Cost-effectiveness of targeted COPD case-finding with the PUMA questionnaire followed by spirometry among ever-smokers in Singapore

Abstract

Background

Chronic obstructive pulmonary disease (COPD) is frequently undiagnosed among ever-smokers. We evaluated the cost-effectiveness of targeted COPD case-finding using the PUMA questionnaire followed by spirometry among ever-smokers aged 50 years and older in Singapore.

Methods

A cohort Markov model compared two strategies: PUMA-based case-finding (PUMA score ≥4) followed by spirometry versus no case-finding, over a lifetime horizon from the healthcare payer perspective, with 3% annual discounting. A willingness-to-pay threshold of SGD75,000/QALY was applied.

Results

Case-finding yielded an incremental cost of SGD 2,475 and an incremental gain of 0.0339 QALYs per person (ICER: SGD 73,065/QALY). Probabilistic sensitivity analysis indicated a 71.1% probability of being cost-effective at SGD 75,000/QALY. The ICER was most sensitive to the odds ratio of exacerbation for individuals with mild and moderate COPD treated with Long-Acting Muscarinic Antagonist (LAMA), and the severity distribution of undiagnosed COPD.

Conclusions

PUMA-based targeted COPD case-finding followed by spirometry is likely to be cost-effective at Singapore’s willingness-to-pay threshold.

Introduction

Chronic obstructive pulmonary disease (COPD) is a major global public health challenge and a leading contributor to morbidity, mortality, and healthcare costs. It frequently coexists with other chronic diseases, including cardiovascular and renal disease^1,2. In Singapore, COPD is the 10^th leading cause of death and imposes a substantial economic burden^3,4.

Despite the availability of evidence-based diagnostic and treatment guidelines, COPD remains substantially underdiagnosed in Singapore. A local primary care study found that only 1.6% of at-risk individuals aged 40 years and older were aware of their COPD status⁵. One reason this diagnostic gap persists is that spirometry, although required to confirm COPD, is often difficult to implement routinely in primary care. Studies conducted in other countries have shown that spirometry is underutilised in primary care settings, and most patients are diagnosed only after presenting with moderate or severe symptoms^6,7. Although earlier detection of COPD can reduce disease burden for patients, caregivers and health systems, population-wide spirometry is impractical in most resource-constrained primary care settings.

Brief questionnaires used for targeted case-finding can help identify individuals at high risk of COPD and prioritise confirmatory spirometry in primary care^8,9. The PUMA (Study of Prevalence and Regular Practice, Diagnosis and Treatment, among General Practitioners in Populations at Risk of COPD) questionnaire is a brief, structured tool that has demonstrated clinical utility for COPD case-finding in primary care internationally¹⁰. We recently evaluated the performance of the PUMA questionnaire in Singapore primary care and found acceptable sensitivity and specificity for targeted COPD case-finding, consistent with findings from other populations^10,11.

Although questionnaire-based case-finding combined with spirometry have been shown to be cost-effective elsewhere, the cost-effectiveness of this strategy in Singapore remains unknown^12,13. Therefore, we assessed the cost-effectiveness of a two-step case-finding strategy using the PUMA questionnaire followed by confirmatory spirometry, compared with standard care in primary care settings using a simulation modelling approach from the healthcare payer perspective.

Methods

Overview

This study was a model-based cost-effectiveness analysis evaluating a targeted COPD case-finding strategy among ever-smokers aged 50 years and older in Singapore, compared with standard care. A decision-analytic Markov model was developed to simulate COPD progression, healthcare utilisation, and costs over a lifetime horizon from the healthcare payer perspective.

Study subjects

The study population was drawn from a multicentre cross-sectional study (the PUMA study) conducted in Singapore’s primary care setting to estimate COPD prevalence and evaluate the case-finding performance of the PUMA questionnaire⁵. Eligible participants were current or former smokers (defined as having smoked at least 100 cigarettes), aged 50 years and older, who had attended a primary care setting in the 12 months before recruitment and had no contraindications to spirometry. Contraindications included a history of myocardial infarction, stroke, aortic or cerebral aneurysm, eye surgery or detached retina within the preceding three months, or adverse reactions to salbutamol inhalers.

All participants completed the PUMA questionnaire comprising four objective risk-factor items (age, sex, smoking history in pack-years, prior spirometry testing) and three subjective symptom-based items (dyspnoea, chronic cough, regular sputum production). Spirometry was performed as the reference standard to evaluate the performance of the PUMA questionnaire and determine an optimal cutoff for identifying high-risk individuals. COPD was defined as a post-bronchodilator FEV1/FVC ratio < 0.7¹⁴. Based on our findings, a PUMA score of ≥5 maximised sensitivity and specificity in Singapore, whereas a PUMA score of ≥4 demonstrated higher sensitivity and negative predictive value and was recommended as the practical threshold for referral to spirometry in the primary care setting. Detailed methodology and results of the PUMA study are described elsewhere⁵.

Both the current study and the PUMA study have received ethical approval from the Institutional Research Board of Nanyang Technological University (IRB-2024-235).

