemiological designs (i.e., case control, randomized control trial (RCT), cohort, etc.) were included. It is not always feasible to conduct utility data collection within a clinical trial, so utility data from non-clinical trial studies was also included. In order to eliminate additive effect of studies using same data source, special effort made to only include the study with the largest sample size.

This systematic review followed MOOSE guideline for observational studies (22). A search strategy was employed for MEDLINE database (Appendix 1) and was adopted for other databases. A hand search and citation-tracking were also conducted.

To ensure consistency in literature review of utility elicitation methodology, general recommendations of the Peasgood and Brazier (9) were followed. EndNote X7.3.1 was used to download citation, and to identify and extract duplicate studies.

Search methods

The systematic review of the literature on utility values for COPD was part of a wider systematic review of economic evidence on COPD, related pharmacological and psychological interventions and progression modelling for patients with COPD. The following electronic databases were searched for relevant articles: MEDLINE, EMBASE (for the period of 1898–2015), Web of Science, CINAHL, ProQuest (which includes PsycINFO and other 61 databases), the Cochrane Library Database (which includes NHS Economic Evaluation Database, Health Technology Assessment Database, Cochrane Database of Systematic Reviews and other three databases), International Society for Pharmacoeconomics and Outcomes Research (ISPOR) and Google Scholar. An attempt was made to decrease the likelihood of publication bias (23) by using dissertation and web sites of key academic institutions such as NICE (National Institute of Clinical Excellence), CCOHTA (Canadian Cooperating Office for Health Technology Assessment), SBU (The Swedish Council on Technology Assessment in Health Care), Health Economic Evaluations Database (HEED, ceased publishing in 2014) and the Cost Effectiveness Analysis Registry at Tufts-New England Medical Centre.

Data extraction and management

Data from included articles were extracted into Excel and Stata spread sheets. The following variables were obtained from each citation: principal author, year of publication, clinical characteristics and demographic of patients, number of patients, country of origin, study design, data collection method, health state utility value measure and utility estimate (mean and standard deviation). In intervention studies, such as randomized control trials, baseline characteristics were used to avoid the potential effect of the intervention on the quality of life estimates. When a demographic or clinical factor splits intervention groups, the entire number of the whole was used where possible. Assessment of study eligibility and extraction of information from each study were carried out by two independent reviewers.

Data analysis

In order to estimate a single mean utility score value for COPD, a meta-analysis was conducted. This was done for COPD as a general condition and for the stages of the disease separately. Point estimates and 95% Confidence Intervals (CI) for utility scores were calculated and displayed in forest plots. Possible publication biases were investigated using funnel plots. Meta-analysis was restricted to EQ-5D Index-elicited utility values, as this was the only utility measure that existed in sufficient numbers for it to be feasible to undertake a meta-analysis. This restriction avoided heterogeneity imposed by elicitation methodology diversity (9).^{Table 1Table 2Table 3}

Meta-analysis was conducted with the command metan (24), using Stata version 13.1. The within-study variability was considered through incorporating random effects model and a mean of a distribution of true effects was estimated. Heterogeneity among the studies was measured using I² statistic = 100% × (Q - df) ⁄Q and 95% CI, indicating the proportion of observed variance due to real differences in utility scores rather than sampling error. Values of 30%–60%, 50%–90% and 75%–100% were considered as moderate, substantial and considerable heterogeneity. If standard errors of utility values were not reported, they were calculated from 95% confidence intervals or standard deviations. If any study did not present enough data for measuring standard error, it was excluded. metabias and metafunnel commands were used to perform the Egger regression asymmetry test for publication bias and draw the funnel plot (25, 26). To demonstrate influence of outlier studies on the overall meta-analysis metaninf command was used.

To conduct pre-specified subgroup analyses, study variables including clinical/participant and conduct of study factors were selected to define subgroups as follows: age, gender, FEV1% predicted, pack-years number of cigarette smoking, number of patients per study, Hospital Anxiety and Depression Scale (HADS) depression index, Borg dysphonia index, Charlson co-morbidity index, level of literacy, length of COPD and Body-mass index, airflow Obstruction, Dyspnoea, and Exercise capacity (BODE) index scores. Interaction tests were conducted only if there were at least two studies in each of the subgroups. Meta-regression was abandoned because of insufficient number of studies in some subgroups. Interaction models to some subgroups of interest were applied and changes in magnitude or direction of the utility values and heterogeneity were reported. The t-test and analysis of variance (ANOVA) were applied for comparing estimated utility means between subgroups.^{Table 2}

Results

Study characteristics

The flow diagram (Figure 1) summarises the selection process of articles to be included. The initial pool of studies comprised 17,565 entries, including three citations captured through hand search (27–29). Of these, 17,570 were excluded after scanning of abstracts. Full text examination of 404 studies was conducted and, after incorporating inclusion and exclusion criteria, 78 studies were selected for review. Thirty-two studies with 49 observations gave estimates of general utility values for COPD population as a whole. Included articles in meta-analysis are tabulated in Tables 1 and 2^{Figure 2}. To adhere to Cochrane handbook recommendation on including studies with multiple intervention groups (multiple observations) in a particular meta-analysis, observations of a single study were combined to create a single value.

Table 1. Characteristics of studies included in meta-analysis.

