Table 3.

Methodological aspects of the articles: variables collected to estimate direct healthcare costs, direct non-medical costs, indirect costs and intangible costs, and limitations

Article	Variables collected	Limitations
Foo J et al. [6]	Sociodemographic characteristics of the patient: sex, age, years working, Patient characteristics related to the disease: COPD severity, mMrc, CAT and comorbidities. Direct healthcare costs: moderate and severe exacerbations, contact with health professionals, COPD medications, and flu vaccination in the past 12 months. Indirect costs: lost productivity and associated annual gain based on the above loss.	The different local unit costs of health services in each country had a limited impact on the variability between different countries. There is also wide variability among COPD patients surveyed in various countries. Direct costs are likely to depend on health systems and patient access to health care. Developed countries where access to care is often free tend to have higher costs. Other factors influencing variations in the data could be related to cultural differences and health practices.
Wacker M. E et al. [7]	Sociodemographic characteristics of the patient: age, sex, smoking and educational level. Patient characteristics related to the disease: FEV1, BMI, concomitant pathologies (MD, cerebrovascular accident (CVA), acute myocardial infarction (AMI), cancer and arthritis). Direct healthcare costs: visits to the doctor (last 3 months) and 13 specialties and prescribed drugs. Indirect costs: work absenteeism (last 12 months) and early retirement.	Regional restriction of the control group, possible selection bias that could have resulted in underestimation or overestimation of excess COPD costs, limiting the potential for external validation. Recall bias in interviews to record direct healthcare costs. Additional health services such as nursing or premature death costs were not considered. Cohort had to be free of moderate-severe exacerbations in the last 4 weeks before participation.
Chen W et al. [8]	Sociodemographic characteristics of the patient: age and sex. Patient characteristics related to the disease: comorbidities. Direct healthcare costs: inpatient costs, outpatient costs and prescribed drugs.	The impact of COPD severity on comorbidity costs was not assessed because information on pulmonary function measures in health data was missing. Lack of information from missing diagnosis registries. The study did not adjust cost estimates using survival for each year of follow-up.
Tachkov K et al. [9]	Sociodemographic characteristics: of the patient age and sex. Patient characteristics related to the disease: severity of COPD. Indirect costs: absenteeism (loss of productivity due to sick leave) and presenteeism (loss of productivity due to reduced work capacity), early retirement or permanent reduction in work capacity. Intangible costs: DALY.	Short follow-up period (1 year). Use of GDP per capita. In the calculation of DALY, YLL was assumed to be 0, because no patients died during the study period. This linked DALY directly to years of disability.
Wacker M. E et al. [10]	Sociodemographic characteristics of the patient: sex, age, smoking and educational level. Patient characteristics related to the disease: GOLD stage, spirometry variables, BMI, comorbidities, and symptoms (cough, expectoration, dyspnea, or no symptoms). Direct healthcare costs: outpatient costs, hospitalizations, and medication costs. Indirect costs: absence from work and early retirement.	The central data are based on self-reported information by study participants (recall bias). Consequently, the economic burden of COPD could be underestimated. History of exacerbations was not included as a possible cost factor. The design does not allow investigating whether COPD-related costs are increased by comorbid conditions.
Quang Vo T et al. [11]	Sociodemographic characteristics of the patient: age, sex and health insurance status. Direct healthcare costs: drugs and medical examinations, laboratory and functional tests and hospitalization costs.	The retrospective design may have led to loss of data. The data used were from a public hospital database; therefore, the results are primarily applicable to public care settings. The accuracy of diagnosis and management was sensitive to the diagnostic criteria used by the reporting physicians.
Lakiang T et al. [12]	Sociodemographic characteristics of the patient: age, marital status, employment status, number of family members, health insurance, monthly income, and educational level. Direct healthcare costs: visits by healthcare professionals, prescribed drugs, and hospitalizations. Direct non-medical costs: medical transportation and meals. Indirect costs: absolute absenteeism, relative absenteeism, work hours lost, absolute absenteeism and relative absenteeism.	The total sample size was not achieved, as many of the patients who were identified as subjects were retired or had not been working for more than one month, so they were excluded from the study. The sample size was compromised regarding the productivity and absenteeism study. The indirect cost was not reported in monetary terms.
Lisspers K et al. [13]	Sociodemographic characteristics of the patient: age and sex. Patient characteristics related to the disease: comorbidities and pharmacological treatment. Direct healthcare costs: COPD drugs, all drugs, nights hospitalized for COPD, nights hospitalized for other causes, hospital visits for COPD, hospital visits for other causes, doctor visits, visits by other health professionals. Indirect costs: loss of income in COPD patients.	The retrospective design of the study means that it may be subject to bias and confounding. The results for absolute cost have limited generalizability outside Sweden, as cost structures, prices, and supply differ between countries. Some sources of costs (personal care and others attributable to family caregivers) may have been overlooked The reference population was identified from patients without COPD who were required to attend the study’s Primary Care centers rather than healthy individuals from the general population.