Case-finding strategies

We compared two strategies. (i) Standard Care (SC): No systematic COPD case-finding was conducted in the primary care setting. Individuals were diagnosed only following symptomatic presentation or after a severe exacerbation. (ii) Annual case-finding: Targeted case-finding with PUMA questionnaire in the primary care setting, followed by spirometry for those who had a PUMA score ≥4 (Fig. 1). Case-finding was repeated annually for individuals without a COPD diagnosis until age 80. Spirometry was assumed to be conducted on-site at general practitioner (GP) clinics.

Fig. 1. Decision tree of the case-finding process (base case).

Notes: Arm 1: targeted case-finding using the PUMA questionnaire followed by spirometry. Arm 2: standard care without systematic case-finding. Under standard care, individuals may be diagnosed either following symptomatic presentation or after exacerbation-related healthcare encounters. Although these pathways exist in practice, they are not separately illustrated to preserve diagram clarity.

Model Structure

We developed a Markov state-transition model in TreeAge Pro 2025 to estimate the long-term cost-effectiveness of COPD case-finding among ever-smokers in Singapore. We used a simulation model as real-world patient-level data were not available to examine the health and economic consequences of implementing COPD case-finding. A cohort Markov model was selected as transitions are driven by disease stage rather than individual heterogeneity.

The model comprised 14 mutually exclusive health states over a lifetime (Fig. 2) grouped into disease-free, undiagnosed disease, diagnosed disease, and death (absorbing state). The simulated cohort comprised 10,000 individuals, who were classified at model entry as COPD-free or as having undiagnosed mild, moderate, severe, or very severe COPD (very severe defined as requiring hospitalisation) based on spirometry impairment, consistent with Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria. For pragmatic reasons, only severe or very severe exacerbations (labelled as severe hereafter for simplicity) were modelled.

Fig. 2. State transition diagram for the cohort Markov model.

Notes: Patients may start in the COPD-free (no COPD) state or any of the undiagnosed COPD states. They can remain in the same state, or transition to another undiagnosed COPD state of higher severity, diagnosed untreated, or diagnosed treated state. Once a patient transitions to a treated state, they may not return to the untreated state. All health states carry an annual probability of death, although transition arrows are not shown for clarity. Model cycle length is one year.

All current smokers diagnosed with COPD were assumed to be referred to a smoking cessation programme. Individuals diagnosed with COPD also received pneumococcal vaccination and annual influenza vaccination and underwent annual spirometry tests thereafter until age 80 or until progression to severe disease, whichever occurred earlier.

Once an individual developed COPD, the model allowed movement to the immediate next worse disease stage. Direct deterioration beyond the next stage within an annual cycle was not allowed because COPD was assumed to progress slowly. Transition from undiagnosed to diagnosed COPD was allowed, but not the reverse. Not all diagnosed patients received treatment, and undiagnosed health states were assumed to have the same baseline transitions to worse undiagnosed GOLD stages as diagnosed, untreated health states. Individuals in any health state were at risk of exacerbation and death. All individuals with severe exacerbations were assumed to receive a COPD diagnosis following the event.

Patients who experienced a severe exacerbation had a higher mortality risk than non-exacerbators¹⁵. The mortality risk for those who had two or more severe exacerbations over a lifetime was higher than for those who had experienced one exacerbation¹⁶.

Individuals with mild/moderate COPD were treated with long-acting muscarinic antagonists (LAMA), and those with severe COPD were treated with LAMA and long-acting beta-agonist combination therapy (LAMA + LABA). Individuals who experienced at least two severe exacerbations were escalated to triple therapy comprising LAMA + LABA and inhaled corticosteroids (LAMA + LABA + ICS). We also assumed that not all diagnosed patients would adhere to the prescribed treatment, as shown in Fig. 3.

Fig. 3.

Example of a pathway in the model.

Both disease-specific and age-adjusted background mortality were incorporated in the model. Health-state utilities and costs were assigned to each state to project quality-adjusted life years (QALYs) and costs across the model time horizon. Individuals who adhered to their treatment were modelled with a reduced probability of disease progression and exacerbation.

The model had a cycle length of one year to capture important COPD-related events such as exacerbations. The base-case time horizon was 50 years, assuming a maximum age of 100 years.

Parameter inputs

Parameters used in the model are shown in Table 1. We used the COPD prevalence and severity distribution for previously undiagnosed individuals from our previous study⁵. Among 359 participants with valid spirometry results, 45 (12.53%) had COPD.

Table 1.

Parameter inputs.

