Arthritis & Rheumatism (Arthritis Care & Research)

Vol. 49, No. 2, April 15, 2003, pp 216 220

DOI 10.1002/art.10998
2003, American College of Rheumatology

ORIGINAL ARTICLE

Declining Use of Orthopedic Surgery in

Patients With Rheumatoid Arthritis? Results of a
Long-Term, Population-Based Assessment
ELEONORA DA SILVA,1 MICHELE F. DORAN,2 CYNTHIA S. CROWSON,2 W. MICHAEL OFALLON,2
2
AND ERIC L. MATTESON

Objective. To describe the use of orthopedic surgery, including joint replacement surgery, in a well-defined, populationbased cohort of patients with rheumatoid arthritis (RA) and to identify characteristics that predict such use.
Methods. A retrospective medical record review was performed of cases of RA incident in Rochester, Minnesota, during
the years 19551995. All joint surgeries were recorded.
Results. Of the total 609 RA incident cases, 242 patients underwent 1 or more (maximum of 20/patient) surgical
procedures involving joints during their followup. Overall, this RA cohort had 7.4 surgeries per 100 person-years of
followup; the cumulative incidence for joint surgery for RA-related joint disease at 30 years was 33.7% SEM 3.8%. The
risk of having a disease-related joint surgery for RA is increased in patients who are women, younger, positive for
rheumatoid factor, and have rheumatoid nodules. When adjusted for duration of followup, patients with RA diagnosed
after 1985 were significantly less likely to have undergone joint surgery for RA (P < 0.001). Survival of patients who
underwent total joint arthroplasty was similar to those who did not.
Conclusion. Reconstructive surgeries are common in RA, although patients diagnosed after 1985 are less likely to require
joint surgery. These findings may reflect trends in medical disease management and have importance for health care
resource utilization planning.
KEY WORDS. Rheumatoid arthritis; Epidemiology; Orthopedic joint surgery.

INTRODUCTION
Rheumatoid arthritis (RA) is a common inflammatory joint
disease that affects about 1% of the population and up to
3% of the population over the age of 65 years (1). The
disease is an immune-mediated process that affects the
synovial and lined appendicular joints as well as the atlantoaxial joint of the cervical spine. It is associated with
marked disability and decreased life expectancy (2 4).
The need for orthopedic surgery is considered a marker of
disease severity (5). Orthopedic procedures, including
joint replacement surgeries, have substantially improved
the overall function and quality of life of patients with RA
1
Eleonora da Silva, MD: Universidade Federale de Sao
Paulo, Brazil; 2Michele F. Doran, MD, Cynthia S. Crowson,
BS, W. Michael OFallon, PhD, Eric L. Matteson, MD, MPH:
Mayo Clinic and Mayo Foundation, Rochester, Minnesota.
Address correspondence to Eric L. Matteson, MD, MPH,
Division of Rheumatology, Mayo Clinic and Mayo Foundation, Rochester, MN 55905. E-mail: matteson.eric@mayo.
edu.
Submitted for publication April 25, 2002; accepted in
revised form June 29, 2002.

216

(5,6). At the same time, they are a major factor in the high
medical cost for patients with this disease (4). It is unknown how use of orthopedic surgery has changed since
the introduction of modern joint replacement surgical
techniques and more aggressive management of RA in the
past 2 decades. The aim of this study was to describe the
use of orthopedic surgery, including joint replacement
surgery, in a population-based incidence cohort of patients
with RA in Rochester, Minnesota. Knowledge of diseaseand patient-associated risk factors for joint surgery provides important insights into the disease course and its
impact in affected patients, with potential consequences
for health care resource utilization planning.

