Vascular Medicine 1999; 4: 1–7

Medical costs of treating venous stasis ulcers: evidence from a retrospective cohort study Jeffrey W Olina, Kathleen M Beusterienb, Mary Beth Childsa, Caroline Seaveyb, Linda McHugha and Robert I Griffithsb,c Abstract: Venous stasis ulcers (VSU) account for approximately 80–90% of lower extremity ulcerations. Given their prevalence and chronic nature, VSU are thought to impose a significant economic burden on Medicare (the USA’s largest health insurance program) and other third party payers. However, comprehensive studies on the costs of VSU treatment are lacking. The objective of this study therefore was to examine comprehensively the direct medical costs of treating patients with a VSU in routine clinical practice. A cohort of 78 patients who presented with a VSU to the Cleveland Clinic Foundation (CCF), a large primary and tertiary referral center, was studied retrospectively. All inpatient and outpatient costs related to VSU treatment that were incurred during the year following VSU presentation or until the ulcer healed, whichever occurred first, were quantified. A total of 71 (91%) patients healed during the study. The average duration of follow-up was 119 days (median: 84 days). The average number of visits per patient was seven (range: 2 to 57). A total of 14 (18%) patients underwent 18 hospitalizations for VSU care. The average total medical cost per patient was $9685 (median: $3036). Home health care, hospitalizations and home dressing changes accounted for 48%, 25% and 21% of total costs, respectively. Total costs were related to duration of active therapy, ulcer size and the presence of at least one comorbidity (p⬍0.05). VSU are costly to manage, especially when time to healing is prolonged. The present findings reflect an underestimate of VSU costs since indirect costs were not examined. Time absent from work, forced early retirement, loss of functional independence and unquantifiable suffering may be additional factors that contribute to the overall burden of VSU. Key words: economics; health care costs; medical resource utilization; venous stasis ulcer

Introduction Venous stasis ulcers (VSU) account for 80–90% of lower extremity ulcerations.1 Venous insufficiency is caused by venous hypertension that occurs as a result of obstruction and/or reflux. Dilatation of the capillaries and leakage of plasma proteins and red blood cells are a direct result of venous hypertension. This leads to fibrin deposition and impaired oxygen transport, producing ischemia and hypoxia that result in cell death and ulcerations.2 Trauma or irritation are often the initiating event in VSU development. Healing time for VSU is unpredictable.3 VSU are subject to delayed healing because delivery of oxygen and other nutrients are impaired by edema and fibrin deposition. Treatment of VSU involves controlling edema using various compression techniques and promoting healing with the application of dressings. Surgical interventions for VSU include wound debridement, vein ligation and skin grafting.4 The actual prevalence of VSU in the USA is unknown,

a

Department of Vascular Medicine, Cleveland Clinic Foundation, Cleveland, OH, bCovance Health Economics and Outcomes Services Inc., Washington, DC, and the cProgram for Medical Technology and Practice Assessment, Johns Hopkins University School of Medicine, Baltimore, MD, USA Address for correspondence: Jeffrey W Olin, Department of Vascular Medicine, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA.

 Arnold 1999

but it has been estimated that 500 000 to 600 000 persons have a VSU.5 Although the economic burden of VSU is often overlooked, VSU treatment costs are substantial.6 In the 2-year period from 1990 to 1992, there were more than 1.3 million outpatient visits in the USA for venous ulcers.6 A projection of costs for 4 months of outpatient treatment from a study in England on the USA population predicted that costs in the USA are between $775 million and $1 billion annually.1 Health Care Financing Administration (HCFA) regulations that restrict payment for compression garments may translate into reduced patient compliance, leading to longer healing times and increased medical costs.7 Although large epidemiological studies have estimated the economic burden of VSU, no studies in the USA have comprehensively examined the direct medical costs associated with treating VSU in a clinical practice setting. Several studies have compared the costs and outcomes of different dressing regimens with follow-up periods ranging from 6 weeks to 6 months.8–11 These studies enumerate dressing costs and, if applicable, nursing costs, but none examined the costs of physician services, outpatient visits, or hospitalizations for VSU care. The present study examined VSU treatment practices over the course of 1 year. The perspective of the economic analysis is that of the health care system and focuses only on direct medical costs. The specific objectives of this research were to estimate the costs of all resources utilized by patients for VSU treatment by delineating inpatient and outpatient components of care, to identify factors that lead 1358-863X(99)VM259OA

2 JW Olin et al

to increased resource utilization and to describe the distribution of resources and associated costs of VSU management over time.

