Cost‐Effectiveness of Specialized Multidisciplinary Heart Failure Clinics in Ontario, Canada

Introduction

Heart failure (HF) is a complex, progressive syndrome characterized by abnormal heart function resulting in poor exercise tolerance, recurrent hospitalizations, and reductions in both quality of life and survival 1. Although tremendous progress has been made in pharmacologic and device therapy, HF patients continue to have a poor prognosis, with an annual mortality ranging from 5% to 50% 1. The incidence of HF is projected to increase, with estimates suggesting a threefold increase in HF hospitalizations over the next decade 2. Alternative targeted health care delivery models have, therefore, been of particular interest in HF, as a means of improving both quality of life and survival 3.

Disease management through specialized multidisciplinary clinics has been shown to improve patient outcomes in several health conditions, including asthma, diabetes mellitus, chronic kidney disease, and cancer 4, 5. The potential benefits of multi-disciplinary care in HF clinics include the improved utilization and compliance with evidence-based medications that prolong survival. Moreover, this model of care may better address the complex interplay between medical, psychosocial, and behavioral factors facing HF patients and their caregivers 3. Several previous randomized studies and meta-analyses have evaluated the efficacy of such clinics, with selected results suggesting a marked reduction in mortality 1, 3, 6. However, interpreting this literature is challenging because the composition of HF clinics and the interventions they offer have varied, as has the population studied 3.

From a health policy standpoint, it remains unclear if the benefit of HF clinics is balanced against the costs of the intervention itself and the subsequent future health care costs associated with more closely managed care. Previous economic evaluations of HF clinics have been restricted to relatively small clinical trials, most with short time horizons 3, 7-11. Accordingly, our objective was to determine the cost-effectiveness of specialized multidisciplinary HF clinics compared to standard care for the long term management of HF patients in Ontario, Canada.

Methods Research Ethics Board Approval

This study was approved by the Institutional Research Ethics Board at Sunnybrook Health Sciences Centre, Toronto, Ontario.

Study Design

We performed a cost-effectiveness analysis to model the costs and outcomes in a cohort of patients discharged after an index hospitalization for HF, comparing two treatment strategies: 1) treatment in a specialized multidisciplinary HF clinic (defined as care involving at least one physician and nurse, one of whom has specialized training in HF) versus, and 2) standard care (defined as care provided by a single practitioner). Outcomes of interest were life expectancy, measured in years, costs (adjusted for inflation to 2008 Canadian dollars using the Bank of Canada Consumer Price Index, http://www.bankofcanada.ca/en/cpi.html), and the incremental cost-effectiveness ratio (ICER), calculated as the incremental cost per life-year gained.

Economic Assumptions

The perspective of this analysis was that of the Ontario Ministry of Health and Long-Term Care (MOHLTC), the single third-party payer for health services in the province. The time horizon for the analysis was 12 years, the period for which accurate estimates of HF natural history in Ontario were available. All health outcomes and costs were discounted at 5% per year (http://www.cadth.ca).

Standard Care Cohort

The target population were patients with a recent hospitalization for HF. For the purpose of estimating survival gain and cost, we identified an actual cohort of all patients in the fiscal year 2005 that were discharged from hospital with a diagnosis of HF in Ontario. Patients were identified based on International Classification of Disease (ICD) Version 10 code I50 in the Canadian Health Institute for Health Information (CIHI) discharge abstract database. We restricted the cohort to patients above the age of 25 years who were residents of Ontario with valid Ontario Health Insurance Plan (OHIP) identification numbers. If an individual had more than one HF hospitalization for 2005, the first admission was defined as the index event. Based on this definition, we identified 16,443 hospitalized HF patients who represented our standard care cohort.

HF Clinic Cohort

A hypothetical HF clinic cohort was modeled, using the same 16,443 patients identified earlier. Life expectancy and costs were estimated in this modeled cohort as described below.

Estimation of Life Expectancy

We used age-gender specific mortality rates from the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) study to estimate the life-expectancy of HF with standard care. The EFFECT study was a chart abstraction of 9943 HF patients, across 44 hospitals in Ontario followed for up to 12 years. Patients in the EFFECT study were from a wide spectrum of clinical settings, including both large tertiary care centers and smaller rural community hospitals, and thus were representative of HF in Ontario. Survival curves were constructed for patients receiving standard care using the age-gender specific life-tables from the EFFECT study 12.

