вход по аккаунту


Survival in systemic lupus erythematosus. a multivariate analysis of demographic factors

код для вставкиСкачать
A Multivariate Analysis of Demographic Factors
We analyzed survival rate and important clinical
outcomes in 411 patients with systemic lupus erythematosus who were seen at our center between 1969 and
1984. All eligible subjects met 4 of the revised American
Rheumatism Association criteria for systemic lupus
erythematosus and all were seen within 2 years of
diagnosis. Mean followup was 75.6 months. Multivariate analysis suggested significant independent effects of
race (P = 0.0139) and socioeconomic status (P =
0.0326) on survival. No evidence of diminished lupusrelated mortality with age was documented. Previously
reported findings of improved survival rate with age
may have been confounded by differences in race distribution between the younger- and older-onset groups.
Systemic lupus erythematosus (SLE) is predominantly a disease of young women, occurring more
frequently in blacks than whites (1,2), and less commonly in children (2,3), men (2,4), and older individuals (1-5). Controversy persists regarding predictors
of disease severity. A poor prognosis has been reported for blacks (6,7), men (3), and children (3), while
late-onset disease generally has been regarded as
benign ( 5 ) . No study of a patient series has yet at~
From the Division of Rheumatology & Immunology, Department of Medicine, and the Center for the Study of Aging and
Human Development, Duke University Medical Center, Durham,
North Carolina.
Supported in part by a Beverly EnterprisedDuke Geriatrics
Fellowship, Durham Veterans Administration Geriatrics Fellowship
Stephanie Studenski, MD; Nancy B. Allen, MD; David S.
Caldwell, MD; John R. Rice, MD; Richard P. Polisson, MD.
Address reprint requests to Stephanie Studenski, MD, Box
31082, Duke University Medical Center, Durham, NC 27710.
Submitted for publication October 2, 1986; accepted in
revised form May 25, 1987.
Arthritis and Rheumatism, Vol. 30, No. 12 (December 1987)
tempted to determine the independent effects of race,
sex, age, and economic status on survival.
This report describes the outcomes and prognoses by age, race, sex, and economic status of a large
series of patients with SLE. Modern methodologic
guidelines (8,9), which are essential for accurate epidemiologic and prognostic analyses, have been incorporated into this study. Most critical among these are
1) inclusion of only “new” cases for prognostication,
2) confirmation of outcome in as high a proportion of
subjects as possible, to minimize losses to followup, 3)
strict definitions of diagnostic and outcome criteria,
and 4) multivariate survival analysis.
Patients. Potential cases for study were identified
from 2 sources within the Duke University Medical Center:
computerized International Classification of Diseases Manual (ICDM) discharge code for SLE from January 1969 to
December 1983, and rheumatology outpatient clinic files.
For inclusion into the study, patients were required
to fulfill 4 or more of the American Rheumatism Association
(ARA) 1982 revised criteria for the diagnosis of SLE (10) and
to have had contact with our facility within 2 years after
initial diagnosis. In order to eliminate patients with syndromes caused by other diseases, some of the ARA revised
criteria were defined with added stringent specifications, as
follows: rashes-only classic photosensitive, malar, or discoid skin lesions were considered positive criteria; arthritis-required physician observation of at least 2 swollen
joints; serositis-in addition to a convincing history of pain,
objective findings by physical examination, chest radiograph, or sonogram were required; renal-only subjects
excreting cellular casts or >0.5 gm proteidday, or exhibiting
biopsy evidence of lupus nephritis were included; neurologic-evidence of seizures or psychosis without other cause
was required (other neurologic syndromes attributed to SLE
Table 1. Demographic characteristics of 41 1 patients with systemic lupus erythematosus
No. (%) age 55 or older
at diagnosis
Mean age at diagnosis
Mean months of followup
% lost to followup
% with high socioeconomic status*
White female
(n = 162)
White male
(n = 49)
Black female
(n = 178)
Black male
(n = 19)
(n = 3)
37 (22.8)
22 (44.9)
13 (7.3)
4 (21)
* See Patients and Methods for description of status determinations.
were not counted as an entry criterion); hemolytic anemiahematocrit <35% and reticulocyte count 2 4 % in the absence
of other cause was required. (Glucose-6-phosphatedehydrogennse deficiency and hemoglobinopathies were specifically
sought, and if they were detected, hemolytic anemia was not
counted as a criterion that was met.)
