COPD Assessment Test

COPD Assessment Test in COPD Trials and Economic Evaluation

The COPD Assessment Test, usually shortened to CAT, asks a person with chronic obstructive pulmonary disease to rate eight aspects of how the disease affects daily life and adds the ratings into one total out of 40. Its developers designed it for assessing and monitoring COPD, trials of COPD treatments report it as an outcome, and the Global Initiative for Chronic Obstructive Lung Disease (GOLD) uses a CAT score of 10 as the cut-point that separates more symptomatic patients in its initial assessment of COPD. The total carries no preference weights, so a change in the CAT is evidence of benefit that has to be converted before it can count in a QALY, and a published mapping to EQ-5D warns that the conversion is likely to understate treatment effects. GOLD now also calls the questionnaire the Chronic Airways Assessment Test (CAAT), and here CAT means this COPD instrument, not computerised adaptive testing or any other use of the letters.

How the eight items were chosen and scored

Jones and colleagues built the CAT in three prospective international studies in Europe and the USA with 1,503 patients. They started with 21 candidate items identified through qualitative research with people who had COPD, and psychometric and Rasch analyses identified eight items that fit a unidimensional model.

The eight items, as listed in the later European validation study by Jones and colleagues, cover cough, phlegm, chest tightness, breathlessness going up hills or stairs, limitation of activities at home, confidence leaving home, sleep and energy. On the questionnaire itself each item has a contrasting statement at each end and six response points, scored 0 to 5, and the developer's user guide gives a scoring range of 0 to 40 for the summed total.

$$\text{CAT} = \sum_{i=1}^{8} x_i, \qquad x_i \in \{0, 1, 2, 3, 4, 5\}$$

where $x_i$ is the score on item $i$ and the sum over the eight items gives a total between 0 and 40, with a higher total meaning a greater impact of COPD on the patient's life.

The developer's user guide states that the CAT was validated as an eight-item questionnaire and that the questions should not be split up or used independently of each other. A study that reports only the breathlessness item, or drops the sleep item, is therefore no longer using the CAT. The test was developed with support from GSK, and the developer's site states that GSK's activities in connection with the test are monitored by a supervisory council that includes external, independent experts.

The GOLD 2025 report notes that the CAT has been renamed the Chronic Airways Assessment Test (CAAT), a change made to make its use easier in other chronic airway diseases. The GOLD 2026 report states that the CAT and the CAAT are equivalent and that their scores are interchangeable, so a model or evidence synthesis can pool results reported under either name.

Reliability, validity and the cut-point of 10

In the development studies the eight items had internal consistency reliability of Cronbach's alpha 0.88, and test-retest reliability in 53 stable patients gave an intra-class correlation of 0.8. In the US sample the CAT correlated with the COPD-specific version of the St George's Respiratory Questionnaire at r = 0.80, a convergent validity result against the older and longer instrument. Stable patients (229) and patients in an exacerbation (67) differed by five units on the 40-point scale, a difference of 12%.

The developer's user guide pairs CAT totals with descriptive impact levels, labelling scores above 20 as high impact and scores above 30 as very high impact, and advises that scores are read alongside the patient's disease severity. These labels describe the size of the burden reported by the patient and are not values attached to health states.

GOLD sets its CAT cut-point as the equivalent of an SGRQ threshold. The GOLD 2025 report describes the SGRQ as the most widely documented comprehensive measure, notes that SGRQ scores below 25 are uncommon in diagnosed COPD and that scores of 25 or more are very uncommon in healthy persons, and gives 10 as the equivalent cut-point for the CAT. The same report states that no equivalent cut-point can be calculated for the modified Medical Research Council (mMRC) breathlessness scale, which it keeps at 2 or more as a threshold for separating less breathlessness from more, and it recommends a comprehensive symptom assessment. The GOLD 2026 report keeps the same cut-point of 10 and describes the CAAT or CAT as an eight-item questionnaire scored 0 to 40.

In GOLD's ABE assessment tool, groups A and B are separated by symptom burden, with a CAT of 10 (or an mMRC grade of 2) as the dividing line, while group E is defined by exacerbation history. Under the 2023 to 2025 rule, group E meant two or more moderate exacerbations or one leading to hospitalisation. The GOLD 2026 report adjusted the criteria for groups A, B and E in response to emerging evidence, and its summary of key changes states that the threshold of one moderate exacerbation should now be used to consider treatment escalation. A decision model that assigns its starting cohort to GOLD groups therefore inherits the CAT cut-point, and group membership can differ between editions.