Estimates for life expectancy of patients treated in HF clinics were obtained from a systematic review and meta-analysis of the literature, which is published separately 13. To ensure that these efficacy estimates were representative of the treatment strategies in our model, the systematic review was restricted to randomized controlled trials of HF clinics consisting, at a minimum, of a nurse and physician, one of whom was a specialist in HF management 13. These trials compared HF clinics to standard care by a single practitioner, and the population was restricted to HF patients after discharge from hospital 13. Summary risk ratio (RR) estimates for mortality and hospitalization were calculated using the random effects model of DerSimonian and Laird (Table 1). The systematic review included eight randomized controlled trials (found at: http://www.ispor.org/Publications/value/ViHsupplementary/ViH13i8_Wijeysundera.asp) 14-21. The meta-analysis concluded that HF clinics are associated with a statistically significant 29% decrease in all-cause mortality (summary RR 0.71; 95% confidence interval [CI] 0.56–0.91) but a nonsignificant 12% increase in overall hospitalizations (summary RR 1.12; 95% CI 0.92–1.35) 13.

Table 1. Model input parameters Parameter Base-case value(95% CI) Source Parameter distribution for PSA RR for all-cause mortality 0.71 (0.56–0.91) Meta-analysis (13) Log-normal RR for all-cause hospitalization 1.12 (0.92–1.35) Meta-analysis (13) Log-normal RR for emergency visit 1 (0.5–1.5) Assumption Log-normal RR for physician assessment/lab test 1.2 (0.7–1.7) Assumption Log-normal RR for same day surgery 1 (0.5–1.5) Assumption Log-normal RR for medication 1 (0.5–1.5) Assumption Log-normal Annual attrition rate from heart failure clinics 0.1 (0–1) Assumption Beta CI, confidence interval; OR, odds ratio; PSA, probabilistic sensitivity analysis.

Survival curves for the HF clinic cohort were then constructed by applying the summary estimate from the meta-analysis to the natural history survival curves constructed from the EFFECT study. Based on expert opinion, we incorporated a 10% annual attrition rate of patients dropping out from the HF clinics into our model. We assumed that the survival benefit afforded by HF clinics only applied to patients who continued to receive care in these clinics. Patients who dropped out of HF clinic care were assumed to have the same mortality rate as those patients receiving standard care. We also assumed that noncompliant patients would not return to HF clinic care.

HF Clinic Costs

Incremental costs associated with treatment provided at HF clinics were identified from an existing HF clinic at the University Health Network (UHN) in Toronto, Ontario which we considered to be representative of specialized multidisciplinary HF clinics in the province. Where selected costs could not be valued, clinical experts were consulted. Briefly, care at the UHN HF clinic is primarily provided by a physician with specific training in HF management and an advanced care nurse practitioner. Care is also provided by allied health care professions as needed. On average, patients had two clinic visits per year; new patients or patients with unstable symptoms were evaluated more frequently.

The types of costs that were considered for the HF clinic are summarized in Table 2. These included costs associated with: 1) health practitioner visits and clinic staffing (including physician, nurse practitioner, pharmacist, dietician, social worker, kinesiologist, and clerical staff); 2) laboratory and imaging tests; and 3) operating and overhead (plant operations, cleaning, waste disposal and pest removal, fire safety, security, building repairs and maintenance, equipment depreciation, administrative fees, utilities). Staffing costs were estimated using a top-down approach based on annual staff salaries including benefits, adjusted by the proportion of time spent in the HF clinic. For laboratory and imaging test, we used a bottom-up approach, assuming that patients would have an electrocardiogram (EKG) every visit, an echocardiogram once a year, and annual screening blood-work assessing renal function, electrolytes and hematologic profile. We used these assumptions to estimate an average cost per 30-day period, which we assumed was constant over the model's time horizon.