Parameter	Base case value	Source
PUMA questionnaire characteristics
Sensitivity	0.8000	5
Specificity	0.5669	5
Epidemiology parameters
Proportion of population with undiagnosed COPD	0.1253	5
Proportion of undiagnosed COPD by severity
Mild COPD	0.4222	5
Moderate COPD	0.5111	5
Severe COPD	0.0444	5
Very severe COPD	0.0222	5
Proportion of Singapore residents that are smokers	0.088	17
Proportion of Singapore residents that are ex-smokers	0.085	17
Probability of developing COPD	0.00814	19
Proportion of PUMA positive patients who go on to do office spirometry	0.4	Where local published data were unavailable, parameters were informed through consultation with two practising primary care physicians involved in COPD management in Singapore.
Probability of quitting after smoking cessation	0.088	25
Probability of having at least 1 severe/very severe exacerbation
Mild COPD	0.104	42
Moderate COPD	0.148	42
Severe COPD	0.197	42
Very severe COPD	0.244	42
Disease progression without exacerbation
GOLD 1 to 2	0.0876	13
GOLD 2 to 3	0.0362	13
GOLD 3 to 4	0.0324	13
Disease progression post severe exacerbation
GOLD 1 to 2	0.164	43
GOLD 2 to 3	0.070	43
GOLD 3 to 4	0.062	43
Mortality for patients with COPD by severity
Mild	0.0116	21
Moderate	0.0117	21
Severe	0.0347	21
Very severe	0.0876	21
HR of mortality for ever smokers	1.66	22
Mortality from severe/very severe exacerbation	0.109	15
HR of mortality from second severe/very severe exacerbation (relative to first exacerbation)	1.9	16
HR of current smokers developing COPD compared to non-smokers	1.46	18
HR of former smokers developing COPD compared to non-smokers	1.21	18
Treatment adherence rate	0.477	28
Treatment effect - LAMA
OR of experiencing an exacerbation	0.85	35
OR of progression to next GOLD stage	0.85	13
Treatment effect - LAMA + LABA
OR of experiencing an exacerbation	0.72	44
OR of progression to next GOLD stage	0.72	Assumed same as OR of experiencing an exacerbation
Treatment effect - LAMA + LABA + ICS
OR of experiencing an exacerbation	0.57	23
OR of progression to next GOLD stage	0.57	Assumed same as OR of experiencing an exacerbation
Cost
Home spirometry	SGD 100	Local fee schedules, published sources and clinician consultation
Office spirometry	SGD 60	Local fee schedules, published sources and clinician consultation
Treatment – LAMA (Annual)	SGD 720	Local fee schedules, published sources and clinician consultation
Treatment – LAMA + LABA (Annual)	SGD 720	Local fee schedules, published sources and clinician consultation
Treatment - LAMA + LABA + ICS (Annual)	SGD 960	Local fee schedules, published sources and clinician consultation
Smoking cessation	SGD 130	30
Consultation charges	SGD 50	Local fee schedules, published sources and clinician consultation
Specialist follow up (Annual)	SGD 600	Local fee schedules, published sources and clinician consultation
Influenza vaccination	SGD 70	31
Pneumococcal vaccination	SGD 125	31
Spirometry test (Annual)	SGD 60	Local fee schedules, published sources and clinician consultation
Hospitalisation due to severe exacerbation	SGD 5,533	45
Utility
Background utility (smoker)	0.84	13
Mild COPD	0.72	13
Moderate COPD	0.70	13
Severe COPD	0.68	13
Very severe COPD	0.59	13
Disutility due to severe/very severe exacerbation
Mild COPD	−0.240	13
Moderate COPD	−0.234	13
Severe COPD	−0.227	13
Very severe COPD	−0.195	13

We calculated incidence of COPD for ever smokers using the hazard ratios of current and former smokers developing COPD compared with the general population, and obtained a weighted average based on the proportions of current and former smokers from Singapore’s National Population Health Survey 2023^17–19. We obtained age-specific all-cause mortality rates from Singapore life tables²⁰. We derived a hazard ratio of mortality for ever-smokers and mortality rates for COPD patients from the literature^17,21,22.

Treatment effects from published randomised controlled trials suggest a 43% reduction in risk of exacerbations for triple therapy with no evidence of differential effect by disease severity²³. Given limited direct evidence on treatment effects on GOLD stage progression, and given that disease progression is strongly associated with exacerbation risk, we conservatively proxied progression effects using exacerbation risk reductions²⁴.

Sensitivity and specificity of PUMA questionnaire

In our previous study, we used ROC analysis to measure case-finding performance of the PUMA questionnaire. The optimal cutoff was determined using Youden’s index in conjunction with consideration of sensitivity, specificity and clinical applicability. Two PUMA score thresholds were evaluated: ≥4 and ≥5. A cutoff of ≥4 demonstrated higher sensitivity (80.0%) but lower specificity (56.7%), while a cutoff of ≥5 yielded lower sensitivity (62.2%) but higher specificity (79.3%) and a high negative predictive value (NPV) of 93.6%. Given the objective of maximising detection in initial case-finding, we selected a threshold of ≥4 for the base-case analysis.

Quit rate after smoking cessation

We assumed that all current smokers were referred to a smoking cessation programme, however, only 8.8% would quit within the first year, based on a previous study where fewer than 40% of individuals who quit received any counselling or medication²⁵. We assumed that the relapse rate was negligible, to reduce model complexity^26,27.

Treatment adherence

We assumed adherence to the inhaled therapies to be 47.7% based on evidence from Hong Kong²⁸. This value is consistent with published studies reporting adherence rates ranging from 22% to 93%.

Utility values

We assigned utility weights to reflect health-related quality of life associated with each health state in the model, with 1.0 representing perfect health and 0.0 representing death. Utility weights for current and former smokers, COPD by severity, and disutility due to exacerbations were sourced from published literature¹³. All costs and QALYs were discounted at 3% per annum, consistent with recommended economic evaluation guidelines²⁹.

Costs

We converted all costs to 2024 Singapore dollars using the healthcare component of Singapore’s Consumer Price Index and discounted at 3% each year.