						Age			FEV1/FVC		FEV1 % pred
First author/		Number of		Co-			SD or	Male		SD or		SD or
year of publication	Country	patients	Population	morbidities	Study design	Mean	range	(%)	Mean	Range	Mean	range	Intervention
Wu et al. 2015 (30)	China	678	COPD in community health centre, after spirometry test	—	Cross-sectional	70.4	10.1	72.9	—	—	54.5	23.0	—
Wilson et al. 2015 (31)	UK	In 73	COPD registries, after spirometry tests	—	RCT	67.3	15.1	41.0	—	—	—	—	Pulmonary rehabilitation
		Co 75				69.3	8.9	50.0
Sundh et al. 2015 (32)	Sweden	373	COPD registry in hospitals	Cardiovascular disease, Chronic bronchitis, Diabetes, Renal failure, depression	Cross-sectional	71.25	—	44.24	—	—	34.72.	—	—
Stoddart et al. 2015 (33)	UK	In 128	COPD registry in hospitals	—	RCT	69.4	8.8	25	—	—	44.0	18.8	Telemonitoring
		Co 128				68.4	8.4	35			40.0	17.0
McDowell et al. 2015 (34)	Ireland	In 55	COPD registry in hospitals	—	RCT	69.8	7.1	41.8	—	—	45.5	13.7	Telemonitoring
		Co 55				70.2	7.4	45.5			43.4	11.3
Donohue et al. 2014 (35)	Multi, 11 countries	Trial 1 353	COPD confirmed by spirometry	—	RCT	62.5	9.05	72	47.5	10.61	49.2	10.82	Pharmaceutical
		353				63.0	8.91	69	46.8	10.78	49.6	10.88
		Trial 2 349				63.2	8.57	76	47.3	10.73	49.4
		348				64.0	8.53	76	47.0	10.72	49.5	10.81
												10.87
Lin et al. 2014 (36)	USA	670	COPD patients from a multicentre study, after spirometry test	Hypertension, Diabetes, Cancer, Dementia, Depression	Cross-sectional	68.5	10.4	57.8	—	—	—	—	—
Ferreira et al. 2014 (37)	Portugal	72	COPD registry in hospital recruited by pneumologists	—	Cross-sectional	68.6	9.5	97.2	—	—	—	—	—
Chen et al. 2014 (38)	Hong Kong	154	COPD outpatient respiratory specialist clinic	Hypertension, Heart disease, diabetes, cancer liver disease …	Cross-sectional	42.9	8.1	98.7	—	—	32.7	9.2	—
Gillespie et al. 2013 (39)	Ireland	In 178	COPD general practice after spirometry test	—	RCT	68.4	10.3	61.6	55.4	11.9	59.7	13.8	Structured
		Co 172				68.8	10.2	65.7	52.9	11.5	57.6	14.3	education pulmonary rehabilitation
Browne et al. 2013	UK	In 73	COPD (Diagnosed	—	RCT	69.3	8.9	50.0	—	—	41 ^*	16 ^*	Maintenance
(40)		Co 75	by physician with spirometry test)			67.3	15.1	41.0					pulmonary rehabilitation
Kruis et al. 2013 (41)	The Netherlands	1086	COPD diagnosed (GP medical records) and spirometry test	Hypertension, Diabetes, Depression, Cardiovascular disease	Cluster RCT	68.3	11.2	53.9	—	—	67.8	—	Integrated COPD management
Taylor et al. 2012	UK	In 61	COPD registry or	—	RCT	69.0	9.8	51.28	0.55	0.15	53.9	22.6	Self-Management
(42)		Co 30	community respiratory clinic after spirometry test			70.5	10	34.2
Garcia-Polo et al. 2012 (43)	Spain	All 115	Stable COPD after spirometry test	Depression, anxiety	Cross-sectional	66.90	8.70	93.00	46.40	12.80	43.70	15.10	—
		HR 64				66.6	9.4	93.8			39.6
		Co 51				67.2	8.0	92.2			46.9
Naberan et al. 2012 (44)	Spain	4552	Stable COPD after spirometry test	No co-morbidity	Cross-sectional observational	67.10	10.00	83.30	59.00	20.00	48.30	21.00	—
Egan et al. 2012 (45)	Ireland	47	Stable COPD	No co-morbidity	longitudinal	—	—	—	—	—	46.8	16.6	Rehabilitation
Starkie et al. 2011 (46)	Multi country	3640	COPD registers after spirometry test	—	RCT - TORCH	64.70	8.40	71.00	—	—	—	—	Pharmaceutical
Fletcher et al. 2011 (20)	Multi, six countries	2426	COPD diagnosed by physician, no spirometry	Hypertension, arthritis, anxiety, depression, diabetes	Cross-sectional	56.4		49.00	—	—	—	—	—
Janssen et al. 2011 (47)	The Netherlands	105	COPD outpatient clinic after spirometry test	—	Cross-sectional	66.30	9.20	61.90	—	—	34.10	13.50	—
Khdour et al. 2011 (48)	Ireland	In	COPD outpatient clinic after spirometry test	—	RCT	66.20	9.80	42.20	56.50	9.50	53.40	16.00	Self-management program
		Co				66.60	9.10	45.00	56.10	10.80	51.30	16.30
Pickard et al. 2011 (49)	USA	120	Diagnosed COPD after spirometry test	—	Cross-sectional	71.20	10.30	98.30	59	22	58.40	24.80	—
Agh et al. 2011 (50)	Hungary	170	Outpatient COPD diagnosed after spirometry test	—	Cross-sectional	63.83	11.24	41.8	—	—	64.21	17.34	—
Heyworth et al. 2009 (51)	UK	280	COPD in general practice. No spirometry test	—	Cross-sectional	—	—	—	—	—	—	—	—
Miravitlles et al. 2009 (52)	Spain	827	Stable COPD patients Primary Care after spirometry test	—	Cross-sectional	69	10	86.5	56.9	10.1	54.6	17.7	—
Skoupá et al. 2009 (53)	Czech Republic	Co: 90	In and outpatient	CHD, Depression	Cross-sectional	65.7	10.9	69	—	—	—	—	—
		In: 90		Diabetes		67.1	10.4	61
Ringback et al. 2008 (29)	Denmark	218	COPD outpatients after spirometry test	Musculoskeletal Cardiac disease	Cross-sectional	69.10	8.10	31.90	—	—	34.10	12.20	Rehabilitation
Stellefson et al. 2008 (54)	USA (control)	41	COPD registries in health clinic	—	RCT	61.51	6.29	39	—	—	—	—	Education
Punekar et al. 2007 (55)	Multi	1381	COPD in general and specialist clinic	—	Cross-sectional	66.00	0.29	66.00	—	—	—	—
	five countries	1322				66.00	0.31	71.00
Rutten- van Molken et al. 2006 (56)	Multi	1235	UPLIFT trial after spirometry	—	RCT	64.50	8.40	73.00	—	—	48.77	12.19	Pharmaceutical
Decramer et al. 2005 (57)	Multi	256	COPD patients- Clinic	—	RCT	62	8	79	—	—	57	9	Pharmaceutical
	EU	267				62	8	79			57	9
Brazier et al. 2004 (58)	UK	225	COPD outpatient clinic after spirometry test	—	Cross-sectional	67.00	10.40		—	—	—	—	—
		230
Monninkhof et al.	The	127	Outpatient clinic after	—	RCT	65	—	85	—	—	56.1	—	Self-management
2004 (59)	Nether‐	121	spirometry test			65		84			58.4		program
	lands

HR: high resource group; Co: control group; EU: European countries; In: Intervention group; RCT: randomised Control Trial; Multi: multicounty; FEV1% pred: predicted amount as a percentage of the forced expiratory lung volume in 1 second.

^*Value before randomization.

Table 2. Utility values estimated in included studies.