Patients were excluded from the study if they had
drug-induceddisease or Felty's syndrome (even if eligible by
ARA SLE criteria).
Renal failure was defined as creatinine clearance <I0
mhinute and/or disease requiring dialysis. Classificationas
an :SLE-related death required at least 1 of the following:
autopsy evidence of an SLE-related death; death attributed
to SLE by the clinicians who cared for the patient; lupusinduced, renal failure-related death; or death because of
active SLE of a major organ system.
Patients were referred to this center primarily from
North Carolina, Virginia, West Virginia, South Carolina,
Georgia, and Florida. For any patient who was not seen
within 1 year of the study, primary physicians were contacted for followup. In the case of patients still lost to
followup, state vital record registries were searched for
death certificates. The Motor Vehicle Administration,Social
Security Administration, and the National Death Index were
not searched.
Late-onset SLE was defined as disease diagnosed at
or after age 55. Insurance resources were used as a proxy
measure of economic status. Patients with private insurance
were classified as high socioeconomic status, while those
having no insurance, Medicare alone, or Medicaid were
classified as low socioeconomic status.
Statistical methods. Data were entered and simple
summary statistics were generated using SAS programs (1 1).
Simple comparisons of proportions employed a Z-test of
binomial proportions, while comparisons of means employed Student's t-tests. Univariate survival analysis employed the product-limit method (Kaplan-Meier) and the
genleralized Wilcoxon (Breslow) test available through
BMDP software (1 2). Multivariate survival analysis involved
the Cox proportional hazards model also available through
BMlDP (12).
Patient characteristics. Of the 770 potential
cases detected during the 15-year study period, 45
were excluded for the wrong ICDM code, 28 for
drug-induced disease, 14 for an unobtainable chart,
192 for inadequate fulfillment of SLE criteria or
Felty's syndrome, and 80 for presentation more than 2
years from diagnosis. The remaining 41 1 patients with
SLE were considered eligible for this study. Three
patients were young women (2 American Indian and 1
Polynesian) who were classified as "other race" and
included as nonwhite females in the survival analysis.
Sixty-eight were men (16.5%), 200 were nonwhites
(48.6%), and 76 were age 2 5 5 at diagnosis (18%)
(Table 1). Mean age at diagnosis was significantly
higher for white male patients than for white female
patients (P < 0.001). Almost one-half of white men
were diagnosed at or older than age 55, compared with
one-quarter or less of the other sex-race groups. Mean
duration of followup was 75.6 months for the group as
a whole, the longest being for white females and the
shortest for black males. Losses to followup occurred
in 13.6% overall. Utilizing insurance status as an
economic indicator, two-thirds of nonwhites were
defined as low socioeconomic status, compared with
one-fifth of white subjects.
Outcome and survival analysis. Important clinical outcome parameters (as defined previously), including renal failure and SLE-related death, were
significantly more prevalent in the nonwhite patient
subgroups (Table 2). Of the 3 women classified as
other race and not included in the table, 1 had renal
disease. None of the 3 died or had renal failure. Renal
disease (as defined above) was present in 60.3% of the
entire group and was more common in nonwhites
(70.6%) than whites (51.6%) (P < 0.0001). Renal
failure occurred in 13.6% of the entire group, more
frequently in males (19.1%) than in females (9.7%) ( P
< 0.0001) and in nonwhites (18.8%) more frequently
than in whites (9.0%) (P < 0.001). Death attributed to
SLE at any time during followup occurred in 19.2% of
the group overall, and, again, was more common in
blacks (25.9%) than in whites (13%) (P < 0.0001).
