Costs and medical care consequences associated with the diagnosis ...

ORIGINAL RESEARCH ARTICLE

Pharmacoeconomics 2005; 23 (7): 733-742 1170-7690/05/0007-0733/$34.95/0  2005 Adis Data Information BV. All rights reserved.

Costs and Medical Care Consequences Associated with the Diagnosis of Peripheral Arterial Disease Kristen Migliaccio-Walle,1 J. Jaime Caro,1,2 Khajak J. Ishak3 and Judith A. O’Brien1 1 2 3

Caro Research Institute, Boston, Massachusetts, USA Division of General Internal Medicine, McGill University, Montreal, Quebec, Canada Caro Research Institute, Montreal, Quebec, Canada

Abstract

Background: Peripheral arterial disease (PAD) is increasingly recognised as an indicator of disseminated atherothrombosis, but its impact on use of healthcare resources is not well understood. Objective: To provide a quantitative description of the resource utilisation and costs incurred following PAD. Methods: Hospitalisations, physician visits and the corresponding direct medical costs were examined in 16 440 patients with a diagnosis of PAD (1985–1995) in Saskatchewan, Canada, and compared with 15 590 reference patients with a diagnosis of myocardial infarction (MI) [1990–1995]. Medical history and patient characteristics were available retrospectively to January 1980 and follow-up to December 2000. Rates and timing of all-cause and cardiovascular hospitalisations and physician visits within discrete periods in the 10 years following PAD diagnosis, and 5 years following MI, were evaluated, as were lengths of stay and predictors of hospitalisation. Results: Average follow-up was 5.9 years among patients with PAD and 3.6 years for MI. Half (55%) of patients with PAD were male versus 64% of reference patients. The mean ages were 67.3 and 66.9 years, respectively. Patients with PAD were hospitalised most frequently soon after diagnosis, with rates subsequently decreasing to 0.14 per month. These rates were similar in the reference group except for the period immediately following MI. The average 5-year cost post-diagnosis (2002 $Can) per patient was $Can41 968 vs $Can48 578 for the reference population. Conclusions: A diagnosis of PAD not only imposes a severe burden on patients and their families, but it also significantly increases the use of healthcare resources and the associated costs. By the end of year 1, this burden is comparable with a diagnosis of MI.

Peripheral arterial disease (PAD) is a common illness in older adults.[1-5] Prevalence estimates are highly dependent on how PAD is defined or identified;[2,5] nevertheless, conservative estimates suggest that 5–10% of the adult population [2,3,5-8] experiences PAD. As it has become clearer that PAD is

evidence of disseminated atherothrombotic disease[9-13] with similar risk factors as myocardial infarction (MI),[6,14] there has been greater interest not only in alleviating the symptoms, but also in treating the underlying condition itself. This has led to the use of antiplatelet agents[9,15] and other drugs such as

734

antihyperlipidaemics, β-adrenoceptor antagonists and calcium channel antagonists.[16-18] While the burden imposed by MI on patients, their families and the healthcare system has been studied and documented extensively, [19] the impact of PAD on resource use and costs is much less well understood. To address this question – one that will gain importance as more treatments for the condition are considered – requires detailed knowledge of the resources consumed by patients with this diagnosis. In this paper, we report the results of analyses of the course over 10 years of a large cohort of patients diagnosed with PAD. These analyses provide a quantitative description of the resource utilisation and costs incurred following this diagnosis. Documentation of this burden provides the basis for examination and understanding of the economic implications of emerging treatments for this disease. Methods The healthcare databases managed by Saskatchewan Health were used to obtain data on patients following a diagnosis of PAD. Saskatchewan Health is a provincial government department in Canada that oversees ten healthcare databases, including all formulary outpatient prescriptions and physician services, hospitalisations and vital statistics for approximately 1 million residents covered by provincial health insurance. The databases can be linked electronically using unique patient identifiers.[20] Residents of Saskatchewan, Canada, covered by Saskatchewan Health, which provides free healthcare for all residents, diagnosed with PAD between 1 January 1985 and 31 December 1995 either during a hospitalisation or a physician visit were eligible if they were aged ≥21 years at the time of diagnosis; no other exclusion criteria were specified. This cohort was selected as part of a larger study[21] of patients with atherothrombosis (MI, ischaemic stroke or PAD) in which the first diagnosis documented during the study period was taken as the identifying diagnosis. Patients with PAD were identified using the International Classification of Diseases (9th Edition) [ICD-9] codes 440, 440.2 or 443.9. The broader three-digit code 443 was allowed if crossed with documentation of a prescription for  2005 Adis Data Information BV. All rights reserved.