Methods Patients The study population comprised a cohort of patients who presented to the Cleveland Clinic Foundation (CCF), a large primary and tertiary referral center in Cleveland, Ohio, with a venous stasis ulcer in 1995. Selected patients all met the following criteria: 1) Assigned an International Classification of Diseases, Ninth Revision (ICD-9) code of 454.0 (venous stasis ulcer) or 454.2 (venous stasis/inflamed ulcer) in 1995. 2) Presented to the CCF for care of VSU in 1995. The VSU met the criteria for a venous ulcer based on the reporting standards of venous disease from the International Consensus Committee on Chronic Venous Disease.12 3) Received all medical care for the VSU from the CCF. This was determined by the frequency of outpatient visits for VSU care, which must have occurred within 6-week intervals. A total of 78 patients were eligible for the study. A clinical evaluation had been performed at the time of VSU diagnosis to ensure that the patients had adequate arterial circulation to enable the ulcer to heal using a standard VSU treatment regimen. In patients with pulses that were difficult to palpate, ankle/brachial indices and pulse volume recordings were obtained to assure that the arterial circulation was adequate to heal the venous ulceration. Data collection A detailed review of patients’ medical charts was performed to identify medical services related to VSU treatment during 1 year following the initial visit or until the ulcer healed, whichever occurred first. For patients with multiple ulcers on one or both legs, only the ulcer with the longest duration was followed. Demographic, clinical and medical resource utilization data were abstracted from the charts. Demographic data included age, gender, weight, previous VSU history and comorbid conditions. Clinical data included ulcer size, location, time since ulcer first occurred and the date that the ulcer healed, if applicable. Medical resource data included outpatient and inpatient visits attributable to VSU care, dressings/supplies used, prescription medications, procedures performed and home health care visits. Assignment of costs to medical resources The sources for assigning costs to medical services included CCF charge data and Medicare databases (the USA’s largest health insurance program). All costs were estimated to reflect 1997 US dollars. Inpatient facility and physician costs Inpatient facility and physician costs were obtained from automated charge data provided by CCF for each inpatient stay. The total facility charge for each inpatient stay was Vascular Medicine 1999; 4: 1–7

then converted to a cost by means of CCF’s hospital costto-charge ratio (0.481) from Medicare’s Prospective Payment System Payment Impact File. Outpatient facility costs To estimate facility costs for an outpatient visit with ulcer debridement (the only outpatient procedure related to VSU treatment that was performed), algorithms were developed that accounted for all variables affecting costs, including equipment, room preparation and non-physician, clinical and administrative staff time. To estimate costs for dressings applied during an outpatient visit, algorithms were developed that accounted for all the supplies needed to change a dressing, which generally included a gauze fluff roll, gloves, dressing and soap. The most common regimen used to treat VSU was the UlcerCare system (Jobst, Toledo, OH, USA). This consisted of either a wet to dry dressing (changed twice a day) or an absorbent dressing (Duoderm: ConveTec, Princeton, NJ, USA; Allevyn: Smith & Nephew Medical Ltd, Hull, UK; Sorbson: Dow Hickam Pharmaceuticals, Sugar Land, TX, USA) changed daily. An inner liner stocking applying 10–15 mm Hg of pressure was worn over the dressing pad day and night. An outer zippered compression stocking (20–30 mmHg compression) was worn during waking hours. Physician outpatient costs Medicare’s Resource Based Relative Value Scale (RBRVS) was used to estimate the cost for outpatient physician services, which included evaluation and debridement. Physicians’ Current Procedural Terminology (CPT) codes corresponding to these services were matched to the corresponding relative value units from the RBRVS by using the Medicare program fee schedule for physician services. The relative value units were then converted to Medicare’s allowed charges for physician services by multiplying the relative unit by the national adjusted conversion factor for the fiscal year 1997. Home dressing supply costs Costs for home dressing changes were estimated based on the prescribed frequency of dressing changes. It was assumed that the supply cost of a dressing change at home was the same as the cost of the supplies used for dressing changes in the outpatient facility. Home health care visits The cost of a home health care visit was estimated from the charge per visit provided by CCF’s home health care agency. This charge was converted to a cost using CCF’s hospital cost-to-charge ratio. It was assumed that the frequency of home health visits corresponded with the frequency of physician-prescribed dressing changes at home. Prescription medications For each medication prescribed, the 1997 Drug Topic’s Redbook13 provided the average wholesale price for the lowest generic brand. This average price was then discounted by 20%. These prices were used to assign costs to the amount of each medication prescribed.

