Direct Health Care Costs Associated With a Multicomponent COPD Exacerbation Intervention

Abstract

BACKGROUND:

Health care costs attributed to COPD have been estimated at $4.7 trillion globally in the next 30 years. With the global burden of COPD rising, identification of interventions that might lead to health care cost savings is an imperative. Although many studies report the effect of COPD self-management interventions on subject outcomes and health care utilization, few data describe their effect on health care costs.

METHODS:

Using data linkage and established case-costing methods with provincial Canadian health databases, we established public health care costs (acute and community) for 12 months following randomization for the 462 participants enrolled in our randomized controlled trial of the Program of Integrated Care for Patients with COPD and Multiple Comorbidities.

RESULTS:

Total median (interquartile range) in-hospital costs in the 12 months follow-up for all (intervention and control) 462 trial participants were CAD $4,769 ($417–16,834) (equivalent to US $3,566 [$312–12,588]). Total costs incurred in the community were higher at CAD $8,011 ($4,749–13,831) (equivalent to US $5,990 [$3,551–10,342]). Controlling for sex, income quintile, Johns Hopkins Aggregated Diagnosis Groups score, and living in an urban locality, we found lower community health care costs but no differences in acute care costs for participants receiving our multicomponent COPD exacerbation prevention management intervention compared to usual care.

CONCLUSIONS:

Controlling for important confounders, we found lower public community health care costs but no difference in acute health care costs with our multicomponent COPD exacerbation prevention management intervention compared to usual care. Community health care costs were almost double those incurred compared to acute health care costs. Given this finding, although most COPD exacerbation management interventions generally focus on reducing the use of acute care, interventions that enable health care cost savings in the community require further exploration.

Introduction

As a chronic disease with rising prevalence and a long-term trajectory,¹ COPD represents an escalating public health problem with significant economic implications to health care systems worldwide.^2,3 COPD is a multifactorial systemic disease comprising muscle wasting, cardiovascular disease, anxiety and depression, decreased fat-free mass, metabolic syndrome, osteoporosis, and chronic infections.⁴ In general, patients with COPD experience numerous comorbidities⁵ that influence COPD prognosis and are affected by COPD exacerbations.

Previously, we reported on a randomized controlled trial evaluating the effectiveness of a multicomponent case manager–led COPD exacerbation prevention/management intervention designed to reduce acute health care utilization.⁶ This multicomponent, case manager–led intervention comprised a 40-min standardized education session based on the Living Well With COPD patient’s learning tool,⁶ action plans for COPD, and other comorbidity exacerbation management as well as case manager–initiated telephone consultations (weekly for 12 weeks and monthly for subsequent 9 months). The control arm comprised usual care. In this trial, when we examined health care utilization, we found reduced risk of emergency department (ED) visits and hospital admission only in those intervention participants requiring these acute health care services. There was no difference when considering all trial participants, with our data highly skewed due to many participants not requiring these services during the 12 months participants were enrolled in the trial. Although our trial was not powered for mortality, risk of death in the intervention group was nearly half that of control participants.

A subsequent randomized controlled trial that similarly combined action plans for both COPD and comorbidities⁷ found no effect on COPD exacerbation frequency but fewer subjects in the intervention with ≥ one respiratory-related hospitalization. Lower probability of respiratory-related hospital admission was also reported in an earlier Cochrane review as well as improved health-related quality of life.⁸ Despite these indications of benefit, little data describe the effect of COPD self-management interventions on subsequent health care utilization costs. This is an important gap in our understanding of programs that are essentially designed to divert costs away from expensive acute health care services through prevention of COPD exacerbation requiring hospitalization.⁹

Therefore, we conducted an a priori planned data linkage to health administrative databases for the province of Ontario, Canada, with the aim of determining the effect of our multicomponent intervention compared to usual care on health care costs at 12 months after randomization. In Canada where we conducted the trial, costs of all medically necessary care are covered for all residents by universal public health insurance funded through general taxation.

QUICK LOOK.

Current knowledge

Health care costs of COPD are excessive, and the global burden of COPD continues to rise due to population aging, worsening air quality, and increasing total numbers of smokers worldwide. Identification of interventions that reduce acute health care utilization producing cost savings is imperative. Although many studies report the effect of COPD self-management interventions on patient outcomes and health care utilization, few data describe their effect on health care costs.