Screening costs

We assumed that PUMA case-finding would be conducted during a routine primary care visit, either at a government polyclinic or a private GP clinic, with no additional cost for completing the PUMA questionnaire. Individuals with positive results were assumed to either undergo office spirometry at a separate visit, or home spirometry (a more expensive option). The costs associated with both office and home spirometry tests were estimated based on averages obtained from several practicing GPs. Based on observed uptake in the PUMA study, we assumed 40% of individuals with positive PUMA result would undergo spirometry.

Treatment costs

We based the cost for smoking cessation programmes on incentives provided under Singapore’s I Quit programme for individuals who successfully quit³⁰. We applied a 30% loading to reflect administrative costs associated with programme delivery. The costs for pneumococcal and influenza vaccinations were based on fees published by Singapore’s Ministry of Health³¹. We estimated costs of inhaled pharmacotherapies from the Healthier SG Medication List and public hospital formularies³¹. We derived the treatment cost for hospitalisation due to severe exacerbation from a recent Singapore burden of disease study³².

Analysis

We conducted the cost-effectiveness analysis from the Singapore healthcare payer perspective, comparing total costs and health outcomes between the COPD case-finding strategy and SC across a lifetime horizon. Outcomes included severe exacerbations averted, total and incremental costs, total and incremental quality-adjusted life years (QALYs), incremental cost-effectiveness ratios (ICERs), COPD-related deaths averted, and net monetary benefit (NMB).

We used a willingness-to-pay (WTP) threshold of SGD 75,000/QALY for our analysis, consistent with threshold commonly used by the Agency for Care Effectiveness (ACE), Singapore’s health technology assessment body²⁹. Most interventions with ICERs below SGD 75,000/QALY were likely to receive positive recommendations from ACE³³.

We validated the model by comparing projected outcomes for the no case-finding arm with published data from other modelling and clinical studies^9,13. We identified comparable studies through a targeted review of the published literature. Key parameters, including ICERs, intervention costs, and health outcome measures, were compared across studies that employed similar methodological frameworks, target populations, and settings. Model files are available from the corresponding author upon reasonable request.

We calculated the annual budget impact of implementing the COPD case-finding programme for a hypothetical cohort of 10,000 at-risk individuals. We used the sensitivity and specificity of PUMA score ≥4 for referral to confirmatory spirometry, the prevalence of COPD among at-risk individuals as reported previously, and the cost of annual spirometry to calculate the annual budget impact of the programme.

Scenario analysis and sensitivity analysis

Scenario analyses examined the impact of several variables, as listed below:

1. Individuals who screened positive on the PUMA questionnaire had the option to undergo office or home spirometry. 30% of these individuals were assumed to undergo home spirometry, and another 30% were assumed to undergo office spirometry.

2. One-time case-finding strategy at the age of 50.

3. Age-dependent COPD incidence rates were used instead of a constant annual probability.

4. A PUMA cutoff of ≥5 was used for referral to spirometry.

5. The odds ratio of disease progression was set to 1 for all treatment regimens, indicating no effect on the reduction of the rate of disease progression.

We performed one-way sensitivity analyses to assess the impact of individual input parameters on cost-effectiveness outcomes, summarised using a tornado diagram.

We also performed probabilistic sensitivity analysis with 10,000 Monte Carlo simulations. We modelled probabilities, proportions, and utilities using beta distributions (bounded between 0 and 1); hazard ratios and odds ratios using log-normal distributions (strictly positive and multiplicative); and cost parameters using gamma distributions (strictly positive and right-skewed), as shown in Table 2. We generated a cost-effectiveness acceptability curve to show the probability that the PUMA questionnaire with spirometry case-finding strategy would be cost-effective across a range of WTP thresholds.

Table 2.

Model parameters with ranges used for sensitivity analyses.