Table 2. Costs associated with heart failure clinic Variable Total cost/year($ CAD 2008) Cost/30 patient-days*($ CAD 2008) Cardiac Technician 38,311 2.86 Physician 176,735 13.20 Clerical (booking) 58,523 4.37 Clerical (charting, data entry) 17,136 1.28 Dietician 4,539 0.34 Kinesiologist 13,322 1.00 Nurse practitioner 42,822 3.20 Pharmacist 9,326 0.70 Social worker 2,731 0.20 Operating costs 6,178 0.46 Utility charge 2,265 0.17 Blood Work 35,255 2.63 Electrocardiogram 32,455 2.42 Echocardiogram 255,860 19.11 Cost per 30 patient-days 52 * Cost per 30 patient-day block was calculated by dividing the 1 year total costs by the total number of patient visits in the clinic for 1 year, and multiplied by (30/365 days) to determine the cost per 30 patient-days. † 1 year cost calculated by product of yearly salary (including benefits) by average proportion of time spent in HF clinic. ‡ Patients assumed to have one echocardiogram per year, and one electrocardiogram (EKG) per visit. Costs Associated with Standard Care

Health-related costs for the standard care cohort were determined using a bottom-up approach by linkage to population-based administrative databases at the Institute for Clinical Evaluative Sciences (ICES), using encrypted unique patient identifiers 22. Administrative records were available up to March 31t, 2008, allowing cost-estimates for a maximum follow-up period of 36 months. We identified all health-related resources utilized by patients within the study period and paid for by the Ontario MOHLTC. The categories of costs included were all-cause physician visits, acute care and chronic care hospitalizations, emergency department visits, and same day surgeries. We included only costs associated with HF related medication use.

Costs associated with physician visits and laboratory tests were determined using data from the claims history in OHIP database, which includes fee-for-service claims submitted by physicians and other licensed health professionals]. It also includes shadow billings from providers of organizations covered by alternate payment arrangements. Because there are regional variations in reimbursements, the median 2008 cost for each physician and laboratory service fee code was used to estimate cost.

The CIHI discharge abstract database has records on the frequency and type of all acute and chronic care hospitalizations in the patients included in our cohort. The CIHI discharge record includes a “most responsible” diagnosis and up to 15 additional diagnosis codes that can be used to estimate comorbidity, as well as procedure codes, length of stay and in-hospital mortality data 22. The cost of hospitalization was estimated using the Resource Intensity Weight (RIW) methodology 22. We multiplied the RIW associated with the case-mix group for each hospitalization by the average provincial cost per weighted case for all Ontario acute and chronic hospitals 22. This method yields a mean cost per hospitalization for cases assigned to a particular case-mix group category.

A similar RIW methodology was employed to determine the costs for emergency department visits and same day surgeries, both using the National Ambulatory Care Reporting Service (NACRS) database 22. NACRS contains administrative, clinical, financial, and demographic data for hospital-based ambulatory care, including emergency department visits, outpatient surgical procedures, medical day/night care, and high-cost ambulatory clinics such as dialysis, cardiac catheterization, and oncology 22.

Finally, data on medication costs were obtained from the Ontario Drug Database (ODB), which has comprehensive drug utilization information on patients more than 65 years, for whom full drug coverage is provided for by the MOHLTC 22. We did not include medication costs associated with patients under the age of 65 years as these would not be covered by the provincial government. We restricted our analysis to HF medications because we did not anticipate that HF clinics would have any impact on non-HF medication use. HF medication classes included angiotension converting enzyme (ACE) inhibitors, angiotension receptor blockers (ARB), beta-blockers, digoxin, spironolactone, diuretics (furosemide, metolazone), hydralazine, and long-acting nitrates.

Health care costs associated with the treatment of these HF patients required modeling, because our follow-up period for observed linked costs was limited to 36 months and, therefore, did not span the 12-year time horizon of the analysis. Based on results of previous studies in cancer care, we expected that health-related costs would not be constant over the lifetime of HF patients 23. Instead, we expected that there would be a phase of high costs associated with the time period immediately after hospital discharge, followed by a phase of clinical stability characterized by relatively constant costs, and finally a phase of increasing costs before death 23. To validate our phased-based costing approach and determine the duration of the postdischarge and predeath phases of increased costs, we performed exploratory analyses of our linked cohort.