	Health quality of life		Average estimated health related quality of life (Utility value)
First author/	measure Instrument	Disease			Data collection	Inclusion and exclusion
year of publication	(number of patients)	Severity	Mean	SD/SE	method	criteria
Wu et al. 2015 (30)	EQ-5D Index	I, II, III, IV	0.726	0.150	Interview
	EQ VAS		66.6	16.2
Wilson et al. 2015 (31)	EQ-5D Index In	I, II, III, IV	0.6	0.3	Interview	>35 years, physician
	Co		0.7	0.2		Labelled diagnosis of COPD, emphysema or chronic bronchitis, >20 pack-year smoking history, FEV1of < 80%.
Sundh et al. 2015 (32)	EQ-5D Index	III, IV	0.6887	0.2749	Interview
	EQ VAS		56.5137	23.0107
Stoddart et al. 2015 (33)	EQ-5D Index In	I, II, III, IV	0.4454	0.0301 (SE)	Interview	No other lung disease
	Co		0.4868	0.0211 (SE)
McDowell et al. 2015 (34)	EQ-5D Index In	II, III	0.49	0.35	Interview	diagnosis of moderate to severe COPD
	Co		0.52	0.30		(GOLD stage 2 or 3 and at least two admissions in past 12 months, not having any other respiratory disease
	EQ VAS In		50.1	18.0
	Co		45.5	23.1
Donohue et al. 2015 (35)	EQ-5D Index Trial 1 Co	II, III	0.70	0.228	Interview
	Co		0.68	0.243
	Trial 2 In		0.70	0.229
	Co		0.70	0.225
Lin et al. 2014 (36)	EQ-5D Index	I, II, III, IV	0.79	0.15	Interview	Diagnosed as COPD,
	EQ VAS		70.6	19.6		Spirometry test data, completion of questionnaires
Ferreira et al. 2014 (37)	EQ-5D Index	Not specified	0.86	0.17	Self-administered	-
	SF-6D		0.81	0.12
Chen et al. 2014 (38)	EQ-5D Index	III, IV	0.644	0.306	Interview	FEV1 30–49 % and < 30%
	SF-6D (UK)		0.629	0.133
	EQ-VAS		55.28	20.24
Gillespie et al. 2013 (39)	EQ-5D Index	II, III	0.762	0.252	—	FEV1 ≥ 30% and ≤ 80%, FEV1/FVC < 70%
			0.801	0.232
Browne et al. 2013 (40)	EQ-5D Index In	Not specified	0.6	0.3	Interview	> 35 years, FEV1 <80%, no significant
	Co		0.7	0.2		co-morbidities, participated in at least 60% of the session of the initial PR programme
Kruis et al. 2013 (41)	EQ-5D Index	I, II, III, IV	0.74	0.26	Interview	Exclusion criteria: terminally ill. dementia,
	EQ VAS		67.0	17.4		cognitive impairment, alcohol or drug abusers, not understanding Dutch
Taylor et al. 2012 (42)	EQ-5D Index In	II, III, IV	0.73	0.04	Self-administered	>35 years, FEV1/FVC < 0.7, FEV1 < 80%,
	Co		0.76	0.04		Exclusion criteria: life threatening co-morbidity, psychological impairment, involvement in previous self-management, lacking of English fluency
Garcia-Polo et al. 2012	Global EQ-5D Index	I, II, III, IV	0.72	±0.31	Interview	General (>40 years, Diagnosed COPD
(43)	EQ VAS		58.6	±20.1		Stable COPD, Current or former smoker ≥10 pack-years)
	High resource EQ-5D Index		0.64	±0.35		High RU (history of admission, 2 ER visits, 2 clinic visits in last year)
	EQ VAS		56	±0.22
	Control EQ-5D Index		0.82	±0.19
	EQ VAS		61.9	±21.2
Naberan et al. 2012 (44)	EQ-5D Index	I, II, III, IV	0.7	0.3	Interview	>40 years, Diagnosed COPD, stable
	EQ VAS		59.3	16.5		No co-morbidity
Egan et al. 2012 (45)	EQ-5D Index	Not specified	0.7	±0.3	Not specified	Diagnosed COPD, stable, no Exacerbation in last month, no co-morbidity
Starkie et al. 2011 (46)	EQ-5D Index	II, III, IV	0.73	0.23	Not specified	Confirmed COPD
Fletcher et al. 2011 (20)	EQ-5D Index	I, II, III	0.636	0.007 (SE)	Face to face or telephone Interview	45–67 years, diagnosed COPD, Current or former smoker ≥ 10 pack-years) or biomass exposure, under prescription
Janssen et al. 2011 (47)	EQ-5D Index	III, IV	0.51	0.33	Not specified	Diagnosed COPD, no hospitalization 4
	EQ VAS		62.9	14.0		weeks later on, no nursing home
	AQoL		0.46	0.28
Khdour et al. 2011 (48)	EQ-5D Index	I II, III, IV			Not specified	>45 years, Diagnosed COPD (>1 year),
	In		0.465	0.301		FEV1 <30–80%, no CHF, no learning
	Co		0.485	0.330		difficulty, no sever mobility problem, terminal illness
Pickard et al. 2011 (49)	EQ-5D Index	I, II, III, IV	0.63	0.27	Not specified	Diagnosed COPD
	EQ VAS		65.3	18.9
Agh et al. 2011 (50)	EQ-5D Index	I, II, III, IV	0.55	0.21	Not specified	>40 years, Diagnosed COPD
						No asthma, allergic rhinitis, lung operation, heart failure, liver failure, renal failure
Heyworth et al. 2009 (51)	EQ-5D Index	Not specified	0.53	0.35	Self-administered Postal survey
	EQ VAS		57.5	19.8
Miravitlles et al. 2009	EQ-5D Index	I, II, III, IV	0.64	0.23	Interview face to	>40 years, after exacerbation, Current or
(52)	EQ VAS		55.81	16.83	face	former smoker ≥ 10 pack-years), not admitted, excluding asthma, no significant cognitive problems
Skoupá et al. 2009 (34)	EQ-5D Index Co	II, III, IV	0.582	0.176	Interview	—
	In		0.377	0.229
	EQ VAS Co		71.5	18.7
	In		37.1	17.7
Ringback et al. 2008 (29)	EQ-5D Index	II, III, IV	0.759	0.174	Not specified	Stable COPD patients, FEV1 < 80% no
	EQ VAS		58.6	16.6		significant cardiac or cognitive problems
Stellefson et al. 2008 (54)	EQ-5D Index	Not specified	0.68	0.57	Not specified	> 50 years, Clinical diagnosis of COPD
	EQ VAS		46.07	17.83		Presence of dyspnea, No formal COPD self-management education exposure within the last 6 months
Punekar et al. 2007 (55)	EQ-5D Index (a)	I, II, III, IV	0.70	0.68-0.71	Not specified	40-75 years, FEV1 40–70% ,
	(b)		0.68	0.66-0.69
Rutten- van Molken	EQ VAS	I, II, III, IV	45.00	16.98	Not specified	>40 years, Current or former smoker ≥10
et al. 2006 (56)			0.67	0.21		pack-years, Diagnosed COPD
	EQ-5D Index		0.76	0.21
Decramer et al. 2005 (57)	EQ-5D Index (a)	II, III	0.76	0.22	Interview face to face
	(b)		0.79	0.19
Brazier et al. 2004 (58)	EQ-5D Index	I, II, III, IV	0.540	0.309	Not specified	Exclusion criteria: Other diseases like
	SF-6D		0.572	0.112		asthma, fibrosis and cancer
						FEV1 >80%
Monninkhof et al. 2004 (59)	EQ-5D Index (a)	II, III, IV	0.81	0.017	Not specified	40-75 years, FEV1 ≤ 80%
	(b)		0.82	0.017

COPDSS: COPD Severity Score; PCS and MCS: Physical (PCS) and Mental (MCS) component; Co: control group; In: intervention group; SD: Standard Deviation; SE: Standard Deviation.

Figure 1. Flow diagram for papers included in meta-analysis.

Seventeen studies reported utility values for some COPD stages (including 10 studies that only reported utility values for stages of COPD) (Table 3). One study (20) used British Thoracic Society (BTS) staging system based on Medical Research Council (MRC) dyspnoea scale. Because of similarity in definition of stages I, II and III in this scaling with stages II, III and IV of GOLD staging system respectively, the equivalent utility values were incorporated in meta-analysis. One study (68) used American Thoracic Society staging system (ATS) 1987. Due to similarity in definition of stages II (moderate) and III (severe) in this scaling with stages III and IV of GOLD staging system respectively, the equivalent utility values were incorporated in meta-analysis. One study (46) followed the GOLD staging definition but it merged stages I and II of COPD patients into one single moderate (II) stage and attributed one single utility value for these groups. Utility value of stage II of this study was omitted from meta-analysis. In one study (55) the ‘severe’ (GOLD-stage III) and ‘very severe’ (GOLD-stage IV) subsets were merged into one single ‘severe’ (stage III) subset. Utility value of stage III of this study was omitted from meta-analysis.