What this paper contributes to our knowledge

Using established patient-level costing methodology, we found lower public community health care costs but no difference in acute health care costs associated with our multicomponent COPD exacerbation prevention management intervention compared to usual care when controlling for sex, income quintile, total Aggregated Diagnosis Groups score of ≥ 7, and living in an urban versus rural locality. Community health care costs were almost double those incurred via accessing acute health care services.

Methods

We used unique encoded identifiers to link to the health administrative databases held at ICES (formerly the Institute for Clinical Evaluative Sciences), Ontario, Canada. These databases hold all data on all patient health care encounters funded by the public health care system in the province of Ontario. ICES is an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze health care data for health system evaluation and improvement. We used the health administrative databases to identify (1) hospitalizations, ICU admissions, and in-hospital death from the Discharge Abstract Database; (2) ED presentations and same-day surgery from the National Ambulatory Care Reporting System; (3) physician billings including procedures from the Ontario Health Insurance Plan physician claims database; (4) prescription costs incurred in the community for individuals eligible for the Ontario Drug Benefit program; (5) in-patient rehabilitation from the National Rehabilitation Reporting System; (6) facility-based continuing (residential) and long-term care services from the Continuing Care Reporting System and the Client Profile Database; (7) in-patient mental health stays from the Ontario Mental Health Reporting System; (8) home care services from the Home Care Database; and (9) death outside of hospital from the Registered Persons Database.

To establish costs, we first confirmed patterns of community and in-hospital health care utilization recorded in the ICES databases according to study allocation. We then calculated the costs of this health care utilization in Canadian dollars using established patient-level costing methodology.¹⁰ We calculated total in-hospital and community (ie, all public health care costs incurred while not a hospital in-patient) costs. In-hospital costs included those incurred through hospital admissions, ED visits, and same-day surgeries. Community costs included those related to general practitioner and specialist ambulatory clinic visits, long-term care, rehabilitation, home care, laboratory and non-physician billings, and out-of-hospital publicly funded prescription costs for those > 65 y of age on receiving benefits covered by the Ontario Drug Benefit program.¹¹ Although costs of the multicomponent intervention were not specifically counted, most costs (eg, clinic visits) were captured via these methods.

The use of data held at ICES was authorized under section 45 of Ontario’s Personal Health Information Protection Act according to privacy regulations of ICES. Recruitment of participants to form the study cohort was approved by the research ethics boards of the Michael Garron Hospital (Toronto, Ontario, Canada) (510-1205-Res-018) and Southlake Regional Health Center (Southlake, Ontario, Canada) (number 0033 1213). Written informed consent for data linkage was obtained from all trial participants.

We present health care costs as medians and interquartile ranges (IQRs) due to skewed data distribution. We examined mortality after randomization using Kaplan-Meier curves. We performed unadjusted comparisons of costs according to study arm allocation using Mann-Whitney U tests. To examine variables within the ICES databases selected a priori as associated with health care costs, we created negative binomial regression models for community and in-hospital costs using the offset option to account for difference in follow-up. Analyses were conducted by an experienced ICES analyst using SAS EG version 7.15 (SAS Institute, Cary, North Carolina). All analyses were 2 tailed, P value of ≤ .05 considered significant.

Results

Of the 470 eligible trial participants, ICES data linkage was possible for 462. We found no difference between intervention and control arm baseline characteristics (sex, age group, income quintile, Johns Hopkins ACG Aggregated Diagnosis Groups [ADGs],¹² geographic profile¹³) (Table 1). Johns Hopkins ADGs are a person-focused, diagnosis-based method of categorizing patients’ illnesses. Unadjusted comparison between groups using ICES data confirmed the numerically lower but non-statistically significance difference in ED visits at 12 months (mean [SD] 1.86 [2.57] visits vs 2.40 [3.45] visits, P = .055) that we found for the main trial primary outcome using participant self-reported health care utilization data. ICES data confirmed no difference in the number of in-patient hospitalizations or days in hospital but confirmed the difference in mortality favoring the intervention arm (Fig. 1).

Table 1.

Demographic Characteristics

Fig. 1.

Kaplan-Meier curves of survival by Program of Integrated Care and usual care for subjects with COPD and multiple comorbidities.