We evaluated the cost per consecutive 30-patient days for patient subgroups that survived 9 to 12 months, 21 to 24 months, and 33 to 36 months postdischarge (Fig. 1). As seen in Figure 1, the mean 30 patient-day costs curves confirmed our hypothesis of discrete cost phases. Inflection points separating the postdischarge and stable phases, and the stable and predeath phases were estimated to occur at 3 months postdischarge, and 6 months before death, respectively. Thirty patient-day blocks of consecutive costs were created within each costing phase, with three blocks for the postdischarge phase, six blocks for the predeath phase, and a single 30 patient-day block of consecutive costs for the stable phase (see Table 3).

image

Exporatory analysis on phases of long term cost associated with HF care.

Table 3. Long-term costs (all costs are reported in 2008 Canadian dollars) 30-day block Observed costs (standard care) Modeled costs (HF clinics) Physician Services Hospitalization ER Same day surgery Medications Overall costs Overall costs Postdischarge phase 1 block postdischarge 1,170 8,725 617 103 59 10,675 11,955 2 block postdischarge 462 2,267 129 47 56 2,961 3,326 3 block postdischarge 373 1,599 105 42 52 2,172 2,438 Stable phase Stable phase 144 384 36 23 31 617 692 Predeath phase 6 block predeath 437 2,344 178 37 66 3,062 3,430 5 block predeath 480 2,721 195 37 67 3,501 3,923 4 block predeath 530 3,241 211 30 65 4,077 4,571 3 block predeath 608 4,162 251 34 63 5,119 5,740 2 block predeath 872 7,389 356 41 57 8,716 9,777 1 block predeath 842 7,020 405 20 21 8,308 9,318 ER, emergency room; HF, heart failure.

We then assigned individual patient costs to each 30-day costing block within the three phases in a hierarchical fashion, first to the postdischarge phase, then to the predeath phase, and finally to the stable phase. For example, if a patient survived for 12 months postdischarge, the mean cost for each of the first 3 months were assigned to each of the corresponding three 30 patient-day blocks of consecutive costs of the postdischarge phase; the mean cost for each of the last 6 months of life were assigned to each of the corresponding six costing blocks in the predeath phase; finally, the remaining 3 months were assigned to the stable category.

Costs for each of the 16,443 patients in our standard care cohort were assigned in this manner. Table 3 summarizes the mean cost for each of the 30 patient-day blocks of consecutive costs. The cumulative lifetime costs for the standard care cohort were estimated by first determining the proportion of the original cohort in each costing block for each 30-day time point in the model over its 12-year time horizon. The total costs at each 30-day time point was then calculated by multiplying the mean cost per block (in Table 3), by the number of patients in the costing block. The cumulative costs were the sum of the costs across all the time blocks.

To model the lifetime costs for the HF clinic group, we adjusted the standard care cost per 30 patient-day block of consecutive costs using estimates from our systematic review (Table 1) and added this to the incremental intervention costs associated with HF clinics, as described earlier. For example, we found that all-cause hospitalization increased by 12% (Table 3). Therefore, the acute care hospitalization component of the mean 30 patient-day cost for standard care in each of the costing blocks in Table 2 was increased by 12%. Only a minority of the studies in the systematic review provided data on medication utilization. These suggested that although HF clinic patients had dose intensification compared to those in standard care, the number of medication classes prescribed was not statistically different. We assumed medication costs to be similar between treatment strategies and tested this in our sensitivity analyses. We expected that care in a specialized HF clinic would result in a greater number of subsequent cardiac investigations, such as cardiac magnetic resonance imaging or coronary angiography; based on expert opinion, we assumed a 20% increase in diagnostic testing in the HF clinic strategy. The modeled costs per 30 patient-days for each of the costing blocks for the HF groups are summarized in Table 3.

Sensitivity Analyses

One-way deterministic sensitivity analyses were performed to evaluate the robustness of our results. The ranges for the sensitivity analysis were obtained from the 95% confidence intervals from the source documentation (Table 3). We also performed a probabilistic sensitivity analysis (PSA), using second-order Monte Carlo simulation with 10,000 trials. Beta distributions were used to define all probabilities, and log-normal distributions were used to define costs and ORs; mean and standard deviations to define distributions were obtained from source documentation. Where standard deviations were not available, we assumed a standard deviation that was 50% of the mean. A cost-effectiveness acceptability curve was produced at varying willingness-to-pay thresholds by drawing parameter values at random from all distributions.