Table 3. Values of utility according to the Spirometry staging and COPD severity staging system in included studies.

		COPD severity	GOLD stages (SD) (range) {SE}
First author/year of publication	Utility Instrument	staging system	Stage I	Stage II	Stage III	Stage IV
Wu et al. 2015 (30)	EQ-5D Index	GOLD	0.786 (0.085)	0.734 (0.158)	0.691 (0.155)	0.655 (0.151)
Kim et al. 2014 (60)	EQ-5D Index	GOLD	0.83 {0.04}	0.88 {0.02}	0.81 {0.03}	0.60 {0.04}
Kim et al. 2014 (61)	EQ-5D Index	GOLD	0.906 {0.006}	0.912 {0.005}	0.857 {0.018}	0.780 {0.071}
Jodar-Sanchez et al. 2014 (62)	EQ-5D Index	GOLD	—	—	—	0.55 (0.33)
Samyshkin et al. 2013 (63)	EQ-5D Index	GOLD	—	—	0.751 (0.738-0.765)	0.657 (0.635-0.678)
Solem et al. 2012 (64)	EQ-5D Index	GOLD	—	—	0.701 (0.182)	0.593 (0.236)
Asukai et al. 2012 (27)	EQ-5D Index	GOLD	0.82 (0.8-0.84)	0.801 (0.794-0.809)	0.774 (0.767-0.782)	0.743 (0.730-0.756)
Fletcher et al. 2011 (20)	EQ-5D Index	BTS	0.836 (0.007)	0.579 (0.009)	0.409 (0.015)	—
Pickard et al. 2011 (49)	EQ-5D Index
	(UK value set)	GOLD	0.73 (0.19)	0.59 (0.32)	0.63 (0.25)	0.63 (0.24)
	(US value set)		0.80 (0.13)	0.70 (0.21)	0.72 (0.19)	0.72 (0.16)
Starkie et al. 2011 (46)	EQ-5D Index	GOLD	—	0.752(0.22)	0.708(0.23)	0.672(0.22)
Menn et al. 2010 (65)	EQ-5D Index	GOLD	—	—	0.62 (0.26)	0.60 (0.26)
	SF-6D		—	—	0.61 (0.13)	0.54 (0.08)
Punekar et al. 2007 (55)	EQ-5D Index (a)	GOLD	0.77 (0.73-0.81)	0.68 (0.626-0.72)	0.62 (0.56-0.68)
	(b)		0.68 (0.64-0.72)	0.72 (0.69-0.75)	0.64 (0.61-0.67)
Rutten-van Molken et al. (2007) (The European Journal of Health Economics)(66)	EQ-5D Index	GOLD	—	0.809 {0.008}	0.762 {0.009}	0.655{0.024}
Rutten-van Molken et al. (2006)	EQ-5D Index
(Chest Journal)(56)	(UK value set)	GOLD	—	0.787 (0.771-0.802)	0.750 (0.731-0.768)	0.647 (0.598-0.695)
	(US value set)			0.832 (0.821–0.843)	0.803 (0.790–0.816)	0.731 (0.699–0.762)
Stahl et al. 2003 (67)	EQ-5D Index	GOLD	0.84 (0.15)	0.73 (0.23)	0.74 (0.25)	0.52 (0.26)
Spencer et al. 2005 (68)	EQ-5D Index	ATS	0.81 {0.02}	0.72 {0.03}	0.67 {0.05}	—
Borg et al.2004 (69)	EQ-5D Index	GOLD	0.8971 (0.1117)	0.7551 (0.2747)	0.7481 (0.2991)	0.5493 (0.3129)

EQ-5D, EuroQol 5D: European Quality of Life questionnaire; SF-12: Short-Form Health Survey-12; SF-36: Short-Form Health Survey-36; VAS: visual analogue scale; GOLD: Global Initiative for Chronic Obstructive Pulmonary Disease; ATS: American Thoracic Society staging system; ERS: European Respiratory Society; BTS: British Thoracic Society.

Approaches and measures in COPD

Three studies (four observations) were omitted (28, 70, 71) from the final analysis due to reporting very extreme EQ-5D elicited utility values (<0.008 and >0.96). Attempts were made to contact these authors but the explanations provided did not fully clarify the reasons for the extreme values. The number of participants for general utility scores ranged from 41 to 4803, with an average of 779. Of these, 63.62% were male and the weighted average age was 66.0 years. The weighted average FEV₁% predicted was 45.61 (95% CI 49.518 to 50.103), which indicated severe airflow obstruction according to GOLD guidelines (2011) (72). Mean pack per year smoking cigarette was 44.90. Identifying specific COPD co-morbidities was not possible. Five studies reported the Charlson co-morbidity index.

Meta-analysis

Forest plot

Figure 2^{Figure 3} represents 32 utility values ordered by date of publication. The mean utility value estimated from random effect meta-analysis was 0.673 (95% CI 0.653 to 0.693). There was substantial heterogeneity in the utility values: I² (variation in ES attributable to heterogeneity) = 97.7%, heterogeneity chi-squared = 1348.12, degree of freedom = 31, p < 0.001 and estimate of between-study variance Tau-squared = 0.0029.

Figure 2. Forest plot (random effect) of utility values for COPD patients, general utility values, effect size.

Funnel plot

There was evidence of potential publication bias in this meta-analysis based on Begg's funnel plot (Appendix Figure 1) and on Egger's test (p value < 0.001), but it should be noted that when between-study heterogeneity is large, none of the bias detection tests work well (73). Test of influence of an individual study on the overall meta-analysis estimate, metaninf, did not show significant outliers

Subgroup analyses -interaction tests

The mean utility values for each state of COPD disease estimated from random effect meta-analysis are presented in Table 4 and Figure 3. The estimated utility value for stage I was 0.820 (95% CI 0.767 to 0.872) and the value constantly declined by increasing the severity of disease; 0.782, 0.721 and 0.624 for stages II, III, and IV respectively. Tests of difference between estimated utility means (Table 5) rejected hypothesis of equality of means between stages of COPD, especially between stages II against III and stages III against IV.

Table 4. Estimated mean utility values in general and four stages of COPD (95% confidence interval).

		Heterogeneity chi-squared /Cochran's Q-test
	Utility value /effect size (95% CI)	χ^2	df	P value	I^2 Heterogeneity statistics	Tau –squared
General utility value	0.673 (0.653–0.693)	1348.12	31	<0.001	97.7%	0.0029
Stage I	0.820 (0.767–0.872)	254.29	7	<0.001	97.2%	0.0041
Stage II	0.782 (0.741–0.823)	563.78	10	<0.001	92.9%	0.0013
Stage III	0.721 (0.688–0.753)	639.18	14	<0.001	97.9%	0.0035
Stage IV	0.624 (0.571–0.677)	516.10	14	<0.001	97.9%	0.0099
Overall stages *	0.724 (0.700–0.749)	3481.83	48	<0.001	98.6%	0.0067

^*The overall stages utility value is the result of pooled effect sizes of meta-analyses of utility scores in stages.

Table 5. Difference between estimated utility value means in subgroups.