PATIENTS AND METHODS

Patients. This study is based on a cohort of 609 patients
with RA incident in Rochester between January 1, 1955
and January 1, 1995. The cohort was assembled using the
facilities of the Rochester Epidemiology Project. All residents of Rochester, Minnesota aged 18 years with a diagnosis of RA based on the 1987 American College of

Joint Surgery for Rheumatoid Arthritis

217

Table 1. Orthopedic surgeries performed in 609 patients with incident rheumatoid arthritis in Rochester, Minnesota

Type of surgeries

Patients having
surgery total;
male/female

RA disease related
Primary total joint arthroplasty
Joint reconstructive procedures
Soft tissue procedures
Trauma and other surgeries
Total joint arthroplasty
Septic joint arthrotomy
Open fracture reduction
Revision of total joint arthroplasty
Total: any surgeries

174; 36/138*
85; 17/68
84; 8/76*
108; 21/87
60; 6/54*
37; 3/34*
10; 2/8
64; 17/47
20; 2/18
242; 55/187

Cumulative incidence at 30 years % SEM

Total
33.7 3.8
17.8 4.0
17.0 3.0
19.8 2.9
12.4 3.4
7.7 2.9
1.7 0.8
12.2 2.5
4.3 2.3
47.0 3.7

Male

Female

25.9 8.8
13.9 10.3
6.7 6.2
14.4 7.0
4.8 3.6
2.3 2.4
1.6 2.1
10.8 3.2
1.8 2.7
38.3 7.9

36.7 4.0
19.4 4.1
20.8 3.3
10.8 3.0
15.2 4.3
9.6 3.6
1.8 0.8
12.7 3.1
5.2 2.9
50.4 4.1

* Significantly more frequent in women than in men.

Rheumatology (ACR; formerly American Rheumatism Association) classification criteria for RA were identified by
searching the computerized diagnostic index for any diagnosis of arthritis (excluding degenerative arthritis or osteoarthritis) (79). The complete (inpatient and outpatient
care by any provider) medical record was reviewed by a
trained nurse abstractor using a pretested data collection
form. All cases were followed longitudinally until December 31, 1998 or until death or migration out of the county.
Description of orthopedic surgeries. The entire medical
record of each patient was reviewed to obtain data on
orthopedic procedures. The date of each procedure was
registered. Surgeries occurring before the incident date of
the diagnosis of RA were excluded. Joint surgery data were
collected for the following joints: right and left temporomandibular joints, shoulders, elbows, wrists, thumbs
(base, metacarpophalangeal joint [MCP], interphalangeal
joint [IP]), other fingers (MCP, proximal interphalangeal
joint [PIP], distal interphalangeal joint [DIP]), hips, knees,
ankles, first toes (metatarsophalangeal joint [MTP] and IP),
MTP 25, toes 25 (PIP, DIP), and cervical spine.
Procedures were recorded as 1) implant arthroplasty
with a total joint arthroplasty (TJA) or hemiarthroplasty
with an implant component; 2) nonimplant arthroplasty:
resection arthroplasty (such as Keller bunionectomy), fusion, wire arthrodesis, Girdlestone procedure, osteoectomy, or osteotomy; 3) soft tissue procedures (STPs): synovectomies, tendon repairs, tendon transfers, tendon
releases, ligament releases (such as carpal tunnel release),
cartilage repair, or meniscus repair; 4) fractures with a
nonimplant surgical repair (plate and pinning), or amputations at a joint level; 5) fractures repaired with an implant TJA; 6) arthrotomy for a septic native joint; 7) arthrotomy for a septic prosthetic joint; 8) diagnostic
arthroscopy or biopsy; 9) revision surgery of an implant
prosthesis; 10) revision surgery for the other procedures;
and 11) cervical spine atlantoaxial fusion or multiple-level
fusions performed for RA-related cervical spine disease.
Data analysis and measurements. The demographic
and disease-specific characteristics of the population were
assessed and the orthopedic surgeries each patient under-

went were recorded. The cumulative incidence of any

orthopedic surgery following the incident diagnosis of RA
was estimated using the method of Gooley et al to account
for the competing risk of death (10). Similarly, the incidence of surgery of specific sites was estimated for those
sites. The Cox proportional hazards model was used to
determine the influence of sex, age, rheumatoid factor
positivity (titer 1:40), rheumatoid nodulosis, and decade
of diagnosis of RA on the rate of first surgery. To avoid
possible bias due to different length of followup and cohort effect, we truncated followup comparably for each
decade cohort. Because the 19851994 cohort was followed until December 31, 1998, we truncated followup of
the other 3 decades (19551964, 19651974, and 1975
1984) at December 31, 1968, December 31, 1978, and December 31, 1988, respectively. The proportional hazards
model was also used to assess the influence of joint surgeries on survival.