Parameter	Base case value	Lower bound	Upper bound	Distribution
PUMA questionnaire characteristics
Sensitivity	0.8000	0.600	1.000	beta
Specificity	0.5669	0.425	0.709	beta
Epidemiology parameters
Proportion of population with undiagnosed COPD	0.1253	0.0940	0.1566	beta
Proportion of undiagnosed COPD
Mild COPD	0.4222	0.3167	0.5278	beta
Moderate COPD	0.5111	0.3833	0.6389	beta
Severe COPD	0.0444	0.0333	0.0555	beta
Very severe COPD	0.0222	0.0167	0.0278	beta
Proportion of Singapore residents that are smokers	0.088	0.066	0.110	beta
Proportion of Singapore residents that are ex-smokers	0.085	0.064	0.106	beta
Incidence rate of COPD for the general population	0.00814	0.00611	0.0102	beta
Proportion of PUMA positive patients who go on to do office spirometry	0.4	0.200	0.600	beta
Probability of quitting after smoking cessation	0.088	0.066	0.110	beta
Probability of having at least 1 severe/very severe exacerbation
Mild COPD	0.104	0.078	0.130	beta
Moderate COPD	0.148	0.111	0.185	beta
Severe COPD	0.197	0.148	0.246	beta
Very severe COPD	0.244	0.183	0.305	beta
Disease progression without exacerbation
GOLD 1 to 2	0.0876	0.0657	0.1095	beta
GOLD 2 to 3	0.0362	0.0272	0.0453	beta
GOLD 3 to 4	0.0324	0.0243	0.0405	beta
Disease progression post severe exacerbation
GOLD 1 to 2	0.164	0.1232	0.2053	beta
GOLD 2 to 3	0.070	0.0522	0.0870	beta
GOLD 3 to 4	0.062	0.0468	0.0780	beta
Mortality for COPD
Mild	0.0116	0.0087	0.0145	beta
Moderate	0.0117	0.0088	0.0146	beta
Severe	0.0347	0.0260	0.0434	beta
Very severe	0.0876	0.0657	0.1095	beta
HR of mortality for ever smokers	1.66	1.49	1.91	log normal
Mortality from severe/very severe exacerbation	0.109	0.105	0.112	beta
HR of mortality from second severe/very severe exacerbation (relative to first exacerbation)	1.9	1.8	2.0	log normal
HR of current smokers developing COPD compared to non-smokers	1.46	1.10	1.83	log normal
HR of former smokers developing COPD compared to non-smokers	1.21	0.91	1.51	log normal
Treatment adherence rate	0.477	0.4	0.581	beta
Treatment effect - LAMA
OR of experiencing an exacerbation	0.85	0.72	1	log normal
OR of progression to next GOLD stage	0.85	0.72	1	log normal
Treatment effect - LAMA + LABA
OR of experiencing an exacerbation	0.72	0.59	0.88	log normal
OR of progression to next GOLD stage	0.72	0.59	0.88	log normal
Treatment effect - LAMA + LABA + ICS (post exacerbation)
OR of experiencing an exacerbation	0.57	0.50	0.64	log normal
OR of progression to next GOLD stage	0.57	0.50	0.64	log normal
Cost
Office spirometry	SGD 60	SGD 45	SGD 75	gamma
Treatment - LAMA (annual)	SGD 720	SGD 540	SGD 900	gamma
Treatment - LAMA + LABA (annual)	SGD 720	SGD 540	SGD 900	gamma
Treatment - LAMA + LABA + ICS (annual)	SGD 960	SGD 720	SGD 1200	gamma
Smoking cessation	SGD 130	SGD 98	SGD 163	gamma
Consultation charges	SGD 50	SGD 38	SGD 63	gamma
Annual specialist follow up	SGD 600	SGD 450	SGD 750	gamma
Influenza vaccination	SGD 70	SGD 53	SGD 88	gamma
Pneumococcal vaccination	SGD 125	SGD 100	SGD 250	gamma
Yearly spirometry test	SGD 60	SGD 45	SGD 75	gamma
Hospitalisation due to severe exacerbation	SGD 5,533	SGD 4,150	SGD 6,916	gamma
Utility
Background utility (smoker)	0.84	0.630	1.000	beta
Mild COPD	0.72	0.540	0.900	beta
Moderate COPD	0.70	0.526	0.877	beta
Severe COPD	0.68	0.510	0.850	beta
Very severe COPD	0.59	0.439	0.732	beta
Disutility due to severe/very severe exacerbation
Mild COPD	−0.240	−0.300	−0.180	beta
Moderate COPD	−0.234	−0.292	−0.175	beta
Severe COPD	−0.227	−0.283	−0.170	beta
Very severe COPD	−0.195	−0.244	−0.146	beta

We also estimated the case-finding-related budget impact for a hypothetical cohort of 10,000 eligible ever-smokers by applying the observed COPD prevalence, PUMA sensitivity and specificity, assumed spirometry uptake among PUMA-positive individuals, and the unit cost of office spirometry. The cost per case detected was calculated by dividing total spirometry costs by the projected number of newly detected COPD cases.

This study followed the Consolidated Health Economic Evaluation Reporting Standards reporting guidelines (Table 3)³⁴.

Table 3.

CHEERS 2022 Checklist.