One-way ANOVA analysis of variance for mean estimated utility by COPD stages	SS	df	MS	F statistics	P value
Between groups	0.2537	3	0.0846	12.40	<0.001
Bartlett's test for equal variances: χ^2(3) = 1.1370 Prob > χ^2 = 0.768
Two-sample t-test with equal
variances	diff	SE	P value Ha: diff < 0	P value Ha: diff ! = 0	P value Ha: diff > 0
Stage I / Stage II	0.042	0.0373141	0.862	0.276	0.138
Stage II / Stage III	0.058	0.032487	0.956	0.088	0.044
Stage III / Stage IV	0.10	0.0306422	0.998	0.004	0.002
Study type: RCT / cross-sectional	0.07	0.0451579	0.931	0.138	0.069
Cigarette: 35-45 Pack-years / 46-55 Pack-years	0.06	0.0457979	0.906	0.188	0.094
FEV1 30–49% / FEV1 50–80%	0.006	0.0505905	0.456	0.912	0.544
Age: < 64 / 65–69	0.016	0.0485167	0.704	0.740	0.297
Year-of-publication
<2008 / 2008–2011	0.130	0.0427302	0.996	0.0088	0.0044
2008–2011/ 2012–2014	0.142	0.031917	0.0002	0.0003	0.9998
2012–2014 / > 2014	0.119	0.0439269	0.9916	0.0168	0.0084

df: degree of freedom; SS: Sum of the Squares; SE: Standard Error; MS: Mean Squire; FEV₁% pred: predicted amount as a percentage of the forced expiratory lung volume in 1 second. The bold p-values are to highlight the significant differences that were observed.

Figure 3. Forest plot (random effect) of utility values for COPD, stages utility, effect size.

Characteristics of study populations

After performing subgroup analysis, there was no evidence of difference in heterogeneity of estimated utility value with age groups of the patients, which was available for all the included studies (Table 6). Some evidences in favour of the effect of study type and cigarette pack-per-year on estimated utility mean were captured (one tailed t-test, Table 5).^{Table 6}

Table 6. Results of interaction tests for subgroup analyses.

		Heterogeneity chi-squared /Cochran's Q test
Group	Utility value /effect size (95% CI)	χ^2	df	P value	I^2 Heterogeneity statistics	Tau-squared
Study type
RCT	0.681 (0.654-0.707)	429.11	12	<0.001	97.2%	0.0020
Cross-sectional	0.669 (0.638-0.700)	873.45	18	<0.001	97.9%	0.0044
Cigarette Pack-years
35-45 Pack-year	0.711 (0.672-0.751)	344.46	5	<0.001	98.5%	0.0024
46-55 Pack-year	0.651 (0.698-0.703)	306.75	6	<0.001	98.0%	0.0046
Not reported	0.665 (0.634-0.696)	681.17	18	<0.001	97.4%	0.0043
FEV1 % pred
FEV1 30-49%	0.658 (0.629-0.687)	293.19	11	<0.001	96.2%	0.0022
FEV1 50-80%	0.658 (0.592-0.725)	350.79	6	<0.001	98.3%	0.0078
Not reported	0.693 (0.661-0.725)	661.61	12	<0.001	98.2%	0.0031
Stages included in the studies
I, II, III, IV	0.682 (0.641-0.723)	435.77	8	<0.001	98.2%	0.0037
I, II, III	0.663 (0.610-0.716)	37.53	1	<0.001	97.3%	0.0014
II, III	0.655 (0.585-0.724)	212.10	4	<0.001	98.1%	0.0058
II, III, IV	0.698 (0.657-0.738)	196.50	5	<0.001	97.5%	0.0023
III, IV	0.618 (0.524-0.712)	26.07	2	<0.001	92.3%	0.0063
Not specified	0.670 (0.584-0.757)	242.13	6	<0.001	97.5%	0.0124
Age
<64	0.692 (0.654-0.731)	381.02	6	<0.001	98.4%	0.0023
65-69	0.678 (0.647-0.708)	814.64	17	<0.001	97.9%	0.0040
>70	0.613 (0.516-0.709)	28.44	2	<0.001	93.0%	0.0067
Charlson Index
< 2.49	0.693 (0.645-0.741)	80.92	2	<0.001	97.5%	0.0018
>2.5	0.615 (0.410-0.821)	23.38	1	<0.001	95.7%	0.0211
Not reported	0.673 (0.650-0.696)	1200.20	26	<0.001	97.8%	0.0032

df: degree of freedom; RCT: Randomized control trial; FEV₁% pred: predicted amount as a percentage of the forced expiratory lung volume in 1 second.

Subgroup analyses were done only when at least two studies were in each subgroup.

Other study characteristics

The interaction tests did not suggest any evidence of difference in utility value and heterogeneity index between the subgroups for country of origin. Interestingly, the general utility value showed a quadratic distribution across year-of-publication (Figure 2). Interaction tests revealed significant change in utility value among groups of year-of-publication but the heterogeneity was remained constant. Utility value was high in studies before 2008, followed by a decline in 2009 to 2011 and a raise in 2012 to 2015. The t-test and ANOVA tests confirmed this trend and the differences (Table 5).

Discussion

This study aimed to summarize utility measures used in COPD and estimate mean utility value for these patients taking the sources of heterogeneity of included studies into account. Thirty-two studies were captured. They reported utility values of COPD based on patient level data. Cross-sectional studies were the dominant type of published studies (nineteen studies). There were in addition, 13 Randomized Control Trial studies. A meta-analysis, controlled for between-study variation, random effect model, calculated mean utility value of 0.673 (95% CI 0.653–0.693) for COPD patients.

This systematic review has revealed substantial diversity in the measuring instrument of HSUV used, and a wide range of utility values in COPD. The utility values ranged from 0.820 (95% CI 0.767–0.872) for stage I to 0.624 (95% CI 0.571–0.677) for stage IV. The meta-analysis indicated a high degree of heterogeneity in utility that was not explained by other factors. The utility score observed in this study is considerably lower than utility score in a general population-based sample, which suggests major impact of COPD on HSUV. For example, a U.S. population-based survey reported a mean utility value of 0.87 (74) on EQ-5D scale. Another representing study from Alberta, Canada, reported a mean utility of 0.91 for individual with no medical problems in a general population survey (75). Similarly, a study presented value set of general population norm of EQ-5D-3L utility value in Queensland, Australia, reported a value of 0.87 (0.86–0.87) (76).

It is well-known that there is inter-instrument variation in the estimation of health utility (77). For this study, in order to reduce diversity and make precise estimation of utility score, meta-analysis was confined only to EQ-5D Index measure. Nevertheless, there was significant utility value diversity between studies which utilized EQ-5D measure (I² = 97.7%).

Clinical and study methodological diversity can both produce heterogeneity, though disaggregation of effects between the two is sometimes very difficult. Patients may be more willing to express the severity of impairment in self-administered than in interviewer-administered questionnaire (78) but the current study did not find evidence against null hypothesis of similarity between these two study subgroups.

Although some included studies did not report spirometry results (40.6%), almost all of them clearly mentioned that COPD diagnostic guidelines were considered and spirometry tests were performed, not only through the registration process (when COPD patient samples were recruited from registry data bases) but also by investigators as part of inclusion criteria. For two studies (20, 51) it was based on General Practitioner diagnosis. An interaction test was performed with subgroup analysis of studies which reported and not reported FEV₁% preb value (Table 6). The test result could not reject null hypothesis of similarity between the two groups. In both groups heterogeneity was very significant and estimated mean utility value were similar.