The cost-effectiveness analysis model was conducted in Microsoft Excel Version 2007 (Microsoft Cooperation, Redmond, WA), and the PSA was conducted using Oracle Crystal Ball Version 11.1.1 (Oracle Corporation, Redwood, CA). Long term health related costs were estimated using SAS Version 9.1 (SAS Institute Inc, Cary, NC).

Results

The characteristic of the 16,443 patients in our cohort are summarized in Table 4. The mean age of the cohort was 76.8 years, with 49.4% being male. A total of 39.8% of patients had an ischemic cardiomyopathy, while 45.9% had diabetes, 84.7% had hypertension and 17.2% had renal insufficiency.

Table 4. Baseline characteristics of cohort of patients with heart failure N = 16,443 (%) Mean Age (years) 95% CI 76.8 (76.6–77.0) % under age 65 95% CI 14.9 (14.4–15.6) Male 95% CI 49.4 (48.4–50.5) Coronary artery disease 95% CI 39.8 (38.9–40.8) Old myocardial infarction 95% CI 31.5 (30.7–32.4) Diabetes mellitus 95% CI 45.9 (44.9–47.0 Hypertension 95% CI 84.7 (83.3–86.1) Cerebrovascular disease 95% CI 8.1 (7.7–8.5) Renal insufficiency 95% CI 17.2 (16.6–17.9) Pulmonary disease 95% CI 18.0 (17.4–18.7) Dementia 95% CI 4.7 (4.3–5.0) Malignancy 95% CI 6.7 (4.3–5.0)

The estimated cost of treatment at a multidisciplinary HF clinic was estimated to be $52 per 30 patient-days, or $624 per patient per year. The individual components of care are summarized in Table 2. The major contributors to the overall cost of care were the physician assessment fee (25.4%) and diagnostic tests performed in the clinic (46.5%), most notably echocardiography (36.8%). Costs associated with nurse practitioner care were only 6.2% of total costs, while those associated with other staff represented nearly 21% of clinic costs.

The mean cost per 30 patient day costing block for long-term costs are presented in Table 3. Within both the postdischarge and predeath phases, there were substantial differences in mean cost between costing blocks. For example, the mean cost was $10,675 in the first 30 days after discharge, followed by a 75% reduction to $2961 for the second month postdischarge. Similarly, in the 6 months before death, there was a steep increase from $3062 in the first predeath costing block, to $8308 immediately before death. The largest contributor to overall health-related future costs was hospitalizations for all the costing blocks. Hospitalization costs were most prominent during the more acute phases of the diseases (i.e., the postdischarge and predeath phases), when they represented more than 80% of total costs. In contrast, in the stable phase hospitalizations represented only approximately 50% of costs, during which time costs associated with medications (5%) and physician services (15%) played a larger role.

At 12 years, nearly all of the patients in either cohort were projected to have died (94.6% in the standard care group versus 92.1% in the HF clinic group). However, death was delayed in the HF clinic cohort. The life expectancy of HF patients treated with standard care was estimated to be 3.21 years. In comparison, as seen in Figure 2, those treated at HF clinics were estimated to have an average survival of 3.91 years, a survival gain of approximately 8.5 months. The cumulative lifetime cost associated with standard care was $53,638 compared to $66,532 for patients in the HF clinic group. Thus, HF clinics cost $18,259 for each additional life-year gained (ICER is $17,427 for costs and health effects not discounted) (Table 5).

image

Survival curves for patients treated in health failure clinic versus standard care.

Table 5. Life expectancy, cumulative costs and incremental cost-effectiveness of heart failure clinics and standard care Undiscounted Cost (CAD 2008) Life expectancy (years) Standard care $61,870 3.87 Heart failure clinic $77,882 4.78 Δ $16,012 0.92 ICER $17,427 Discounted (costs and life expectancy: 5%) Cost (CAD 2008) Life expectancy (years) Standard care $53,638 3.21 Heart failure clinic $66,532 3.91 Δ $12,895 0.71 ICER $18,259 HF, heart failure; ICER, incremental cost-effectiveness ratio; Δ, difference.