Medical costs of treating venous stasis ulcers 3

Statistical methods Medical resource utilization and direct medical costs were summarized using descriptive statistics, including mean and median values with 25% to 75% interquartile ranges. Additionally, the follow-up period was divided into 3month intervals and costs incurred per interval were summarized for the total sample and for the subgroups of patients with ulcer sizes less than or equal to, versus greater than, 2.8 cm2. Associations among total costs and demographic characteristics and duration of follow-up were evaluated using the Spearman correlation coefficient (␳) and the Mann–Whitney U test, the non-parametric alternative to the Student’s t-test, as appropriate. All data analyses were performed using version 6.12 of the Statistical Applications Software (SAS Institute, Cary, NC, USA).

Results Table 1 reports the demographic characteristics of the 78 patients who met the inclusion criteria. The mean age was 67 years and 54% were female. The majority of the patients were white (68%) and covered by Medicare (63%). A total of 29 (37%) patients had at least one comorbidity, the most frequent of which was obesity, which occurred in 21 (27%) patients. Ulcer characteristics at the initial visit are shown in Table 2. Ulcer size was available for 59 patients, for whom the median ulcer size was 2.8 cm2 (mean: 9.4 cm2; range: 0.1–90 cm2). The median duration of time that the ulcer was present prior to the initial visit was 0.5 months (range: 0–180 months). Table 1 Baseline demographic characteristics. Characteristic

Total sample (n = 78)

Mean age (SD)

67 (16)

Gender (%) Female Male

42 (54) 36 (46)

Previous history of VSU (%)

51 (65)

Mean body mass indexa (SD)

32 (9)

Raceb (%) White Black Hispanic Other

50 21 2 1

Insurance (%) Medicare Private Medicaid

49 (63) 28 (36) 1 (1)

Comorbid conditionsc (%) Obesity Diabetes Hypertension Atherosclerotic heart disease Peripheral arterial disease Carotid artery disease Cancer

21 16 29 12 14 9 7

(68) (28) (3) (1)

(27) (21) (37) (15) (18) (12) (9)

Table 2 Characteristics of ulcer at baseline visit. Characteristic

Total sample (n = 78)

Lega (%) (n = 78) Left Right

51 (65) 27 (35)

Site (%) (n = 75) Calf Ankle Foot

42 (56) 32 (43) 1 (1)

Leg area (%) (n = 75) Medial Lateral Anterior Posterior Circumferential

39 14 15 4 3

(52) (19) (20) (5) (4)

Size (cm2)b (n = 59) Mean (SD) Median (range)

9.4 (18.9) 2.8 (0.1–90)

Ulcer duration prior to baseline (months) (n = 78) Mean (SD) Median (range)

6.9 (26) 0.5 (0–180)

a Four patients had ulcers on both legs; for these patients the ulcer with the longest follow-up period was included in the analysis. b If ulcer size at baseline visit was missing, ulcer size at first follow-up visit was used if available (applied to nine cases).

The mean duration of follow-up was 119 days (median: 84 days; range: 7–356 days). A total of 71 (91%) patients’ ulcers healed during the study. The mean time to healing among these patients was 108 days (median: 77 days). Overall, there were 528 CCF visits, including hospitalizations. The mean number of visits per patient was seven (range: 2 to 57 visits). The proportions of patients who used different health services are shown in Table 3. Outpatient debridement was performed for 24 (31%) patients, and 14 (18%) had a total of 18 hospitalizations for VSU care. The mean length of stay was 11 days with a range of 3 to 27 days. Seven hospitalized patients underwent debridement and skin grafting for treatment of their ulcers. Two patients’ ulcers were grafted twice. The other hospitalized patients’ ulcers were treated with one or more of the following: bed rest, leg elevation, debridement, whirlpool treatments, antibiotics and dressing changes. Table 3 Patient utilization of health services for venous stasis ulcers (n = 78). Health service

n (%)

Outpatient visita Outpatient debridement Hospitalization Home health care Prescription medications Oral antibiotics Oral pain medications Topical medications

78 24 14 21 20 13 12 10

(100%) (31%) (18%) (27%) (26%) (17%) (15%) (13%)

a

Missing data on 20 patients. Missing data on 4 patients. Patients could have more than one comorbid condition.

b c

Vascular Medicine 1999; 4: 1–7

a Average number of visits per patient is seven (range: 2 to 57 visits).