This study did not show any association between degree of airflow obstruction (FEV₁% pred) and general utility score. This may be explained by the chronic nature of COPD that leads many patients to adjust their lifestyle in accordance with their daily living ability and minimize their sense of functional impairment (79). Another possible reason is related to the limitation of preference-based measures in measuring HSUV in COPD disease. It has been shown that these measures have some limitations in tracing the impact of a disease over time, due to the floor effects with the SF-6D and ceiling effects with the EQ-5D (80). Guyatt et al. (81) pointed out that responsiveness of generic measures to treatment effects in randomized trials in chronic respiratory disease is likely to be limited and may not be valid for measuring longitudinal differences over time. Hesselink et al. (82) reported that changes in FEV₁% pred was weakly correlated with HSUV changes during a 2-year follow-up of COPD patients. These findings were consistent with the results of previous studies (83–85), which implied clinical measures such as FEV₁% pred provided limited information about health condition and were not well correlated with health status of COPD patients. Consistent with these evidences, the new approach of the updated 2014 GOLD report suggests that progression and severity of the COPD disease cannot be drawn in a single-shot picture using only one diagnostic criteria and a combined COPD assessment is needed for prognosis of the disease (72). The combined assessment approach takes three elements into consideration: spirometric test, risk of exacerbations and one of the following disease-specific Hr-QoL measures: COPD Assessment Test (CAT) or COPD Control Questionnaire (CCQ). This method, in conjunction with an assessment of potential co-morbidities, provides a better approach for COPD staging and individualization of the disease management.

Given the current state of knowledge three systematic literature reviews of utility values for COPD disease were published. (79, 86, 87). The aim of these studies was to summarize utility/disutility values in COPD by severity of the disease. Due to the following methodological variations their estimations were different from the current study: 1) In two of these studies, estimated mean utility values for stages of disease were derived from simple mean calculation without incorporating variances around utility values in each included study; in other word meta-analysis was not statistical approach. 2) The current study performed a more comprehensive and, up-to-date systematic literature review and captured more valuable studies for the general and stage specific utility values. 3) In the current study appropriate statistical tests were used to demonstrate sources of heterogeneity and differences in estimated utility values by sub-group analyses. 4) The current study tried to adhere to general recommendations of Peasgood and Brazier (9) in selection of included studies and running meta-analysis.

Another five literature reviews were captured that focused mainly on QoL and outcomes considering variety of interventions in COPD (83, 88–91). The most recent literature review (91) was a qualitative study covering humanistic and economic burden of COPD. In the humanistic section, the study focused on 32 non-RCT studies, of which almost 30% of them were conference abstracts. Different types of HR-QoL measures were included. No quantitative analyses were carried out by this study. Some suggested associations between study characteristics and patient conditions such as demographic, disease symptoms, co-morbidities, resource use and cost were proposed. This study recommended that a comprehensive quantitative study is needed for a reliable conclusion.

In comparison with the findings from the past, current systematic literature review has significant clinical and research implications. In reference to Peasgood and Brazier's critical paper (9) this study tried to overcome major concerns related to meta-analysis of utility estimates in chronic diseases. Very restricted inclusion and exclusion criteria (such as excluding values that were not the appropriate utilities) were applied to capture unbiased study population. Especial attempted were made to generate a pool of utility values elicited from similar health state of COPD patients population. Adopting EQ-5D as the only elicitation method ensured consistency in methodological estimation of utility. All available study characteristics were reported transparently and justification for choosing data from studies were clearly explained. So, modellers can choose the most appropriate estimated value.

There are a few limitations applied to this research. First, the form of aggregated data (study level not individual information) assembled in this study meant that it was not possible to do a more comprehensive meta-regression analysis to investigate correlation of study characteristics (48), demographic diversity (44, 51, 92), clinical staging (25, 53, 93) or health condition differences such as co-morbidities with heterogeneity. Second, COPD patients have a higher prevalence of osteoporosis, anxiety/panic attacks, heart trouble, heart attack, and heart failure, than smokers or non-smokers general population (94, 95). Co-morbidity measured by Charlson Index was only considered by five studies that were included (41, 43, 44, 47, 52). Third, the review did not include non-English language publications unless English versions of their abstracts were available.

For the future research, consideration of specific limitations of some HSUV measure instruments (e.g., celling effect and limited sensitivity in EQ-5D) are essential; using EQ-5D-5L instead of EQ-5D-3L may overcome this limitation.

In conclusion, this study shows considerable inconsistency in utility measures among COPD related published literature. It confirms that the utility value in COPD is considerably lower than the general population. However, the effects of contributing factors such as spirometry assessment and co-morbidities on utility value remain largely unclear. This paper suggests that careful consideration should be taken into account when using systematic method (meta-analysis) for calculation of input parameters in health economic analysis. In case of high level of heterogeneity, appropriate sensitivity analyses are recommended for more accurate health economic appraisals.

Acknowledgments

The authors thank Ms. Rachel Sore (Statistical Consulting Centre, University of Melbourne) for her contribution for statistical analysis.

Declaration of interest statement

The authors report no conflicts of interest. All the authors interpreted data, read and approved the final manuscript. The authors alone are responsible for the content and writing of the paper.

Funding

The first author, Foruhar Moayeri, received PhD scholarship funding from University of Melbourne Faculty of Medicine, Dentistry and Health Sciences.

Appendix 1

^{Table A1}

Table A1. Summary of Medline search strategy.

1	COPD
2	Chronic obstructive pulmonary disease
3	Chronic obstructive lung disease
4	Emphysema
5	Chronic bronchitis
6	1 or 2 or 3 or 4 or 5
7	"quality of life"
8	CUA
9	Euroqol-5d
10	eq-5d
11	eq5d
12	Qaly
13	“Quality -adjusted life year”
14	Health state
15	QoL
16	Utility
17	7 OR 8 OR 9 OR 10 OR 11 OR 12 OR 13 OR 14 OR 15 OR 16
18	6 AND 17

^{Appendix Figure A1}

Appendix Figure A1. Funnel plot of general utility values, included studies of COPD.

Appendix 2: Excluded citations

Not using patient level data

1.	Atsou K, Hejblum G, Chouaid C. Effectiveness and cost-utility estimates of Tiotropium treatment and rehabilitation programs in French patients with Chronic Obstructive Pulmonary Disease. Value in Health 2011; 14(7):A495–A496.
2.	Rutten-van Molken MPMH, Hoogendoorn M, et al. Holistic preferences for 1-year health profiles describing fluctuations in health: the case of chronic obstructive pulmonary disease. PharmacoEconomics 2009; 27(6):465–477.
3.	Schunemann HJ, Stahl E, Austin P, Akl E, Armstrong D, Guyatt GH. A comparison of narrative and table formats for presenting hypothetical health states to patients with gastrointestinal or pulmonary disease. Med Decis Mak 2004; 24(1):53–60.