Topic	No.	Item	Location where item is reported
Title
	1	Identify the study as an economic evaluation and specify the interventions being compared.	Title, Page 1
Abstract
	2	Provide a structured summary that highlights context, key methods, results, and alternative analyses.	Abstract, Page 2
Introduction
Background and objectives	3	Give the context for the study, the study question, and its practical relevance for decision making in policy or practice.	Introduction, Line 5
Methods
Health economic analysis plan	4	Indicate whether a health economic analysis plan was developed and where available.	Not reported
Study population	5	Describe characteristics of the study population (such as age range, demographics, socioeconomic, or clinical characteristics).	Methods - Study subjects
Setting and location	6	Provide relevant contextual information that may influence findings.	Methods - Study subjects
Comparators	7	Describe the interventions or strategies being compared and why chosen.	Methods – Case-finding strategies
Perspective	8	State the perspective(s) adopted by the study and why chosen.	Methods - Analysis
Time horizon	9	State the time horizon for the study and why appropriate.	Methods - Model Structure, Eighth Paragraph
Discount rate	10	Report the discount rate(s) and reason chosen.	Methods - Utilities Values and Costs
Selection of outcomes	11	Describe what outcomes were used as the measure(s) of benefit(s) and harm(s).	Methods - Analysis
Measurement of outcomes	12	Describe how outcomes used to capture benefit(s) and harm(s) were measured.	Methods - Analysis
Valuation of outcomes	13	Describe the population and methods used to measure and value outcomes.	Methods - Analysis
Measurement and valuation of resources and costs	14	Describe how costs were valued.	Methods - Costs
Currency, price date, and conversion	15	Report the dates of the estimated resource quantities and unit costs, plus the currency and year of conversion.	Methods - Costs
Rationale and description of model	16	If modelling is used, describe in detail and why used. Report if the model is publicly available and where it can be accessed.	Methods - Model Structure
Analytics and assumptions	17	Describe any methods for analysing or statistically transforming data, any extrapolation methods, and approaches for validating any model used.	Methods - Analysis
Characterising heterogeneity	18	Describe any methods used for estimating how the results of the study vary for subgroups.	Not reported
Characterising distributional effects	19	Describe how impacts are distributed across different individuals or adjustments made to reflect priority populations.	Methods - Analysis
Characterising uncertainty	20	Describe methods to characterise any sources of uncertainty in the analysis.	Methods - Analysis
Approach to engagement with patients and others affected by the study	21	Describe any approaches to engage patients or service recipients, the general public, communities, or stakeholders (such as clinicians or payers) in the design of the study.	Not reported
Results
Study parameters	22	Report all analytic inputs (such as values, ranges, references) including uncertainty or distributional assumptions.	Tables 1 and 2
Summary of main results	23	Report the mean values for the main categories of costs and outcomes of interest and summarise them in the most appropriate overall measure.	Results, first paragraph
Effect of uncertainty	24	Describe how uncertainty about analytic judgments, inputs, or projections affect findings. Report the effect of choice of discount rate and time horizon, if applicable.	Results - Scenario Analysis and Sensitivity Analysis
Effect of engagement with patients and others affected by the study	25	Report on any difference patient/service recipient, general public, community, or stakeholder involvement made to the approach or findings of the study	Not reported
Discussion
Study findings, limitations, generalisability, and current knowledge	26	Report key findings, limitations, ethical or equity considerations not captured, and how these could affect patients, policy, or practice.	Discussion
Other relevant information
Source of funding	27	Describe how the study was funded and any role of the funder in the identification, design, conduct, and reporting of the analysis	Declarations
Conflicts of interest	28	Report authors conflicts of interest according to journal or International Committee of Medical Journal Editors requirements.	Declarations

Results

In the base-case analysis for 50-year-old ever-smokers (Table 4a), annual COPD case-finding using the PUMA questionnaire followed by spirometry in primary care was cost-effective compared with no case-finding, with an ICER of SGD 73,065/QALY. The case-finding strategy was associated with an incremental cost of SGD 2,475 and an incremental gain of 0.0339 QALYs per person over a lifetime horizon. At the population level, the case-finding strategy was projected to avert 251 severe exacerbations (a 2.87% reduction) per 10,000 individuals compared with standard care.

Table 4.

a. Base-case analysis results over a lifetime horizon, per person.

a
Strategy	Cost (SGD)	Incremental cost (SGD)	QALYs	Incremental QALYs	ICER	Net monetary benefit (NMB) (SGD)	Severe/very severe exacerbations	Severe/very severe exacerbations averted	COPD related deaths	COPD related deaths averted
No case-finding	4,854	–	14.9450	–	–	1,116,019	0.8717	–	0.3592	–
Case-finding with PUMA questionnaire followed by spirometry	7,329	2,475	14.9788	0.0339	SGD 73,065/QALY	1,116,084	0.8466	2.87%	0.3592	0.00%

b. Results from scenario analysis compared to base case
Scenario	Incremental Cost	Incremental Effectiveness	ICER	Severe/very severe exacerbations averted
1) Base-case analysis	SGD 2,475	0.0339	SGD 73,065/QALY	2.87%
2) 30% doing office spirometry and 30% doing home spirometry	SGD 2,846	0.0422	SGD 67,313/QALY	3.37%
3) One-time targeted case finding at age 50	SGD 602	0.0187	SGD 32,132/QALY	0.95%
4) Using age-dependent incidence rate for COPD	SGD 2,016	0.0247	SGD 81,526/QALY	2.44%
5) PUMA score of ≥5 instead of ≥4 for referral to spirometry	SGD 2,248	0.0292	SGD 76,854/QALY	2.58%
6) Treatment has no effect in reduction of rate of disease progression (OR of disease progression set to 1 for all regime)	SGD 2,482	0.0323	SGD 76,749/QALY	2.76%

Scenario analysis

Scenario analysis results are shown in Table 4b.

When home spirometry was offered in addition to office spirometry following a positive result on PUMA questionnaire, the case-finding strategy remained cost-effective with an ICER of SGD 67,313/QALY. This scenario was associated with a higher incremental cost of SGD 2,846 and a gain of 0.0422 QALYs per person, and was projected to reduce severe exacerbations by 3.37%.

A one-time case-finding strategy at age 50 yielded a substantially lower ICER of SGD 32,132/QALY, indicating greater cost-effectiveness relative to repeated annual case-finding. However, it also yielded the smallest incremental benefit (0.0187 QALYs) and the lowest proportion of exacerbations averted (0.95%). When age-dependent COPD incidence rates were used instead of a constant annual probability, the ICER increased above the WTP threshold, to SGD 81,526/QALY. This scenario was associated with a marginally lower incremental cost of SGD 2,016 and lower incremental gain of 0.0247 QALYs compared with the base case.