Merino M et al. [14]	Sociodemographic characteristics of the patient: sex, age, smoking, educational level and employment category. Patient characteristics related to the disease: spirometry parameters, disease severity, exacerbations in the last 12 months, number and description of comorbidities and BMI. Sociodemographic characteristics of the caregiver: gender, age and relationship to the patient. Direct healthcare costs: medical tests (blood tests, pre-post spirometry, alpha-I anti-trypsin, X-rays, CAT scan and blood gas), pneumologist visits, Emergency Department visits, hospital admissions and prescribed drugs. Direct non-medical costs: permanent stay in a home or residential center, attendance at a day care center, professional home care, informal care, use of medical transportation, home oxygen, and number of informal caregivers. Indirect costs: labor productivity losses.	The severity level of COPD was missing in most (58.3%) of the participants’ medical records. Resource use corresponds to those associated with COPD in the last 12 months, but it is possible that a proportion is not strictly associated with COPD but with its related comorbidities. Recall bias in relation to work hours lost in the last 12 months of working patients (7.9%).
Patel JG et al. [15]	Sociodemographic characteristics related to the patient: age, sex, geographic region, type of plan, type of eligibility (leave + short-term disability, leave only or short-term disability only) and type of industry. Patient characteristics related to the disease: comorbidities. Direct healthcare costs: Assessments, hospitalizations, doctor’s visits, visits to other professionals and prescribed drugs. Indirect costs: absenteeism and short-term disability.	The small sample size limits the interpretation of the impact of exacerbation frequency on absenteeism in the current study. The models were adjusted for comorbidities included in the CCI, rather than a complete list of possible comorbidities, which means that some conditions may not have been considered. The indirect cost estimate only considers short-term disability and not absenteeism, presenteeism, or long-term disability. This may underestimate the costs to an employer. In addition, the short-term disability data were extracted from the MarketScan database whose purpose and consistency have not been confirmed. The exacerbator status assessment used medical claims to define exacerbator groups in the same period as the outcomes. This potentially confounded the impact of exacerbations on direct costs.
Stafyla E et al. [16]	Sociodemographic characteristics of the patient: age, sex, occupational exposure, employment status and smoking. Patient characteristics related to the disease: BMI, years with COPD, spirometry parameters, respiratory symptoms, comorbidities, GOLD stage and GOLD group. Direct healthcare costs: pharmacological treatment, medical visits, respiratory function tests, vaccinations, and laboratory tests. Direct non-medical costs: transport and home oxygen.	This study was designed to estimate the direct cost of COPD during the maintenance phase and not to assess the cost of exacerbations, which contributes a significant proportion of total cost in many studies. Due to the absence of related data, indirect costs could not be calculated. A large proportion of patients visited NHS units free of charge, so the cost of medical visits was underestimated. A final limitation could be the small population size.
Zhu B et al. [17]	The number of studies included in the review was 47 with an average quality assessment score of 7.70 out of 10. Direct healthcare costs, direct non-medical care costs, indirect costs and intangible costs were evaluated: 13 studies evaluated quality of life.	N/A*
Changhwan Kim M. D et al. [18]	Direct healthcare costs: Outpatient and inpatient care, emergency room visits and prescribed drugs. (Formal + informal) Direct non-medical costs: transportation, home care and nursing costs. Indirect costs: premature mortality and productivity losses.	Sample size. Patient survey data were used to estimate direct nonmedical and indirect costs for the first time in Korea. The survey data were entirely patient-reported (recall bias). An attempt was made to recruit patients from primary clinics to tertiary hospitals. Most of the sample were tertiary hospital patients, so direct medical costs per patient were higher in tertiary care hospitals than in primary and secondary care facilities. Overall costs may be overestimated.
Rehman A et al. [19]	The number of studies included in the review was 40. Direct healthcare costs were evaluated.	N/A
Viinanen A et al. [20]	Sociodemographic characteristics of the patient: age and sex. Patient characteristics related to the disease: BMI, spirometry parameters, comorbidities, years of disease follow-up and deaths due to COPD or other causes. Direct healthcare costs: medical visits (outpatient + emergency), hospitalization days, medical procedures and surgeries, and laboratory.	It is likely that costs associated with medical care were underestimated, since all hospital days were valued equally in the analyses, however, severe exacerbations of COPD are treated with costly treatment that could not be assessed in this study. Indirect costs were not considered. Data were taken from specialty care hospital records, patients with severe COPD were likely referred or their exacerbations treated in specialty care whereas patients with non-severe COPD would have had a reason to go to specialty care (diagnostic purposes), therefore patients with non-severe COPD in this study likely had higher associated health costs than non-severe patients treated at the Primary Care level. Risk of data collection from unconscious patients that may affect population size and other outcomes.
Woo L et al. [21]	The number of studies included in the review was 10. Direct and indirect costs were evaluated.	N/A
Iheanacho I et al. [22]	The number of studies included in the review was 73. Direct healthcare costs and direct non-medical costs (22 articles) and indirect costs (1 article) were evaluated.	N/A
Rehman A et al. [23]	The number of studies included in the review was 19. Direct and indirect costs were evaluated.	N/A

Source.Own elaboration

*N/A ➔ Not applicable

*COPD: Chronic obstructive pulmonary disease

*mMrc: Modified scale of Medical Research Council

*CAT: COPD Assessment Test

*FEV1: Peak expiratory volume in the first second

*BMI: Body mass index

*MD: Mellitus diabetes

*CVA: Cerebrovascular accident

*AMI: Acute myocardial infarction

*DALY: Disability adjusted life years

*GDP: Gross domestic product

*GOLD: Global initiative of chronic obstructive lung disease

*NHS: National health service