RESULTS
Between 1955 and 1995, there were 609 Rochester residents aged 18 years who fulfilled the 1987 ACR criteria
for RA, forming the incident RA cohort (9). Of these patients, 26.9% (164) were male and 73.1% (445) were female. Rheumatoid factor was positive in 341 (56.0%) of
the patients, and 148 (24.3%) patients developed rheumatoid nodules. The average age at diagnosis was 58.0 years.
The average age at first surgery for RA-related joint disease
was 58.9 years (range 28.9 85.1 years). The mean followup was 13.9 years (range 0.0 40.9 years).
A total of 242 (39.7%) patients underwent 1 or more
(maximum of 29) surgical procedures involving joints, of
which 174 (28.6%) were directly related to RA (that is, not
related to trauma, sepsis, or other causes). The cumulative
incidence at 30 years for surgical procedures for RA was
33.3% SEM 3.8%. The number and types of surgeries
are contained in Table 1. Primary TJA for RA was performed in 85 patients; other joint reconstructive procedures (JRP) were performed in 84 patients, and STPs for
RA-related joint disease were done in 108 patients. Only 2
patients underwent cervical spine fusion for RA-related
disease. Trauma and other joint surgeries performed in 60

218

da Silva et al

Table 2. Types and frequencies of orthopedic procedures performed for rheumatoid arthritis-related joint disease*
Number of surgeries

Type of
surgery

Total
number of
patients

Total
number of
procedures

1
n (%)

2
n (%)

3
n (%)

4
n (%)

5 or more
n (%)

TJA
JRP
STP

85
84
108

169
239
216

39 (45.9)
29 (34.5)
57 (25.0)

30 (35.3)
23 (27.4)
27 (25.0)

6 (7.0)
6 (7.1)
7 (6.5)

3 (3.5)
9 (10.7)
8 (7.4)

7 (8.2)
17 (20.2)
9 (8.3)

* TJA primary total joint arthroplasty; JRP non-TJA joint reconstructive procedures; STP soft tissue procedures.

patients (30 year cumulative incidence 12.4% SEM

3.4%) included TJA for fractures in 37 patients, arthrotomy for septic arthritis in 10 patients, and revision of a
previous TJA in 20 patients (Table 1).
The median number of RA-related procedures per patient in the 174 patients who had surgery was 2 (minimum
of 1/patient and maximum of 20/patient). Many patients
underwent multiple types of surgeries, and many patients
had multiples of each type of surgery. For example, 85
patients had primary TJA; of these, 39 (45.9%) had only 1
primary TJA, 30 (35.3%) had 2 TJAs, and 6 (7%) had 3
TJAs. The results for primary TJA, JRP, and STP are shown
in Table 2. The hip was the most frequently operated joint
in this cohort (Table 3). For surgery related to RA itself, the
knee joints were the most frequently operated, followed by
the joints of the wrist, first toes, and fingers (Table 4).
In univariate Cox proportional hazards models, the risk
of having a disease-related joint surgery for RA was increased in patients who are younger (relative risk [RR]
0.84, 95% confidence interval [95% CI] 0.76 0.93 per 10
year increase in age at diagnosis of RA, P 0.001), in those
who are positive for rheumatoid factor (RR 1.73, 95% CI
1.272.37, P 0.001), and in those with subcutaneous
nodules (RR 2.84, 95% CI 2.00 4.03, P 0.001). Multivariate analyses did not provide different results. There
was a borderline increase in the time to first joint surgery

in women (RR 1.47, 95% CI 1.022.12, P 0.04). However, women had significantly more joint surgeries than
men, with a rate of 8.6/100 RA-related surgeries per person-year; for men the rate was of 4.0/100 per person-year
(P 0.0001).
When evaluated as a continuous variable, there was no
effect of year of diagnosis of RA on the risk for joint surgery
(P 0.19). However, when evaluated by decade of diagnosis of RA, there was a clear reduction in the risk for all
types of joint-related surgery (P 0.0001) and for each
subgroup considered separately (TJA, P 0.001; JRP, P
0.002; STP, P 0.01) for the cohort of patients diagnosed
19851994. For the decade 19551964, there were slightly
fewer surgeries (likely reflecting the fact that TJA became
more feasible for technical reasons during the mid-1960s);
and for the decade 19851994, a markedly lower risk was
seen for an RA-related joint surgery than for the patients
diagnosed with RA during 19651974 and 19751984 (P
0.0001; Figure 1).
At some point during the followup period, most patients
had taken a disease-modifying antirheumatic drug
(DMARD). However, we could not with certainty determine whether the use of a DMARD, or any specific
DMARD, affected the risk for having orthopedic surgery in
this study (data not shown).
There were 333 deaths during the followup period. The