Using a PUMA score of ≥5 instead of ≥4 for referral to spirometry resulted in an incremental cost of SGD 2,248 and incremental gain of 0.0292 QALYs. The ICER increased to SGD 76,854/QALY, which is above the WTP threshold. Lastly, in a scenario where treatment had no effect in reduction of rate of disease progression, the ICER increased to SGD 76,749/QALY, with higher incremental cost (SGD 2,482) and lower incremental effectiveness (0.0323) compared with the base case.

Sensitivity analysis

In the deterministic sensitivity analysis (Fig. 4), the ICER was most sensitive to: (i) the odds ratio of exacerbation for individuals with mild and moderate COPD treated with LAMA, (ii) the probability of a previously undiagnosed COPD patients having moderate COPD, and (iii) the probability of a previously undiagnosed COPD patients having mild COPD. The case-finding strategy remained cost-effective at the SGD 75,000/QALY threshold provided the odds ratio of exacerbation with LAMA did not exceed 0.855. The other parameters that influenced the cost-effectiveness of the case-finding strategy are the probability of severe exacerbation for individuals with mild and moderate COPD, cost of LAMA treatment, physician consultation costs, and the odds ratio of disease progression for individuals with mild and moderate COPD treated with LAMA. For the case-finding strategy to be cost-effective, at least 36.85% of individuals with a PUMA score of ≥4 needed to undergo confirmatory spirometry. The number needed to screen to detect one COPD case is 25.

Fig. 4. Tornado diagram from one-way deterministic sensitivity analysis.

Notes: Variation in the odds ratio of experiencing a severe or very severe exacerbation with LAMA had the greatest influence on the base-case ICER, with the ICER exceeding the WTP threshold when the odds ratio increased above 0.855. ICER, incremental cost-effectiveness ratio; SGD, Singapore dollars; WTP, willingness-to-pay.

Probabilistic sensitivity analysis confirmed the robustness of the base-case result. Figures 5 and 6 show that at the WTP threshold of SGD 75,000/QALY, the PUMA and spirometry case-finding strategy had a 71.1% probability of being cost-effective compared with no case-finding.

Fig. 5. Cost-effectiveness acceptability curve from probabilistic sensitivity analysis (10,000 Monte Carlo simulations).

Note: The curve shows the probability that PUMA-based targeted case-finding is cost-effective compared with no case-finding across a range of willingness-to-pay thresholds. QALY, quality-adjusted life year.

Fig. 6. Cost-effectiveness plane for the comparison of targeted PUMA-based case-finding with no case-finding, based on 10,000 Monte Carlo simulations.

Notes: Each point represents one simulation. Points in the northeast quadrant indicate that case-finding is both more costly and more effective than no case-finding.

Budget impact

We estimated that implementing the PUMA-led case-finding strategy across primary care networks in Singapore generate an annual case-finding related budget impact of SGD 114,977 per 10,000 eligible ever-smokers screened. This corresponds to approximately SGD 287 per COPD case detected, based on projected spirometry use and newly identified cases.

Discussion

Principal findings

We constructed a Markov model to evaluate the cost-effectiveness of a targeted COPD case-finding strategy using the PUMA questionnaire, followed by confirmatory spirometry for individuals with a positive PUMA result, among ever-smokers in Singapore. Our results suggest that over a lifetime horizon, COPD case-finding was cost-effective compared with no case-finding, with an incremental cost of SGD 2,475 and a gain of 0.0339 QALYs per person, corresponding to an ICER of SGD 73,065 per QALY. Targeted case-finding was also projected to reduce severe exacerbations at the population level.

Cost-effectiveness was most sensitive to the odds ratio of exacerbation for individuals with mild and moderate COPD treated with LAMA, and the probability that a previously undiagnosed patient had moderate or mild COPD.

Interpretation in context

Our findings are consistent with international evidence suggesting that targeted COPD case-finding among high-risk individuals may be a cost-effective strategy. Prior economic evaluations, predominantly from Western healthcare systems, have reported mixed results regarding the relative cost-effectiveness of questionnaire-only, spirometry-only, or combined case-finding approaches. A recent meta-analysis suggested that combined strategies may be less cost-effective than single-modality case-finding when compared with no case-finding⁹; however, these findings were characterised by substantial heterogeneity and limited applicability to Asian health systems.

In practice, questionnaire-based case-finding offers a pragmatic first step in primary care by identifying individuals most likely to benefit from confirmatory spirometry. In settings where spirometry capacity is limited, this two-step approach may improve efficiency by prioritising diagnostic resources while maintaining adequate case detection rates. Our study found that a two-step approach implemented during routine primary care visits is a cost-effective method for early COPD case-finding.

Policy implications

The value of early COPD detection in our model is largely driven by the utility benefit received from symptom relief in newly diagnosed patients, and by reduced disease progression with treatment. Evidence from large clinical trials shows that inhaled therapies improve symptoms and quality of life across COPD severities, supporting the clinical plausibility of these benefits³⁵.