Minimal clinically important difference of the CAT

A minimal clinically important difference answers how large a change in one patient's CAT score counts as meaningful, which is a different question from where the GOLD cut-point sits. Kon and colleagues ran three prospective studies at two centres in London between April 2010 and December 2012 and combined anchor-based and distribution-based methods.

In their first study, pulmonary rehabilitation changed the mean CAT score by minus 2.5 (95% confidence interval minus 3.0 to minus 1.9), and the change correlated with changes in the SGRQ and the Chronic Respiratory Questionnaire. Linear regression against anchors gave estimates of about minus 1.2 to minus 2.8, receiver operating characteristic curves consistently identified minus 2, and distribution-based estimates ran minus 3.3 to minus 3.8. They concluded that the most reliable estimate of the minimum important difference of the CAT is 2 points.

The developer's user guide uses the same figure, stating that a difference or change of 2 or more units over 2 to 3 months in a patient suggests a clinically significant difference or change in health status. In a trial the 2-point figure can define responders, and the share of responders in each arm becomes an input to a model alongside or instead of the mean difference.

The CAT in cost-utility analysis and mapping

A CAT total is a count of item scores with no preference weights, so equal steps are not known to be equally valued and 0 is not anchored to full health. The score cannot serve as a health state utility value in a cost-utility analysis. When a COPD trial has collected the CAT but not the EQ-5D, a mapping function can predict EQ-5D utilities from CAT scores, following the general design of a mapping study.

Hoyle, Tabberer and Brooks, writing from the Office of Health Economics and GSK, developed such an algorithm in 2016 for use in COPD health technology assessment. They used EQ-5D-3L and CAT data from two multinational phase III trials and compared ordinary least squares, multinomial logistic regression and a hybrid of the two. Ordinary least squares performed about as well as the more complex models on root mean squared error and mean absolute error. Within-sample validation showed over- and underestimation in the utility range 0.5 to 0.9, although every mean absolute error was 0.100 or below.

Their conclusion matters for any model that leans on the algorithm. Without EQ-5D data the ordinary least squares algorithm may be used to estimate utility for treatment models, but it is likely to underestimate treatment effects, and the authors recommended that utilities for UK COPD technology assessments be derived directly from the EQ-5D. A CAT-based utility is therefore a fallback that is likely to bias the incremental QALY towards zero, which raises the incremental cost-utility ratio of an effective treatment.

Illustrative worked example of a CAT responder analysis and a mapped QALY gain

All numbers in this example are illustrative and come from no trial or published algorithm. A 12-month trial randomises 150 patients to a new inhaler and 150 to usual care. The mean CAT change is minus 3.0 with the new inhaler and minus 1.0 with usual care, and the responder threshold is a fall of at least 2 points.

1. Classifying one patient. A patient whose CAT total was 18 at baseline and 15 at 12 months has fallen 18 − 15 = 3 points, so the patient is a responder. Both totals are 10 or more, so the patient stays on the more symptomatic side of the GOLD symptom split despite responding.

2. Responder proportions. If 90 of 150 patients respond with the new inhaler and 60 of 150 with usual care, the proportions are 90 / 150 = 0.60 and 60 / 150 = 0.40. The difference is 0.60 − 0.40 = 0.20, so 1 / 0.20 = 5 patients must be treated for one extra responder.

3. Mean difference and mapped utility. The difference in mean CAT change is 3.0 − 1.0 = 2.0 points. Suppose an illustrative linear mapping in which each CAT point is worth 0.01 of EQ-5D utility. The mapped utility gain is 2.0 × 0.01 = 0.02, and if it lasts the full year the QALY gain is 0.02 × 1 = 0.02.

4. Cost per QALY.

$$\text{ICUR} = \frac{\Delta C}{\Delta Q} = \frac{500}{0.02} = 25000$$

where $\Delta C$ is the illustrative incremental cost of GBP 500 per patient and $\Delta Q$ is the mapped QALY gain of 0.02.

The mapped analysis gives GBP 25,000 per QALY. If EQ-5D had been collected directly and showed a utility gain of 0.025 for the year, the ratio would be 500 / 0.025 = 20,000, or GBP 20,000 per QALY. The direction of this change is what Hoyle and colleagues warned about, since a mapping that underestimates treatment effects can move a treatment across a decision threshold. Kon's 2-point MCID describes change within a patient, so a difference of 2 points between arm means is a different quantity and cannot be read as proof that the average patient gained a meaningful benefit. Either way, a CAT difference says nothing about value until it is turned into QALYs and set against cost.

