TABLE 1

Summary of findings

Health state/outcome
(categories of values and preferences)
Estimates of outcome importance range across studies/pooled mean (95% CI)Participants/studiesCertainty of evidenceInterpretation of findings
Exacerbation (utility# measured with visual analogue scale)0.259–0.580/0.462 (0.453–0.471)####1991/8####+++−
Moderate certainty due to inconsistency####
Most people find exacerbation of COPD probably has a large impact on their lives. There is probably no important variability for this assessment.
Exacerbation (EQ-5D utility+)0.430–0.683/0.519 (0.502–0.537)¶¶¶¶927/3¶¶¶¶++−−
Low certainty due to inconsistency and indirectness¶¶¶¶,######
Most people find exacerbation of COPD probably has a large impact on their lives. There is probably no important variability for this assessment.
Exacerbation (disutility)§Visual analogue scale
 1 nonserious exacerbation: −0.037 (0.005)
 2 nonserious exacerbations: −0.068 (0.005)
 1 serious exacerbation: −0.090 (0.007)
 1 nonserious and 1 serious exacerbation: −0.130 (0.007)
Time trade off
 1 nonserious exacerbation: −0.010 (0.007)
 2 nonserious exacerbations: −0.021 (0.007)
 1 serious exacerbation: −0.042 (0.009)
 1 nonserious and 1 serious exacerbation: −0.088 (0.009)
239/1++++
High certainty
Most people find exacerbation of COPD has an impact on their lives, which grows larger as the severity of exacerbation progresses. There is probably no important variability for this assessment.
Level 1 dyspnoea (utility measured with visual analogue scale)ƒ0.751146/1ƒ++++
High certaintyƒ
Most people find level 1 dyspnoea has a small to moderate impact on lives. There is probably no important variability for this assessment.
Level 2 dyspnoea (utility measured with visual analogue scale)ƒ0.65645/1ƒ+++−
Moderate certainty due to imprecisionƒ
Most people find level 2 dyspnoea probably has a moderate impact on lives. There is probably no important variability for this assessment.
Level 3 dyspnoea (utility measured with visual analogue scale)ƒ0.5267/1ƒ++−−
Low certainty due to very serious imprecisionƒ
Most people find level 3 dyspnoea probably has a large impact on lives. There is probably no important variability for this assessment.
Adverse events (discrete choice)##Two studies suggested that patients consider adverse events as important outcomes. One study suggested adverse events were more important than onset time of medicine, ease of use, rescue medicine use. Another suggested adverse events were more important than costs of treatment, extent to which the patient sees the same doctor each time and extent to which the doctor treats the patient as an entire person. Both studies concluded symptom relief to be more important than adverse events.564/2+++−
Moderate certainty due to risk of bias##
People probably consider adverse events as an important outcome. There is likely no important variability for this assessment.
Extent of symptom relief (discrete choice)##Two studies compared extend of symptom relief with other outcomes. Extent of symptom relief was considered the most important outcome in these two studies.564/2+++−
Moderate certainty due to risk of bias##
Most people probably find symptom relief as important outcome. There is probably no important variability for this assessment.
Extent of symptom relief (forced choice)¶¶In a survey on expectation of treatment, greater symptomatic relief was chosen by 82.3% of the participants, thus the most important outcome. Extent of symptom relief was considered the second most important outcome in one cross-sectional study (less preferred “not to be kept alive on life support when there is little hope for a meaningful recovery”). Another study reported 58.0% of the participants would prefer treatment focusing on relieving pain and discomfort rather than extending life.1640/3+++−
Moderate certainty due to risk of bias¶¶
Most people probably find symptom relief as important outcome. There is probably no important variability for this assessment.
Very severe COPD (utility measured with visual analogue scale)++0.321–0.651/0.345 (0.335–0.354)++++746/7++−−
Low certainty due to risk of bias++++++ and inconsistency++
Most people find very severe COPD seems to have a large impact on lives. There is probably important variability for this assessment.
Very severe COPD (EQ-5D utility)§§0.520–0.740/0.681 (0.667–0.694)§§§§898/10+++−
Moderate certainty due to inconsistency§§§§
Most people find very severe COPD probably has a large impact on lives. There is probably important variability for this assessment.
Severe COPD (utility measured with visual analogue scale)ƒƒ0.446–0.689/0.508 (0.501–0.515)ƒƒƒƒ4683/8++−−
Low certainty due to risk of bias++++++ and inconsistency ƒƒƒƒ
Most people find severe COPD probably has a moderate to large impact on lives. There is probably important variability for this assessment.
Severe COPD (EQ-5D utility)###0.620–0.810/0.741 (0.734–0.749)#####4352/11+++−
Moderate certainty due to inconsistency#####
Most people find severe COPD probably has a moderate to large impact on lives. There is probably important variability for this assessment.
Moderate COPD (utility measured with visual analogue scale)¶¶¶0.589–0.726/0.639 (0.635–0.642)¶¶¶¶¶9664/10++−−
Low certainty due to risk of bias++++++ and inconsistency¶¶¶¶¶
Most people find moderate COPD probably has a moderate impact on lives. There is probably important variability for this assessment.
Moderate COPD (EQ-5D utility)+++0.680–0.890/0.821 (0.815–0.826)+++++4620/9+++−
Moderate certainty due to inconsistency+++++
Most people find moderate COPD probably has a moderate impact on lives. There is probably important variability for this assessment.
Mild COPD (utility measured with visual analogue scale)§§§0.680–0.811/0.738 (0.732-0.746)§§§§§3623/8++−−
Low certainty due to risk of bias++++++ and inconsistency§§§§§
Most people find moderate COPD probably has a small to moderate impact on lives. There is likely important variability for this assessment.
Mild COPD (EQ-5D utility)ƒƒƒ0.770–0.900/0.873 (0.863–0.883)ƒƒƒƒƒ2067/7+++−
Moderate certainty due to inconsistency ƒƒƒƒƒ
Most people find moderate COPD probably has a small to moderate impact on lives. There is probably important variability for this assessment.