From a policy perspective, our results provide quantitative support for considering structured COPD case-finding within primary care-led disease management in Singapore. The PUMA questionnaire is brief (requires only 2–3 min to complete) and can be self-administered or supported with minimal assistance from trained healthcare staff during patient registration, thereby reducing the burden on physicians during consultations. This approach aligns with existing workflows in public polyclinics and private GP practices, where brief patient assessments and health risk evaluations are routinely integrated into chronic disease management. Incorporating a COPD case-finding questionnaire into these processes would require minimal operational adjustments while enhancing case-finding efforts.

For practices without on-site spirometry capabilities, referral pathways to polyclinic-based respiratory clinics or hospital pulmonary function laboratories would be required. In Singapore’s integrated cluster-based healthcare system supported by electronic medical records, such referrals are already routine for other diagnostic procedures, and the additional volume from PUMA-positive patients is likely to be manageable. If home spirometry becomes widely adopted and supported by robust clinical evidence, both private and public healthcare institutions could offer it as an additional testing option. Insights from our unpublished qualitative study suggest that individuals place high value on the convenience and accessibility of home-based testing, highlighting its potential to improve spirometry uptake and patient engagement.

Reimbursement considerations will be central to implementation. Questionnaire-based case-finding could be incorporated into Singapore’s subsidised health screening protocols without additional patient cost. For private GPs, the Chronic Disease Management Programme (CDMP) provides guidelines for screening, while the Community Health Assistance Scheme (CHAS) provides means-tested subsidies for chronic disease management^36,37. Physician remuneration for screening activities and care coordination could potentially be considered within the CHAS subsidy structures to support implementation.

Pilot implementation in selected polyclinics and CHAS-partner GPs would provide valuable insights into real-world uptake among eligible patients, identify workflow barriers, assess the feasibility of spirometry referral processes, and refine protocols before broader rollout. Such pilot studies could also provide important data on patient and provider acceptability, which would strengthen the evidence base for broader programme implementation.

Limitations

Our study has several limitations. First, treatment effects, disease progression rates, adherence rates, and utilities were derived from international studies due to the lack of Singapore‑specific data, which may not fully reflect the experience of Singapore’s multi‑ethnic population and healthcare‑seeking behaviour. Where possible, data from comparable high‑income or Asian settings were used, though local epidemiological and registry data would improve future modelling accuracy. Second, we adopted a healthcare system perspective and did not include broader societal costs such as productivity losses, which are difficult to estimate given the slow progression of lung function impairment and contextual variation in labour and caregiving norms. Third, we did not model several cost components, including ventilatory support, long‑term oxygen therapy, pulmonary rehabilitation, and selected COPD medications. Fourth, we did not explicitly model costs arising from multimorbidity, although age‑specific background mortality partially captures competing risks of non‑COPD death; omission of these costs may underestimate the incremental cost and ICER of COPD case‑finding^38,39. Fifth, smoking cessation was modelled using real‑world effectiveness rather than full uptake, and relapse was not included, which may underestimate variability in outcomes. Sixth, use of a lower PUMA threshold (≥4) may increase false‑positive referrals and associated spirometry costs. In addition, COPD definition using a fixed FEV1/FVC ratio, while consistent with current GOLD guidelines, may misclassify older adults compared with the lower‑limit‑of‑normal criteria, potentially affecting prevalence and cost‑effectiveness estimates; however, both choices support earlier diagnosis in line with national primary care strategy^40,41. Seventh, the analysis focused on ever-smokers; while non‑tobacco environmental exposures contribute to COPD burden globally, this reflects the Singapore context and existing evidence for smoker‑targeted case‑finding. Lastly, we did not assess operational barriers to real‑world implementation. The assumed 40% spirometry completion rate among PUMA‑positive individuals may be affected by physician time constraints, spirometry availability and follow‑up coordination, which future real‑world or survey‑based studies could better characterise^38–41.

Conclusion

Targeted COPD case-finding using the PUMA questionnaire followed by spirometry among ever-smokers in Singapore is likely to be cost-effective at Singapore’s willingness-to-pay threshold of SGD 75,000/QALY. By facilitating earlier identification of individuals with undiagnosed disease, case-finding allows timely initiation of appropriate treatment with downstream benefits in symptom control, and reduced exacerbations. Moreover, the limitations in the study may lead to conservative estimates of costs and benefits; therefore, real-world implementation may demonstrate even greater value than projected in this analysis. Overall, these findings suggest that COPD case-finding represents a high-value strategy for improving long-term health outcomes among at-risk individuals. Future research should evaluate the operational feasibility of this strategy in primary care and its real-world impact on patient outcomes and healthcare utilisation.

Author contributions

LES and JMN obtained funding to the study. AS, IO, LES, and VMQ conceived the study. AS, GPT, IO, KT, LES, and VMQ participated in data collection. IO conducted data analysis and wrote the first draft of the manuscript, with inputs from AS, LES, and VMQ. All authors critically reviewed the intellectual content of the manuscript and approved the final version for publication.

Funding

This study was supported by AstraZeneca Singapore. The funder had no role in study design, data collection, analysis, interpretation, or preparation of the manuscript.

Data Availability

The data used to support the findings of this study are cited within the manuscript.

Competing interests

The authors declare no competing interests.