Migliaccio-Walle et al.

pentoxifylline, on the assumption that all patients who received this prescription were diagnosed with PAD.[22] Pentoxifylline is not indicated for any other diagnoses covered by ICD-9 code 443. Patient history was available back to 1 January 1980 and follow-up was through December 2000 or until a patient could no longer be followed because of emigration or death. Co-morbid conditions present at the time of diagnosis that were considered in the analyses included prior MI, angina, prior stroke, prior transient ischaemic attack (TIA), hypertension, diabetes mellitus, atrial fibrillation, heart failure and high cholesterol. All these diagnoses were identified using ICD-9 codes except for high cholesterol, which was identified based on documentation of a prescription for a cholesterol-lowering drug prior to the index event. Data on smoking history or laboratory values were not available in these databases. The date of first diagnosis, either at hospital admission or physician visit, depending on where the index diagnosis was made, was taken as the date of entry into the study. Reference Population

Analyses were also carried out in a reference population comprised separately of Saskatchewan residents aged ≥21 years experiencing an MI (ICD-9 410.x) between 1 January 1990 and 31 December 1995. The first occurrence of an MI during the study period was taken as the reference event. This does not necessarily represent the first event ever. Patient history was available as described for the study population. Outcomes

Resource use, total costs and survival were estimated for each patient over time following the index diagnosis of PAD. Eight periods of time were considered in these analyses: four in year 1 (30 days, 60 days, months 3–6, months 7–12) and each year thereafter through year 5. Inpatient (hospitalisations and procedures, excluding long-term care) and physician services were evaluated to determine the resource use. Hospitalisations and physician visits were categorised into two groups: all-cause- and cardiovascular disease (CVD)-related. Services were considPharmacoeconomics 2005; 23 (7)

Burden of Peripheral Arterial Disease

ered CVD related if the primary diagnosis was for ischaemic stroke (ICD-9 433–434, 436 or 362.3), MI, PAD, TIA (ICD-9 435), stable or unstable angina (ICD-9 411, 413) or heart failure (ICD-9 428). Procedures were divided into angiography (Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures [CCP] Codes 48.96-48.98, 50.8), coronary artery bypass graft (CABG) [CCP 48.1–48.2], percutaneous transluminal coronary angioplasty (PTCA) [CCP 51.59] and PAD-related procedures. The latter included any one of the following procedures: amputation (CCP 96.1), embolectomy (CCP 50.0), aortoiliofemoral bypass (CCP 51.25), intra-abdominal bypass (CCP 51.1, 51.26), other peripheral bypass (CCP 51.29), endarterectomy (CCP 48.01, 48.04, 50.1), arteriography (CCP 48.96, 48.97) or arterial surgery (CCP 48.1–48.9, 50.0–50.7). Hospitalisations as a result of bleeds were also considered in the costing analysis. Services with a primary diagnosis of intracranial haemorrhage, gastrointestinal haemorrhage or unspecified haemorrhage were considered bleed related.

735

ed by dividing the number of resources consumed by the number of patients using the resource in the period multiplied by the number of months in the period. This has been standardised to a 1-month duration to enable comparisons over time (otherwise the varying lengths of the periods would create apparent differences). ‘Hazard’ is a measure of the risk of consuming that resource (e.g. the likelihood of being hospitalised) – at least once – and takes into account actual follow-up time. It is calculated by dividing the number of patients who consume a resource by the patient time accumulated in the period. This can be converted to the cumulative risk over any particular time using standard exponential hazard equations. The third measure, ‘total rate’, considers both the total use (like the ‘frequency’) and the follow-up time, so that it addresses the entire rate of use. It is the number of resources consumed divided by the patient time accumulated in the period. Cox proportional hazards analysis was used to evaluate the impact of potential determinants of resource use, and Kaplan-Meier analyses were employed to determine the cumulative proportion of patients hospitalised over time.