4 JW Olin et al

A comparison of the demographic characteristics between patients who did and did not utilize home health care services showed that substantially more females used home health care. Among those using home health, 71% were female and 29% were male. Among patients who were hospitalized, a higher percentage had a previous history of VSU (93%) compared with those not hospitalized (62%). In addition, a higher percentage of patients who were hospitalized were black (62%). Table 4 shows patient and cohort total costs by type of health service. The mean ⫾ SD total cost per patient was $9685 ⫾ $14 136 (median: $3036; interquartile range: $1137 to $10 854). The mean ⫾ SD hospitalization cost per patient was $2445 ⫾ $7313. When only those patients who were not hospitalized were included (n = 64), the mean ⫾ SD total cost per patient was $5965 ⫾ $10 201 (median: $1935; interquartile range: $1002 to $5869). Among patients who did not receive home health care (n = 57), the mean ⫾ SD total cost per patient was $4523 ⫾ $8562 (median: $1660; interquartile range: $976 to $4739). The mean ⫾ SD hospitalization costs associated with skin grafting were $11 654 ⫾ $5031. Figure 1 shows the distribution of costs across the medical services. Home health care, hospitalizations and home dressing changes accounted for 48%, 25% and 21% of total costs, respectively. Total costs were strongly related to duration of followup (␳ = 0.67, p ⬍ 0.0001) and less strongly related to ulcer size (␳ = 0.34, p ⬍ 0.01). Patients with at least one comorbidity incurred higher costs than those with no comorbidities ($11 214 versus $6245 per patient, respectively: p ⬍ 0.05, Mann–Whitney U test). No significant correlations were observed between total costs and age, body mass index and duration of ulcer prior to the initial visit. Although the difference was not statistically significant (p = 0.08, Mann–Whitney U test), patients with a history Table 4 Costs of health services per patient for venous stasis ulcers (n = 78). Health service

Total outpatient Facility Physician

Mean (SD) US$

524 (651) 251 (388) 273 (274)

Total hospitalization

2445 (7313)

Total home dressing (dressing and compression stocking/liner)

2035 (2035)

Total prescription medication Oral antibiotics Pain medications Total topical medications (corticosteroids, antiinfectives and antibiotics)

364 (204–608) 171 (94–301) 201 (109–364) 0 (0–0) 1296 (679–2743)

29 (73)

0 (0–18)

18 (62) 11 (33)

0 (0–0) 0 (0–0)

3 (13)

0 (0–0)

Total home health care 4652 (10 364) TOTAL

Median (interquartile range) US$

9685 (14 136)

Vascular Medicine 1999; 4: 1–7

0 (0–3042) 3036 (1137–10 854)

of VSU incurred higher costs than those without a VSU history ($11 007 versus $3470, respectively). Figure 2 shows the mean total costs per patient during each 3-month follow-up interval. The mean costs per patient were similar across the first three follow-up intervals. Patients incurred an average cost of $5736, $5088 and $5108 during the 0–90-day, 91–180-day and 181–270-day intervals, respectively. The average cost decreased to $2602 during the 271–365-day interval. Figure 3 shows the mean total costs per patient during each 3-month follow-up interval stratified by an ulcer size of 2.8 cm2, which was the median size at baseline. During the initial 90-day followup interval, the costs incurred for patients with ulcer sizes ⬎2.8 cm2 were almost twice as high as those incurred by patients with ulcer sizes ⭐2.8 cm2 (p ⬍ 0.05, Mann–Whitney U test).