Table 3. Joints involved in orthopedic surgery in 609 patients with rheumatoid arthritis

Joints

Patients having surgery,

by joint total;
male/female

Total

Male

Female

Hips
Knees
Wrists
First toes (MTP, IP)
Other fingers (MCP, PIP, DIP)
Feet, MTPs 25
Thumbs (base, MCP, IP)
Toes 25 (PIP, DIP)
Elbows
Ankles
Shoulders
Cervical spine fusion
Temporomandibular

84; 12/72*
81; 23/58
75; 13/62
48; 4/44*
56; 11/45
44; 5/39
42; 4/38*
27; 2/25*
16; 3/13
9; 1/8
12; 4/8
2; 0/2
1; 0/1

17.7 3.8
16.0 3.0
14.2 2.7
10.0 2.6
11.3 3.2
9.6 3.1
8.5 3.0
6.1 2.3
2.9 1.2
1.7 1.0
2.5 1.6
0.4 0.4
0.2 0.2

7.6 2.9
16.4 7.0
9.4 5.2
3.7 5.7
9.9 9.6
4.3 5.7
4.5 9.3
1.2 1.0
1.9 3.0
0.7 0.9
2.7 1.7
0.0
0.0

21.5 4.8
16.0 3.3
16.1 3.2
12.4 2.9
11.8 2.5
11.5 3.6
10.0 2.6
7.9 3.0
3.3 1.4
2.0 1.3
2.5 2.1
0.5 0.5
0.2 0.3

Cumulative incidence at 30 years % SEM

* Significantly more frequent in women than in men.

Number of surgeries. Patients may for example have had from 1 to 5 MCP joints operated on in a single surgery session, but each surgery session is
counted only once for the respective joint group. MTP metatarsophalangeal; IP interphalangeal; PIP proximal interphalangeal; DIP distal
interphalangeal; MCP metacarpophalangeal.

Joint Surgery for Rheumatoid Arthritis

219

Table 4. Joints involved in orthopedic surgery for rheumatoid arthritis-related joint disease

Joints

Patients having surgery,

by joint total;
male/female

Total

Male

Female

Hips
Knees
Wrists
First toes (MTP, IP)
Other fingers (MCP, PIP, DIP)
Feet, MTPs 25
Thumbs (base, MCP, IP)
Toes 25 (PIP, DIP)
Elbows
Ankles
Shoulders
Cervical spine fusion
Temporomandibular

26; 4/22
63; 16/47
59; 9/50*
44; 3/41*
40; 7/33
35; 4/31*
42; 4/38*
25; 1/24*
14; 1/13
4; 0/4
11; 3/8
1; 0/1
1; 0/1

5.5 2.5
13.1 3.1
11.5 2.7
9.1 2.4
8.1 2.2
7.7 2.7
8.5 3.0
5.8 2.3
2.7 1.2
0.9 0.9
2.4 1.6
0.2 0.2
0.2 0.2

2.0 1.3
12.0 7.2
6.9 5.2
3.1 5.7
5.8 5.3
3.7 5.7
4.5 9.3
0.6 0.7
1.0 2.3
0.0
2.1 1.6
0.0
0.0

6.8 3.3
13.5 3.4
13.2 3.2
11.3 2.6
8.9 2.4
9.2 3.1
10.0 2.6
7.7 3.0
3.3 1.4
1.3 1.2
2.5 2.1
0.3 0.3
0.3 0.3

Cumulative incidence at 30 years % SEM

* Significantly more frequent in women than in men.

See Table 3 footnote.

survival of patients undergoing RA-related joint surgery

was similar to those who did not.