Total median (IQR) in-hospital costs in the 12 months following randomizations for all (intervention and control) 462 trial participants were CAD $4,769 ($417–16,834) (equivalent to US $3,566 [$312–12,588]). Total costs incurred in the community were higher at CAD $8,011 ($4,749–13,831) (equivalent to US $5,990 [$3,551–10,342]). Unadjusted comparisons identified no difference in in-hospital or community publicly funded health care costs between participants receiving the multicomponent COPD exacerbation prevention management intervention compared to participants in the control group (Table 2). Regression modelling indicated community health care costs were lower in those subjects allocated to the intervention arm after controlling for sex, income quintile, total ADG score of ≥ 7 (indicative of greater comorbidity) and living in an urban versus rural locality. There was no association with in-hospital costs by study arm allocation. Male sex and having a higher ADG score were associated with higher community and in-hospital costs (Table 3).

Table 2.

Unadjusted Health Care Costs at 1 Year

Table 3.

Adjusted Health Care Costs at 12 Months Following Randomization

Discussion

This a priori planned evaluation using health administrative databases in the province of Ontario, Canada, suggests that our multicomponent case manager–led COPD self-management program with action plans for COPD and comorbidity reduced public health care costs incurred through access to community services compared to participants randomized to usual care when adjusting for important covariates. Conversely, and contrary to our a priori hypothesis when designing the trial, we found no difference in in-hospital costs adjusting for the same covariates. Median community health care costs were almost double in-hospital costs, indicating the burden patients with COPD and multi-morbidity place on community resources.

Our trial was not designed to assess cost effectiveness of our intervention, and therefore, these results need to be interpreted with caution. The COPD Patient Management European Trial (COMET) did demonstrate cost savings of EUR €37.50/subject/y for subjects with severe or very severe COPD when considering unplanned all-cause hospitalization days, mortality, and quality-adjusted life expectancy.¹⁴ This trial evaluated a very similar case manager–led multicomponent self-management program but also included home telemonitoring with an e-health platform to aid in early exacerbation detection. As with our findings, the COMET trial also failed to demonstrate a difference in use of acute health care services but did demonstrate a reduction in mortality.¹⁵

We identified a median overall direct annual health care cost of CAD $6,460 (US $4,830). This figure is substantially less than a previous United States study evaluating health care insurance claims in working-age people with COPD (US $11,984).¹⁶ Similarly, a 2023 study using data from 1,073 subjects with COPD within the Medical Expenditure Panel Survey reported an all-cause medical cost/subject/y of US $19,449 of which US $6,145 was for prescription drugs.¹⁷ This difference in direct costs likely reflects differences in health care costings in Canadian versus United States health care systems. As well, medication costs were not included for approximately 25% of our cohort due to ineligibility for the Ontario Drug Benefit program and, therefore, not captured within ICES databases.

Despite no difference in acute health care costs, we did identify a reduction in community health care costs associated with our multicomponent case manager–led COPD self-management program. Recently the global economic burden of COPD has been estimated at $4.33 trillion from 2020–2050, with the largest economic burden predicted for China ($1.36 trillion) and the United States ($1.04 trillion).¹⁸ Given these substantial costs and that COPD prevalence continues to rise due to population aging, worsening air quality, and increasing total numbers of smokers worldwide,¹⁹ identification of exacerbation interventions that produce health care cost savings remains an imperative in COPD exacerbation management.

Strengths of our study include the use of extensively validated provincial health care databases that provide comprehensive data on both acute and community-incurred health care costs. Another strength is the data linkage from a prospectively recruited cohort who had participated in a randomized controlled trial of a self-management program for subjects with COPD and comorbidity. Our study has limitations. First, this study is subject to the limitations inherent to administrative database studies in terms of data availability and quality. However, reassuringly the patterns of health care utilization we identified using the databases reflected those we identified through participant self-report in real time throughout the trial. Second, our data do not reflect costs incurred by family members needing to provide care in the home to these participants including costs to them associated with lost income due to needing to provide care. They also do not consider private out-of-pocket and third-party insurance costs. Therefore, overall costs may be underestimated. Finally, our data may not be generalizable to jurisdictions such as the United States without publicly funded health care systems for acute and community health care provision. In such jurisdictions, there will be greater reliance on private health insurance and different cost models. In countries with similar models of publicly funded health care services, cost savings we identified in community services are likely to be replicable.

Conclusions

We identified lower public community health care costs but no difference in acute care costs in the 12 months following randomization associated with our multicomponent COPD exacerbation prevention management intervention compared to usual care. Community health care costs were almost double those incurred via accessing acute health care services. Given this finding, although most COPD exacerbation management interventions generally focus on reducing the use of acute care, interventions that enable health care cost savings in the community require further exploration.