Discussion This study demonstrated that VSU care entails substantial medical resource utilization and costs, especially in cases where time to healing is prolonged. The average cost per month of care was approximately $2400. Total cost per patient was distributed quite evenly during the follow-up interval, suggesting that reducing time to ulcer healing could result in a proportionate decrease in costs. Almost 50% of the total VSU costs were associated with home health care and one-quarter were due to hospitalizations for VSU care, yet only a minority of patients were hospitalized (18%) or received home health care (27%). The findings showed that females were more likely to use home health care services. Hospitalized patients were likely to have a previous history of VSU and were disproportionately black. However, the numbers of patients who used home health care and who were hospitalized in this study were small and thus it is difficult to reach conclusions about specific demographic characteristics that may be related to costly health services. Further research in this area may be helpful in efforts to reduce the costs of VSU management. A factor consistently found to be related to time to healing, as well as the recurrence of ulcers, is compliance.14–17 A prospective study performed by Erickson and colleagues showed that patients with VSU who complied with the treatment regimen had faster healing and fewer recurrent ulcers compared with less compliant patients.14 Another study showed that only 4% of patients who complied with stocking usage developed a recurrent ulcer compared with 79% who were non-compliant. The patients reported the major reason for non-compliance to be the cost of compression stockings.15 The present study selected patients who were compliant with scheduled office visits because the inclusion criteria required VSU visits to occur within 6 weeks of one another. Therefore, the patients selected may have healed faster, resulting in less utilization of medical services. The observed median time to healing of 77 days is consistent with Erickson’s study of a nurse-managed/physician-supervised treatment program that showed a median time to healing of 3 months for compliant patients.14 It is possible that the treatment costs for VSU have been overestimated because patients who have had failed treat-

Medical costs of treating venous stasis ulcers 5

Topical Medications < 1% Prescription Medications < 1%

Outpatient Visits 5%

Home Dressings 21%

Home Health Care 48%

Hospitalizations 25%

Figure 1 The distribution of total costs across the medical services.

$7000

$6000

$5736 $5088

$5108

Average Total Costs

$5000

$4000

$3000

$2602

$2000

$1000

$0 0-90 days n = 78

91-180 days n = 37

181-270 days n = 19

271-365 days n=9

Follow-up Interval

Figure 2 Average total costs incurred during 3-month follow-up intervals.

ment for their ulcers elsewhere may be treated at the Cleveland Clinic Foundation and therefore require more intensive and costly therapy. In contrast, the reported costs are underestimates in that they did not account for the significant indirect costs related to VSU: time absent from work, forced early retirement, loss of functional independence and other unquantifiable suffering.18–20 The study also did not include the costs associated with recurrent ulcers, which occur in many patients.14.15,17 The magnitude of the potential costs associated with non-healing ulcers argues for Vascular Medicine 1999; 4: 1–7

further research on how to improve compliance among VSU patients and facilitate access to compression devices that help prevent VSU. As demonstrated in a previous study by Hume and colleagues on the costs of Boston’s Visiting Nurse Association services for VSU care, compression was not routinely ordered and was not funded by the Health Care Financing Administration.7,21 The finding that home health care accounted for a significant proportion of the total VSU medical costs is consistent with results from Hume’s study that extrapolated

6 JW Olin et al

$8000 $7000

Ulcer size = or 2.8 cm2

$5860*

$6000 $5000

$4537

$4873

$4584

$4000

$3783 $3317

$3000 $2000 $1253

$1242

$1000 $0 0-90 days

n = 30

n = 29

91-180 days

n = 10

n = 18

181-270 days

271-365 days

n=5

n=2

n=8

n=4

Follow-up Interval

Figure 3 Average total costs incurred during 3-month follow-up intervals according to ulcer size. Note that the ulcer size is missing for 19 patients. (*p ⬍ 0.05, Mann–Whitney U test.)

Visiting Nurse Association costs to the entire population of the USA and yielded a home health cost estimate exceeding $16 million per month, reflecting a substantial burden on the Medicare budget.21 A similar study in the UK showed district nurse time to be the largest contributor to VSU medical costs for the National Health Service.22 Venous obstruction and/or valvular insufficiency, which may lead to VSU, can occur as a sequela of previous venous thrombosis. Several investigators have studied cohorts of patients who have experienced venous thrombosis. One such study, conducted recently by Bergqvist and colleagues in Sweden, examined the medical records of 257 patients with a history of deep venous thrombosis. Similar to the present study, the authors examined medical resource utilization data and costs. They estimated the average present value of the expected lifetime costs of treating postthrombotic complications. VSU occurred in 14% of the patients and were the most expensive type of complication at a cost of SEK 43 630 or US $7932 per patient.23 The present study examined the direct medical costs associated with VSU care in a USA clinical practice setting. Based on the course of medical resource utilization observed, minimizing the time to healing would help reduce overall VSU costs. Further research is needed to provide insights on ways to improve compliance and identify factors that are predictive of costly VSU management.