Grading of Recommendations Assessment, Development and Evaluation Working Group grades of evidence: here we assessed the certainty of evidence on mean outcome importance. We use “certainty of evidence”, “certainty in estimates”, “quality of evidence” and “strength of evidence” interchangeably. High certainty: we are very confident that the true value of outcome importance lies close to that of the estimate. Moderate certainty: we are moderately confident in the estimate; the true value of outcome importance is likely to be close to the estimate but there is a possibility that it is substantially different. Low certainty: our confidence in the estimate is limited; the true value of outcome importance may be substantially different from the estimate. Very low certainty: we have very little confidence in the estimate; the true value of outcome importance is likely to be substantially different from the estimate EQ-5D: EuroQol-5D; COPD: chronic obstructive pulmonary disease. #: utilities represent the value individuals place on different outcomes; they are measured on an interval scale, with 0 reflecting states of health equivalent to death/worst imaginable health and 1 (or 100 in some cases) reflecting perfect health/best imaginable health. : eight studies [19–26] used EQ-5D visual analogue scale to elicit health state values on exacerbation of COPD. +: three studies [21–23] used EQ-5D utility to elicit the importance of outcome. §: Rutten van Molken et al. [27] reported the disutility due to the exacerbations; the measurement tools included visual analogue scale and time trade off; the researchers estimated the disutility due to exacerbation using random effects regression analysis. ƒ: Kim et al. [28] reported the utility of dyspnoea, according to the levels of breathlessness (level 1, short of breath during strenuous activities; level 2, stopping to catch breath after a few minutes walking; level 3, breathless when dressing or washing); in a total sample of 200, the numbers of participants experiencing level 1, 2 and 3 breathlessness were 146, 45 and seven, respectively; due to small sample size, we downgraded the certainty of evidence by one level for the estimates of level 2 breathlessness and two levels for level 3 breathlessness. ##: Bulcun et al. [29] compared extent of symptom relief with extent to which the doctor gives sufficient time to listen to the patient, possibility of experiencing adverse effects from treatment, costs of treatment, extent to which the patient sees the same doctor each time and extent to which the doctor treats the patient as an entire person; Kawata et al. [30] recruited 515 patients for an online voluntary survey on the comparison of importance of symptom relief, speed of symptom relief, rescue medicine use and side-effects; participants' eligibility and their answers were considered as having serious risk of bias. ¶¶: three studies [31–33] asked directly what participants would prefer in facing a COPD treatment decision; the questions included expectation of treatment, reasons to continue or not continue with treatment and preferred treatment characteristics; the assessment was at risk of bias due to unclear reliability and validity features. ++: [27, 28, 34–37]. §§: [27, 28, 35–42]. ƒƒ: [27, 28, 34–37, 43, 44]. ###: [28, 35–40, 43, 44]. ¶¶¶: [27, 28, 34–37, 43, 44–46]. §§§: [27, 28, 34–37, 43, 45]. ƒƒƒ: [28, 35–37, 42, 43, 45]. ####: across eight included studies, the point estimates range from 0.259 to 0.580; using inverse-variance method to pool the estimates, the I2 (95.5%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study populations cannot explain the source of heterogeneity (the participants in the studies were exacerbation patients, exacerbation patients not needing hospitalisation, ambulatory patients and hospitalised patients due to exacerbation). ¶¶¶¶: across three included studies, the point estimates range from 0.430 to 0.683; using inverse-variance method to pool the estimates, the I2 (95.5%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study populations cannot explain the source of heterogeneity (the participants in the three studies were exacerbation patients, exacerbation patients not needing hospitalisation and ambulatory patients). ++++: the point estimates range from 0.321 to 0.651; using inverse-variance method to pool the estimates, the I2 (98.5%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. §§§§: the point estimates range from 0.520 to 0.740; using inverse-variance method to pool the estimates, the I2 (80.2%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. ƒƒƒƒ: the point estimates range from 0.446 to 0.689; using inverse-variance method to pool the estimates, the I2 (98.8%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. #####: the point estimates range from 0.620 to 0.810; using inverse-variance method to pool the estimates, the I2 (94.5%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. ¶¶¶¶¶: the point estimates range from 0.589 to 0.726; using inverse-variance method to pool the estimates, the I2 (97.9%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. +++++: the point estimates range from 0.680 to 0.890; using inverse-variance method to pool the estimates, the I2 (97.8%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. §§§§§: the point estimates range from 0.680 to 0.811; using inverse-variance method to pool the estimates, the I2 (88.0%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. ƒƒƒƒƒ: the point estimates range from 0.770 to 0.900; using inverse-variance method to pool the estimates, the I2 (91.3%) and statistical test (<0.001) suggest potential heterogeneity across studies; the difference in study population cannot explain the source of heterogeneity. ######: we rated down the quality of evidence for indirectness because an indirect measurement tool (EQ-5D) was used to elicit the utility of outcomes. ++++++: we downgraded the certainty in evidence because of low response rate observed by Lin et al. [35] and the potentially biased sampling strategy by asking physicians to provided recruited patients by Boros and Lubinski [34].