DISCUSSION
Reconstructive surgeries for RA are common, with 28.6%
of 609 RA patients having undergone such a procedure.
The cumulative incidence at 30 years was 33.7%. Soft
tissue and joint reconstructive procedures were most often
performed, but primary TJAs were also frequent, performed in 85 of 609 patients with a cumulative incidence
at 30 years of 17.8%. Surgery of the knee for RA-related
disease was more frequent than any other joint or group of
joints (multiple MCP or MTP joints operated on during 1
operative session considered together).
The likelihood of surgery was higher in patients who
were younger at disease diagnosis; there was a 16% decrease in risk for every 10 year increase in age at incidence.
A woman with RA was 1.5 times as likely to have an
orthopedic surgery than was a man, a finding noted by

Figure 1. Cumulative incidence of surgery (% of patients with

surgery) by decade, January 1, 1955January 1, 1995.

other authors (11). The risk for joint surgery was 70%
higher for those with rheumatoid factor positivity and was
2.8 times higher in patients with subcutaneous nodules
than those without. If the requirement for joint surgery can
be viewed as a marker of severe disease and poor outcome
(at least with respect to joint integrity), these findings
would seem to confirm other epidemiologic observations
of younger age at disease onset, female sex, and rheumatoid factor positivity as markers of disease severity (12).
Certainly the decision to undergo joint replacement surgery is complex and not simply a reflection of joint damage
and attendant loss of function. Women had more surgeries
than men in absolute and relative terms, with increased
utilization of small joint (hand and foot) surgery in women
accounting for the major part of this difference. We did not
examine psychosocial factors including occupational, educational, or marital status, or outcome of previous joint
surgery as predictors of the need for joint surgery. Whether
this increased use of surgery by women reflects functional,
cosmetic, pain perception, greater disease extent, or in the
case of previous surgery, greater satisfaction with functional and cosmetic outcomes is uncertain. It is clear that
there are significant sex differences in the rates of utilization of joint surgery.
There are no directly comparable studies of joint surgery
for patients with RA. In another study of 1,600 patients
with RA, 25% underwent TJA within 21.8 years of disease
onset, whereas in the present study we found that 122 of
609 had TJA (20.3%; 85 patients with primary TJA and 37
patients with TJA due to trauma) (13). A cross-sectional
epidemiologic study of 1,629 patients with RA seen in a
private practice in France during a 6-month period in 1996
revealed that 24% of patients had 1 or more surgical procedures for joint disease during their disease course (mean
disease course, 8 years; mean number of procedures, 3), a
figure somewhat lower than that in our study (14). Referral
and temporal trends in disease management may explain
these differences.
We attempted to evaluate whether the use of DMARDs

220
or glucocorticoids had an influence on the risk for orthopedic joint surgery. There were important methodologic
limitations that complicated this analysis, including the
fact that this was a retrospective study not designed to
address this question, and that indication and disease severity probably confounded the results regarding use of
these drugs. We could not with certainty determine
whether glucocorticoid or DMARD use, including individual DMARDs, had any influence on the risk for joint surgery.
There was a clear difference in the utilization of and risk
for having joint surgery in patients with RA diagnosed in
the decade after 1985. This may well reflect improvements
in disease management, and could be an indication that
modern therapies for RA are having a positive impact on
important measures of outcome. This trend could reflect a
change in the nature of the disease, possibly that it has
milder expression in more recent years, or could reflect a
change in rheumatologic and orthopedic practice patterns.
Certainly enthusiasm for hip replacement and knee replacement surgeries increased in the 1970s as surgical
techniques and prosthetic materials improved, but the
overall trend toward less surgery in more recent years is
noteworthy. We did not specifically evaluate the presence
of risk factors for osteoporotic and other fractures necessitating joint surgery in these patients with RA; this has
been the subject of other reports (1517).
Our results suggest that survivorship among RA patients
undergoing surgery was similar to the RA population at
large. There are no comparable studies, although in a case
series of nonambulatory Japanese patients with RA undergoing total knee arthroplasty or total hip arthroplasty, only
8 were alive after 10 years of followup (18).
From this and other studies it is evident that patients
with RA have high rates of utilization of orthopedic services. That these rates are higher than expected from the
general population is reflected in a study of total hip
arthroplasties done in Rochester, Minnesota between 1969
and 1990: 63.2% of these procedures were for degenerative
joint disease, 19.8% for fractures, and 19.8% for RA (19).
Our longitudinal, population-based assessment data support and extend these findings.
Because of the demographic make up of the community
(more than 90% white), estimates of the need for joint
surgery in nonwhites would be unstable; however, the
majority of RA cases nationally are among whites. We
cannot demonstrate that the medical and surgical practice
in Rochester is optimal, but most (98%) surgeries were
performed at Mayo Clinic. Although the generalizability of
our findings is limited, utilization rates from the above
mentioned study of total hip arthroplasty in Rochester
were consistent with European data and slightly higher
than in other studies from the US (19).
Although it is uncertain how the need for joint surgeries
in patients with RA will be affected by changes in disease
therapies, it is clear that the need for such services will