Acknowledgement This study was funded by an unrestricted grant from Novartis Pharmaceuticals Corporation.

References 1 Phillips TJ, Dover JS. Continuing medical education: leg ulcers. J Am Acad Dermatol 1991; 25: 965–87. 2 Browse NL, Buenard KG. The cause of venous ulceration. Lancet 1982; ii: 243–45.

Vascular Medicine 1999; 4: 1–7

3 Falanga V. Venous ulceration. Chronic Would Care Previews 1996; 8: 101–108. 4 Black SB. Venous stasis ulcers: a review. Ostomy/Wound Management 1995; 41(8): 20–32. 5 Burton CS. Venous ulcers. Am J Surg 1994; 167 (suppl): 37S–41S. 6 Margolis DJ, Cohen JH. Management of chronic venous leg ulcers: a literature-guided approach. Clin Dermatol 1994; 12: 19–26. 7 Hume M. Venous ulcers, the vascular surgeon and the Medicare budget. J Vasc Surg 1992; 16: 671–73. 8 Cameron J, Hoffman D, Poore S et al. A retrospective trial in the treatment of venous leg ulcers. Wounds 1996; 8: 95–100. 9 Kikta MJ, Schuler JJ, Meyer JP et al. A prospective, randomized trial of unna boots versus hydroactive dressing in the treatment of venous stasis ulcers. J Vasc Surg 1988; 7: 478–83. 10 Ohlsson P, Larsson K, Lindholm C, Moller M. A cost effective study of leg ulcer treatment in primary care. Comparison of saline-gauze and hydrocolloid treatment in a prospective randomized study. Scand J Prim Health Care 1994; 12: 295–99. 11 Thomas S. Cost-effective management of leg ulcers. Community Outlook 1990; Mar 14: 21–22. 12 Porter JM, Moneta GL. An international consensus committee on chronic venous disease. Reporting standards in venous disease: an update. J Vasc Surg 1995; 21: 635–45. 13 1997 Drug Topic’s Redbook. Montvale, Medical Economics Co. Inc., 1997. 14 Erikson CA, Lanza DJ, Karp DL et al. Healing of venous ulcers in an ambulatory care program: the role of chronic venous insufficiency and patient compliance. J Vasc Surg 1995; 22; 629–36. 15 Samson RH, Showalter DP. Stockings and the prevention of recurrent venous ulcers. Dermatol Surg 1996; 22: 373–76. 16 Ricotta JJ, Dalsing MC, Ouriel K et al. Research and clinical issues in chronic venous disease. Cardiovasc Surg 1997; 5: 343–49. 17 Mayberry JC, Moneta GL, Taylor LM, Porter JM. Fifteen-year results of ambulatory compression threapy for chronic venous ulcers. Surgery 1991; 109: 575–81. 18 O’Donnell TF, Brose NL, Burnard KG, Thomas ML. The socioeconomic effects of iliofemoral venous thrombosis. J Surg Res 1997; 22: 483–88. 19 Charles H. The impact of leg ulcers on patients’ quality of life. Prof Nurse 1995; 10(99): S71–S73. 20 Phillips T, Stanton B, Provan A, Lew R. A study of the impact of leg ulcers on quality of life: financial, social, and psychologic implications. J Am Acad Dermatol 1994; 31: 49–53.

Medical costs of treating venous stasis ulcers 7

21 Hume M, Basmajian A. Venous ulcers, the nurse and the health care budget. J Vasc Nurs 1993; 11: 23–24. 22 Bosanquet N. Cost of venous ulcers: from maintenance therapy to investment programmes. Phlebology 1992; 7 (suppl I): 44–46.

Vascular Medicine 1999; 4: 1–7

23 Bergqvist D, Jendteg S, Johansen L et al. Cost of long-term complications of deep venous thrombosis of the lower extremities: an analysis of a defined patient population in Sweden. Ann Intern Med 1997; 126: 454–57.