How the CAT differs from the SGRQ and other measures

The CAT and the SGRQ both measure the impact of COPD on health status, and the SGRQ was the reference against which the CAT was validated. The SGRQ is longer, scores each response with an empirically derived weight on a 0 to 100 scale, has its own 4-unit threshold for important change, and carries a longer record as a trial endpoint. The CAT is a simple sum of eight items and is quick enough for routine consultations. GOLD treats a CAT of 10 as equivalent to an SGRQ of 25, but that equivalence is a cut-point and does not convert one score into the other across the range.

The CAT is a patient-reported outcome measure, a disease-specific instrument and a non-preference-based measure, and it records health-related quality of life as the patient experiences COPD. It differs from the mMRC scale, which grades breathlessness alone, and GOLD has stated that no mMRC cut-point equivalent to the SGRQ threshold can be calculated. It differs from the EQ-5D, a generic instrument with no respiratory content whose value sets turn responses into utilities.

Limitations and common misreadings of CAT scores

Several problems recur when CAT scores are carried into economic work. Most of them come from treating a short, unweighted total as more than it is, or from carrying a figure into a setting its source did not study.

• Reading the cut-point as a threshold for benefit. A CAT of 10 is GOLD's dividing line for symptom burden at initial assessment, whereas the 2-point figure concerns change. A patient whose score falls 14 to 11 has improved by more than the MCID while staying above the cut-point.
• Choosing one MCID without saying why. Kon and colleagues found anchor-based estimates near 2 and distribution-based estimates of 3.3 to 3.8, and a responder analysis changes with the threshold chosen.
• Treating a mapped utility as a measured one. Hoyle and colleagues found the mapping likely to underestimate treatment effects, so a model built on CAT-mapped utilities should show how far the result depends on the mapping and should prefer direct EQ-5D data where they exist.
• Using parts of the questionnaire. The developer states the eight items should not be split up, so subscores built from single items are outside the validated instrument.
• Taking impact labels for utilities. The user guide's high and very high impact labels describe reported burden and carry no preference weights.

Sources

• Global Initiative for Chronic Obstructive Lung Disease. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease, 2025 Report. https://goldcopd.org/wp-content/uploads/2024/11/GOLD-2025-Report-v1.0-15Nov2024_WMV.pdf
• Global Initiative for Chronic Obstructive Lung Disease. GOLD Report 2026, Key Changes Summary. November 2025. https://goldcopd.org/wp-content/uploads/2025/11/KEY-CHANGES-GOLD-2026-10Nov2025.pdf
• Global Initiative for Chronic Obstructive Lung Disease. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease, 2026 Report (version 1.3, 8 December 2025). https://goldcopd.org/wp-content/uploads/2026/01/GOLD-REPORT-2026-v1.3-8Dec2025_WMV2.pdf
• Global Initiative for Chronic Obstructive Lung Disease. 2023 GOLD Pocket Guide (version 1.2). https://goldcopd.org/wp-content/uploads/2023/03/POCKET-GUIDE-GOLD-2023-ver-1.2-17Feb2023_WMV.pdf
• GSK. COPD Assessment Test (CAT) Healthcare Professional User Guide. https://www.catestonline.org/content/dam/global/catestonline/documents/CAT_HCP%20User%20Guide.pdf (accessed 17 October 2026)
• Hoyle CK, Tabberer M, Brooks J. Mapping the COPD Assessment Test onto EQ-5D. Value in Health. 2016;19(4):469-477. https://pubmed.ncbi.nlm.nih.gov/27325339/
• Jones PW, Harding G, Berry P, Wiklund I, Chen WH, Kline Leidy N. Development and first validation of the COPD Assessment Test. European Respiratory Journal. 2009;34(3):648-654. doi:10.1183/09031936.00102509
• Jones PW, Brusselle G, Dal Negro RW, et al. Properties of the COPD assessment test in a cross-sectional European study. European Respiratory Journal. 2011;38(1):29-35. https://publications.ersnet.org/content/erj/38/1/29
• Kon SSC, Canavan JL, Jones SE, et al. Minimum clinically important difference for the COPD Assessment Test: a prospective analysis. Lancet Respiratory Medicine. 2014;2(3):195-203. doi:10.1016/S2213-2600(14)70001-3