da Silva et al
remain high in the foreseeable future, with important impacts on patient quality of life and health care resource
utilization planning. Patients recently diagnosed with RA
who are treated according to modern therapeutic concepts
may require less disease-related joint surgery than their
predecessors.

REFERENCES
1. Silman AJ, Hochberg MC. Epidemiology of the rheumatic
diseases. Oxford: Oxford University Press; 1993.
2. Symmons DP. Mortality in rheumatoid arthritis. Br J Rheumatol 1988;27 Suppl 1:44 54.
3. Gabriel SE, Crowson CS, OFallon WM. Mortality in rheumatoid arthritis: have we made an impact in 4 decades? J Rheumatol 1999;26:2529 33.
4. Mitchell DM, Spitz PW, Young DY, Bloch DA, McShane DJ,
Fries JF. Survival, prognosis, and causes of death in rheumatoid arthritis. Arthritis Rheum 1986;29:706 14.
5. Anderson R. The orthopedic management of rheumatoid arthritis. Arthritis Care Res 1996;9:23 8.
6. Clayton ML, Smythe CJ. Surgery for rheumatoid arthritis. New
York: Churchill Livingstone; 1992.
7. Gabriel SE, Crowson CS, OFallon WM. The epidemiology of
rheumatoid arthritis in Rochester, MN, 19551985. Arthritis
Rheum 1999;42:41520.
8. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,
Cooper NS, et al. The American Rheumatism Association
1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:31524.
9. Doran MF, Pond GR, Crowson CS, OFallon WF, Gabriel SE.
Trends in incidence and mortality in rheumatoid arthritis in
Rochester, Minnesota over a 40-year period. Arthritis Rheum
2002;46:62531.
10. Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of
failure probabilities in the presence of competing risks. Stat
Med 1999;18:695706.
11. Weyand CM, Schmidt D, Wagner U, Goronzy JJ. The influence
of sex on the phenotype of rheumatoid arthritis. Arthritis
Rheum 1998;41:81722.
12. Jacoby RK, Jayson MI, Cosh JA. Onset, early stages, and prognosis of rheumatoid arthritis. BMJ 1993;2:96 100.
13. Wolfe F, Zwillich S. The long-term outcome of rheumatoid
arthritis: a 23-year prospective, longitudinal, study of total
joint replacement and its predictors in 1,600 patients with
RA. Arthritis Rheum 1998;41:1072 82.
14. Sany J, Dropsy R, Daures J-P. A cross-sectional epidemiological survey of rheumatoid arthritis patients seen in private
practice in France: descriptive results (1629 cases). Rev Rhum
[Engl Ed] 1998;65:46270.
15. Hooyman JR, Melton LJ III, Nelson AM, OFallon, WM , Riggs
BL. Fractures after rheumatoid arthritis: a population-based
study. Arthritis Rheum 1984;2712:1353 61.
16. Cooper C, Coupland C, Mitchel M. Rheumatoid arthritis, corticosteroid therapy and hip fracture. Ann Rheum Dis 1995;
54:49 52.
17. Michel B, Bloch D, Wolfe F, Fries JE. Fractures in rheumatoid
arthritis: an evaluation of associated risk factors. J Rheumatol
1993;20:1666 9.
18. Yoshino S, Shiga H, Nakamura H, Nagashima M. Long term
outcome of total joint arthroplasty in nonambulatory patients
with rheumatoid arthritis. J Rheumatol 1999;26:1076 9.
19. Madhok RD, Lewallen DG, Wallrichs SL, Ilstrup DM, Kurland
RL, Melton LJ III. Trends in the utilization of primary total hip
arthroplasty, 1969 through 1990: a population-based study in
Olmsted County, Minnesota. Mayo Clin Proc 1993;68:11 8.