Not reporting variance of mean or sample size

4.	Asukai Y, Baldwin M, Mungapen L. Utility values for COPD patients based on the EQ-5D questionnarie from three indacaterol phase III stdies. Thorax 2012; 67:A100–A101.
5.	Borg S, Ericsson A, Wedzicha J, Gulsvik A, Lundback B, Donaldson GC, et al. A computer simulation model of the natural history and economic impact of chronic obstructive pulmonary disease. Value Health 2004; 7(2):153–167.
6.	Dollerup J, Poulsen PB, Godtfredsen NS, Grann O, Pors B, Andersen KK, et al. Improvements in quality of life for patients undergoing COPD rehabilitation at municipality health care centres in DENMARK. Value in Health 2009; 12(7): A305–A305.
7.	Igarashi A, Makita H, Fukuda T, Akazawa M, Kato Y, Tsutani K, et al. EQ-5D based QOL assessment in patients with chronic obstructive pulmonary diseases (COPD) in Japan. Value in Health 2009; 12 (3): A127.
8.	Spencer M, Briggs AH, Grossman RF, Rance L. Development of an economic model to assess the cost effectiveness of treatment interventions for chronic obstructive pulmonary disease. Pharmacoeconomics 2005; 23(6):619–637.
9.	Stavem, K. Reliability, validity and responsiveness of two multiattribute utility measures in patients with chronic obstructive pulmonary disease. Quality of Life Res 1999; 8(1–2):45–54.

Non-EQ-5D Index utility measures

10.	Boros PW, Lubinski W. Health state and the quality of life in patients with chronic obstructive pulmonary disease in Poland: A study using the EQ-5D questionnaire. Polskie Archiwum Medycyny Wewnetrznej 2012; 122(3):73–81.
11.	Bourbeau J, Ford G, Zackon H, Pinsky N, Lee J, Ruberto G. Impact on patients' health status following early identification of a COPD exacerbation. Eur Respir J 2007; 30(5):907–913.
12.	Cleland JA, Lee AJ, Hall S. Associations of depression and anxiety with gender, age, health-related quality of life and symptoms in primary care COPD patients. Fam Pract 2007; 24(3):217–223.
13.	Harper R, Brazier JE, Waterhouse JC, Walters SJ, Jones NMB, Howard P. Comparison of outcome measures for patients with chronic obstructive pulmonary disease (COPD) in an outpatient setting. Thorax 1997;52(10):879–887.
14.	Kaplan RM, Ries AL, Reilly J, Mohsenifar Z, Natl Emphys Treatment Tra Res G. Measurement of health-related quality of life in the national emphysema treatment trial. Chest 2004; 126(3):781–789.
15.	Mazur W, Kupiainen H, Pitkaniemi J, Kilpelainen M, Sintonen H, Lindqvist A, et al. Comparison between the disease-specific Airways Questionnaire 20 and the generic 15D instruments in COPD. Health and Quality of Life Outcomes. 2011;9.
16.	Mittmann N, Chan D, Trakas K, Risebrough N. Health utility attributes for chronic conditions. Disease Mgmt Health Outcomes 2001; 9(1):11–21.
17.	Paddison JS, Cafarella P, Frith P. Use of an Australian Quality of Life Tool in Patients with COPD. COPD 2012; 9(6):589–595.
18.	Petrillo, J., Cairns J. Development of the EXACT-U: a preference-based measure to report COPD exacerbation utilities. Value in Health 2011; 14(4):546–554.
19.	Rodriguez Gonzalez-Moro JM, de Lucas Ramos P, Izquierdo Alonso JL, López-Muñiz Ballesteros B, Antón Díaz E, Ribera X, et al. Impact of COPD severity on physical disability and daily living activities: EDIP-EPOC I and EDIP-EPOC II studies. Int J Clin Pract 2009; 63(5):742–750.
20.	Rutten-van Molken MPMH, Hoogendoorn M, Lamers LM. Holistic preferences for 1-year health profiles describing fluctuations in health the case of chronic obstructive pulmonary disease. Pharmacoeconomics 2009; 27(6):465–477.
21.	Ryynanen O-P, Soini EJ, Lindqvist A, Kilpelainen M, Laitinen T. Bayesian predictors of very poor health related quality of life and mortality in patients with COPD. BMC Med Inform Decis Mak 2013;13.
22.	Schunemann HJ, Goldstein R, Mador MJ, McKim D, Stahl E, Puhan M, et al. A randomised trial to evaluate the self-administered standardised chronic respiratory questionnaire. Eur Respir J 2005; 25(1):31–40.
23.	Torrance G, Walker V, Grossman R, Mukherjee J, Vaughan D, La Forge J, et al. Economic evaluation of ciprofloxacin compared with usual antibacterial care for the treatment of acute exacerbations of chronic bronchitis in patients followed for 1 Year. Pharmacoeconomics 1999; 16(5):499–520.
24.	Galaznik A, Chapnick J, Vietri J, Tripathi S, Zou KH, Makinson G. Burden of smoking on quality of life in patients with chronic obstructive pulmonary disease. Expert Rev Pharmacoecon Outcomes Res 2013; 13(6):853–860.
25.	Santana M-J, S-Parrilla J, Mirus J, Loadman MA, Lien DC, Feeny D. An assessment of the effects of Iyengar yoga practice on the health-related quality of life of patients with chronic respiratory diseases: A pilot study. Can Respir J 2013; 20(2):E17–E23.
26.	Koskela J, Kilpelainen M, Kupiainen H, Mazur W, Sintonen H, Boezen M, et al. Co-morbidities are the key nominators of the health related quality of life in mild and moderate COPD. BMC Pulm Med 2014; 14:102.
27.	Koskela J, Kupiainen H, Kilpelainen M, Lindqvist A, Sintonen H, Pitkaniemi J, et al. Longitudinal HRQoL shows divergent trends and identifies constant decliners in asthma and COPD. Respir Med 2014; 108(3):463–471.
28.	Hashim Ali Hussein, S., Nielsen, L. P., Konow Bøgebjerg Dolberg, M., & Dahl, R. Serum magnesium and not vitamin D is associated with better QoL in COPD: A cross-sectional study. Respir Med 2015; 109(6):727–733.
29.	Hutchinson AF, Graco M, Rasekaba TM, Parikh S, Berlowitz DJ, Lim WK. Relationship between health-related quality of life, comorbidities and acute health care utilisation, in adults with chronic conditions. Health Qual Life Outcomes 2015;13.