Costs

Unit costs corresponding to inpatient care (including procedures) for CVD- and bleed-related diagnoses were obtained from current Ontario Case Costing Project (OCCP) case mix groups reports;[23,24] the Alberta Ambulatory Care Costs[25] were used for related ambulatory procedures (e.g. cardiac catheterisation). These costs were then applied on a patient-by-patient basis to the resources consumed in each period to estimate the average total cost per patient in each period following diagnosis (all costs are reported in 2002 Canadian dollars). No discounting is applied because there is no comparison where differential timing is involved. Analyses

The mean number of resources consumed per patient in each period was calculated, as were the rates of resource use. Rates of hospitalisation were calculated in several ways: frequency; hazard; and total rate. The measure ‘frequency’ gives an idea of how much the resource is used in each period by patients who consume at least one unit. It is calculat 2005 Adis Data Information BV. All rights reserved.

Results Demographic Characteristics

Of the 16 440 patients identified with a diagnosis of PAD, one was excluded from the analysis for administrative reasons. The majority of patients (71.2%) were identified by ICD-9 code 440, followed by 443 (18.7%), 443.9 (7.0%) and 440.2 (3.1%). Patient characteristics are shown in table I. Almost all patients (92.8%) had at a history of another condition or were aged ≥65 years at the time of diagnosis. The mean age of patients was 67.3 (SD 9.2) at the time of diagnosis. The average follow-up time for these patients was 5.9 years, during which 26.9% of patients died in hospital, 33.6% of which occurred while hospitalised for a CVD-related condition. Follow-up ended prematurely in 62.7% of patients, mostly because of death (92.6%); 7.4% of patients emigrated. We also identified 15 590 patients who had experienced an MI, of whom 31 were excluded for administrative reasons. Of these patients, 56.3% Pharmacoeconomics 2005; 23 (7)

736

Migliaccio-Walle et al.

Hospitalisations

Of patients diagnosed with PAD, 10.7% were hospitalised an average of 1.08 times in the first month following diagnosis. This translates to a rate of 1.43 hospitalisations per patient year (table II). Throughout the remainder of year 1, 9.0%, 20.5% and 24.4% of patients were hospitalised in month 2, months 3–6 and months 7–12, respectively. The highest frequency of use in this period was in month 2: 1.07 hospitalisations per month per patient hospitalised, a rate of 1.18 per patient year. Approximately one-third of patients required hospitalisation in each year thereafter, at a frequency eight times Table I. Patient characteristics and medical history at the time of diagnosis of patients diagnosed with peripheral arterial disease (PAD) vs myocardial infarction (MI)a Item

PAD population (no. [%])b

MI population (no. [%])b

Evaluable patients

16 439

15 559

Mean age [y] (SD)

67.3 (9.2)

66.9 (11.1)

Male

9029 (55)

9935 (64)

Patients with a qualifying diagnosis in hospital

1659 (10)

8752 (56)

atrial fibrillation

862 (5)

902 (6)

angina

5446 (33)

6426 (41)

heart failure

4177 (25)

3892 (25)

diabetes mellitus

3172 (19)

3682 (24)

hypercholesterolaemia

1128 (7)

1495 (10)

hypertension

9554 (58)

9484 (61)

ischaemic stroke

2110 (13)

999 (6)

prior MI

2868 (17)

Medical history

transient ischaemic attack 2217 (13)

3668 (24) 1369 (9)

a

Data from healthcare databases managed by Saskatchewan Health, Canada.

b

Unless otherwise indicated.

 2005 Adis Data Information BV. All rights reserved.

PAD pop - any PAD pop - CVD MI pop - any MI pop - CVD

100 90 80 Cumulative proportion (%)

were diagnosed in hospital (table I). Almost all patients (92.1%) had a history of another condition or were aged ≥65 years at the time of diagnosis; 59.3% had already experienced a CVD-related event. The mean age was 66.9 years (SD 11.1) at the time of the event. The average follow-up time for these patients was 3.6 years, during which 15.3% died in hospital, 42.1% of which occurred while hospitalised for a CVD-related condition. Half of the patients (52.2%) ended follow-up early: 93.6% of them died and 6.4% emigrated.

70 60 50 40 30 20 10 0 0

12

24

36

48 60 72 Time (months)

84

96

108 120

Fig. 1. Cumulative proportion of patients with peripheral arterial disease (PAD) or myocardial infarction (MI) hospitalised at least once over the follow-up period. CVD = cardiovascular disease.

lower than earlier in follow-up (0.14–0.15 per month per patient hospitalised during the year). These rates were as high as those observed in the reference population (table III); only in the very acute period following the index MI was a slightly greater proportion of patients hospitalised (p < 0.0001) versus PAD patients. The cumulative proportions of patients hospitalised are presented in figure 1. Among patients diagnosed with PAD