Deterministic (one-way) sensitivity analyses demonstrate that these results were robust, across the range of plausible values. Specifically we did not find that our results varied if medication and diagnostic tests costs associated with specialized HF clinics increased by 50%. Importantly, if the mortality benefit associated with HF clinics was assumed to be the limits of the 95% confidence interval from the systematic review (RR 0.56–0.91), the HF clinic strategy remained cost-effective. In addition, HF clinics remained cost-effective if the proportion of patients who dropped out annually was varied from 1% to 90%. Of 10,000 simulations of the PSA, 99.4% were cost-effective at a willingness to a pay threshold of $50,000, as seen in the cost-effectiveness acceptability curve displayed in Figure 3.

image

Cost-effectiveness acceptability curve.

Discussion

We performed a cost-effectiveness analysis from the perspective of the MOHTLC of Ontario comparing multidisciplinary HF clinics to standard care for patients discharged after a hospitalization for HF. We found that HF clinics were associated with an improvement in estimated life expectancy of approximately 8.5 months over the 12-year time horizon of our model, a substantial increase given the poor prognosis associated with this condition. This survival benefit balanced against the increased costs associated with the implementation of the multidisciplinary clinic itself and a small increase in future hospitalizations. In contrast to previous economic evaluations of HF clinics, our study examined a large, real-world cohort over a long time horizon 3, 7-11. Moreover, ours is the first study in the literature to use accurate administrative datasets to estimate long term health related costs 3, 7-11. These results were robust across a wide plausible range of parameters, and alternative assumptions regarding costs and benefits of HF clinics, thereby providing evidence to suggest that specialized multidisciplinary clinics are a cost-effective means of providing ambulatory care to HF patients.

The prognosis for patients with HF has improved over the last two decades with the introduction of neurohormonal modulating therapies such as angiotensin-converting enzyme (ACE) inhibitors, β-blockers, and aldosterone inhibitors as the mainstay of pharmacological therapy for this complex condition. In the past 5 years, improvements in device therapy with the use of automated implantable cardiovertor defibrillators (AICD) for the prevention of arrhythmic deaths and resynchronization therapy in suitable candidates has further reduced mortality. Nonetheless, despite the availability of these therapies uptake remains poor in part because the optimal use of these treatments requires close supervision by appropriately trained personnel. The majority of HF patients in Canada are treated by primary care physicians, who may lack the expertise or time to optimize their patients' medications or identify suitable candidates for advanced device therapy 24.

Multidisciplinary clinics likely improve disease management through a number of mechanisms. Given the focus on one particular disease and enhanced ability for close monitoring, patients at a HF clinic may be more likely to receive appropriate medications and, more importantly, receiving optimal doses 1, 6. Dose intensification to the levels used in clinical trials is critical in order for patients to realize the maximum benefit of these medications. Such dose intensification is facilitated by the specialized supervision available at HF clinics. Furthermore, these complex patients often have concomitant medical, behavioral and social challenges, all of which need to be addressed 1, 6. As such, the availability of allied health professions such as pharmacists, dieticians, social workers and exercise therapist likely contribute to the survival benefit associated with HF clinics.

Current American and Canadian practice guidelines suggest as a Type 1 recommendation that certain subsets of HF patients, specifically those recently admitted to hospital for a HF exacerbation, should be referred to a specialized HF clinic 1, 6. Our study reinforces this recommendation by suggesting that this benefit was cost-effective compared to the traditional willingness to pay threshold of $50,000. This cost-effectiveness persisted despite an apparent increase in long-term hospitalizations and their associated costs.

Our study has important implications for HF care. Given the current climate of limited health care resources, it is essential that any new treatment strategy demonstrate a favorable incremental cost for its additional health benefit. We found that HF clinics had an ICER of approximately $18,000 per life-year gained, which compares favorably to other recently adopted cardiac technologies, such as drug eluting stents (ICER > $27,000 per quality-adjusted life-year gained) 25-27. As our perspective was that of the third party payer (MOHLTC), we did not incorporate costs, such as caregiver expenses or productivity costs. Given the mortality benefit of HF clinics, and the evidence that disease management strategies improve functional status, we expect that a greater proportion of patients treated at HF clinics would be able to return to work. However, as the majority of o

留言 (0)

沒有登入
gif