Citations reported utility score from other primary studies

30.	Hettle R, Wouters H, Ayres J, Gani R, Kelly S, Lion M, et al. Cost-utility analysis of tiotropium versus usual care in patients with COPD in the UK and Belgium. Respir Med 2012; 106(12):1722–1733.
31.	Neyt M, Devriese S, Thiry N, Van den Bruel A. Tiotropium's cost-effectiveness for the treatment of COPD: a cost-utility analysis under real-world conditions. BMC Pulm Med 2010;10:47.
32.	Price D, Asukai Y, Ananthapavan J, Malcolm B, Radwan A, Keyzor I. A UK-based cost-utility analysis of indacaterol, a once-daily maintenance bronchodilator for patients with COPD, using real world evidence on resource use. Appl Health Econ Health Pol 2013; 11(3):259–274.
33.	Price D, Gray A, Gale R, Asukai Y, Mungapen L, Lloyd A, et al. Cost-utility analysis of indacaterol in Germany: A once-daily maintenance bronchodilator for patients with COPD. Respir Med 2011; 105(11):1635–1647.
34.	Schuenemann HJ, Norman G, Puhan MA, Stahl E, Griffith L, Heels-Ansdell D, et al. Application of generalizability theory confirmed lower reliability of the standard gamble than the feeling thermometer. J Clin Epidemiol 2007; 60(12):1256–1262.
35.	Szende A, Leidy NK, Stahl E, Svensson K. Estimating health utilities in patients with asthma and COPD: evidence on the performance of EQ-5D and SF-6D. Qual Life Res 2009;18(2):267–272.
36.	Miravitlles M, Naberan K, Cantoni J, Azpeitia A. Socioeconomic Status and Health-Related Quality of Life of Patients with Chronic Obstructive Pulmonary Disease. Respiration 2011; 82(5):402–408.
37.	Miravitlles M, Cantoni J, Naberan K. Factors associated with a low level of physical activity in patients with chronic obstructive pulmonary disease. Lung 2014; 192(2):259–265.
38.	Miravitlles M, Huerta A, Alberto Fernandez-Villar J, Alcazar B, Villa G, Forne C, et al. Generic utilities in chronic obstructive pulmonary disease patients stratified according to different staging systems. Health Qual Life Outcomes 2014;12.
39.	Miravitlles M, Huerta A, Valle M, Garcia-Sidro P, Forne C, Crespo C, et al. Clinical variables impacting on the estimation of utilities in chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis 2015; 10:367–376.
40.	Miravitlles, M., Molina, J., Quintano, J. A., Campuzano, A., Pérez, J., & Roncero, C. Factors associated with depression and severe depression in patients with COPD. Respir Med 2014; 108(11):1615–1625.
41.	Wilke S, Janssen DJA, Wouters EFM, Schols JMGA, Franssen FME, Spruit MA. Correlations between disease-specific and generic health status questionnaires in patients with advanced COPD: a one-year observational study. Health Qual Life Outcomes 2012;10.

Citations with mixed utility value for COPD and other type of obstructive pulmonary diseases

42.	Guyatt GH, King DR, Feeny DH, Stubbing D, Goldstein RS. Generic and specific measurement of health-related quality of life in a clinical trial of respiratory rehabilitation. J Clin Epidemiol 1999; 52(3):187–192.
43.	Roy AN, Madhavan S. Patient Reported Health-related Quality of Life in Co-morbid Insomnia: Results from a Survey of Primary Care Patients in the United States. Primary Health Care, 2014; 4: 160.

Citation with extreme utility values

44.	Calverley PMA, Rabe KF, Goehring UM, Kristiansen S, Fabbri LM, Martinez FJ, et al. Roflumilast in symptomatic chronic obstructive pulmonary disease: two randomised clinical trials (vol 374, pg 685, 2009). Lancet 2010; 376(9747):1146.
45.	O'Reilly JF, Williams AE, Rice L. Health status impairment and costs associated with COPD exacerbation managed in hospital. Int J Clin Pract 2007; 61(7):1112–1120.
46.	Koo H-K, Park J-H, Park HK, Jung H, Lee S-S. Conflicting Role of Sarcopenia and Obesity in Male Patients with Chronic Obstructive Pulmonary Disease: Korean National Health and Nutrition Examination Survey. PLoS One 2014; 9(10).

Citation not reporting detailed data

47.	Covelli H, Bhattacharya S, Cassino C, Conoscenti C, Kesten S. Absence of electrocardiographic findings and improved function with once-daily tiotropium in patients with chronic obstructive pulmonary disease. Pharmacotherapy 2005; 25(12):1708–1718.
48.	Fabbri LM, Calverley PMA, Luis Izquierdo-Alonso J, Bundschuh DS, Brose M, Martinez FJ, et al. Roflumilast in moderate-to-severe chronic obstructive pulmonary disease treated with longacting bronchodilators: two randomised clinical trials. Lancet 2009; 374(9691):695–703.
49.	Hunger M, Thorand B, Schunk M, Doering A, Menn P, Peters A, et al. Multimorbidity and health-related quality of life in the older population: results from the German KORA-Age study. Health and Qual Life Outcomes 2011; 9.
50.	Miller JD, Malthaner RA, Goldsmith CH, Goeree R, Higgins D, Cox PG, et al. A Randomized clinical trial of lung volume reduction surgery versus best medical care for patients with advanced emphysema: A two-year study from Canada. Ann Thorac Surg 2006; 81(1):314–321.
51.	Mueller TA, Wiren A, Small M, Cristino J, Pike J. Impact of cough and/or sputum symptoms on health-related quality of life in COPD patients: An observational, cross-sectional study in Europe and the USA. Value Health 2009; 12(7): A306–A306.
52.	Paterson C, Langan CE, McKaig GA, Anderson PM, Maclaine GDH, Rose LB, et al. Assessing patient outcomes in acute exacerbations of chronic bronchitis: The measure your medical outcome profile (MYMOP), medical outcomes study 6-item general health survey (MOS-6A) and EuroQol (EQ-5D). Qual Life Res 2000; 9(5):521–527.
53.	Ramsey SD, Patrick DL, Albert RK, Larson EB, Wood DE, Raghu G. The cost-effectiveness of lung transplantation – a pilot study. Chest 1995; 108(6):1594–1601.

Citations with inconsistent utility values with their reference article

54.	Chong J, Karner C, Poole P. Tiotropium versus long-acting beta-agonists for stable chronic obstructive pulmonary disease. Cochrane Database System Rev 2012(9); DOI: 10.1002/14651858.CD009157.pub2
55.	Gani R, Griffin J, Kelly S, Rutten-van Molken M. Economic analyses comparing tiotropium with ipratropium or salmeterol in UK patients with COPD. Primary Care Respir J 2010; 19(1):68–74.

Citations with patients at exacerbation state

56.	Miravitlles M, Izquierdo I, Herrejon A, Vicente Torres J, Baro E, Borja J, et al. COPD severity score as a predictor of failure in exacerbations of COPD. The ESFERA study. Respir Med 2011; 105(5):740–747.
57.	Goossens LMA, Nivens MC, Sachs P, Monz BU, Rutten-van Molken MPMH. Is the EQ-5D responsive to recovery from a moderate COPD exacerbation? Respir Med 2011; 105(8):1195–1202.
58.	Cross J, Elender F, Barton G, Clark A, Shepstone L, Blyth A, et al. A randomised controlled equivalence trial to determine the effectiveness and cost-utility of manual chest physiotherapy techniques in the management of exacerbations of chronic obstructive pulmonary disease (MATREX). Health Technol Assess 2010; 14(23):1-147, iii–iv.
59.	Solem CT, Sun SX, Sudharshan L, Macahilig C, Katyal M, Gao X. Exacerbation-related impairment of quality of life and work productivity in severe and very severe chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis 2013; 8:641–652.
60.	Antoniu SA, Puiu A, Zaharia B, Azoicai D. Health status during hospitalisations for chronic obstructive pulmonary disease exacerbations: the validity of the Clinical COPD Questionnaire. Expert Rev Pharmacoecon Outcomes Res 2014; 14(2):283–287.

Citations without spirometry confirmation test

61.	Arne M, Janson C, Janson S, Boman G, Lindqvist U, Berne C, et al. Physical activity and quality of life in subjects with chronic disease: Chronic obstructive pulmonary disease compared with rheumatoid arthritis and diabetes mellitus. Scand J Prim Health Care 2009; 27(3):141–147.

No defined specific inclusion and exclusion criteria

62.	Tsiachristas A, Cramm JM, Nieboer AP, Rutten-van Molken MPMH. Changes in costs and effects after the implementation of disease management programs in the Netherlands: variability and determinants. Cost Effect Resour Alloc 2014; 12(17).