Adding the dimension of time to the event of
Table 2. Clinical outcomes in 408 patients with systemic lupus erythematosus (SLE)*
White female
(n = 162)
White male
(n = 49)
Black female
(n = 178)
Black male
(n = 19)
82 (51)
12 (7.4)
27 (55.1)
7 (14.3)
9 (18.4)
5 (10.2)
126 (71)
31 (17.4)
44 (24.7)
7 (3.9)
13 (68.4)
Renal diseaset
Renal failure
Deaths due to SLE
Deaths due to other causes
19 (11.7)
5 (3.1)
6 (31.6)
7 (36.8)
0 (0)
* See Patients and Methods for definitions of outcomes. Values are the number (%) of patients.
t As defined by American Rheumatism Association criteria.
SLE-related death, survival rate was evaluated univariately (Figure 1). No significant difference in survival
was delineated by age category (age <55 versus age
253,although a trend toward an increase in SLErelated mortality with older age was discernible ( P =
0.0925). A significant difference in survival by race (P
= 0.002), but no difference in survival by sex ( P =
0.2248), was documented.
The results of a Cox proportional hazards analysis, allowing for simultaneous control of the effects
of multiple factors on survival, are noted in Table 3.
The coefficient divided by the standard error corresponds to the Z-statistic, a measure of the normal
distribution. An absolute value 21.96 corresponds to a
P value less than 0.05. The sign on the coefficient
divided by the standard error points to the sequence of
categories within the variable. A positive sign indicates the second category (e.g., nonwhite race), while
a negative value indicates the first category (e.g., male
sex). The marginal effects of race, sex, age, and
insurance status were analyzed. Nonwhite race (P =
0.0139) and low socioeconomic status (P = 0.0326)
contributed independently to mortality. No significant
effect of sex (P = 0.0985) or age (P = 0.0833) on
survival was proven, although trends toward increasing mortality in males and older SLE patients were
Systemic lupus erythematosus, a clinical syndrome of varying severity, remains difficult to define.
We have analyzed the effects of age, race, sex, and
economic status on major outcomes and survival in
411 patients at a single university center, utilizing
modern standards of case assembly, outcome determination, and analytic technique. These methodologic
standards are of critical importance for accurate prognostication (8,9). Sackett et al have suggested that the
inclusion of only new cases of disease in survival
studies is the most critical element in assembling pa-
tients for prognostic studies (8). Patients referred later
in their disease course may be systematically different
from -patients whose diseases have been newly diagnosed. For example, inclusion of patients referred to a
tertiary care setting late in their disease course for
complex procedures, such as dialysis or renal transplantation, may skew results toward a poorer outcome
for the entire group. Conversely, longstanding cases of
mild disease might have the opposite effect on findings
of outcome and survival. The studies by Siege1 et al
(13), and Wilson et al(l4) and this investigation are the
only outcome and survival series to date that have
restricted entry to new cases of SLE. The time period
for eligibility as a new case of SLE has not been
uniformly agreed upon. We used “within 2 years of
diagnosis” to be consistent with the study by Wilson et
al (14) and to eliminate longstanding cases while maintaining an adequate sample size for analysis.
Case assembly also requires careful attention to
well-defined inclusion and exclusion criteria. The
ARA has established criteria for the classification of
SLE for just such a purpose (10). Some survivorship
studies have not been restricted to cases that were
eligible by these criteria. Interpretation of the criteria
remains variable as well, especially as they apply to
cutaneous manifestations, renal disease, and neurologic involvement. For example, in some studies,
generalized urticaria, elevated levels of serum creatinine or microscopic hematuria, and peripheral neuropathies have been deemed consistent with ARA criteria for SLE. These variations inevitably result in a
heterogeneity of disease that makes outcome analysis
difficult. Cases identified from a single tertiary care
institution may be systematically different from those
seen elsewhere, as demonstrated by the Multicenter
Lupus Survival Study Group (15). We acknowledge
the fact that the cases assembled in our patient series
may not reflect the clinical experience in other geographic areas or clinical practice settings.
Followup and outcome determination are other
areas of central importance. In general, the more
2 ,Ot
Followup (months 1
0-4 = M a l e
& 20
l3 0 0
Fo I I ow u p ( months )
1; 3
Figure 1. Cumulative proportional survival rates of 41 1 systemic lupus erythematosus
patients by age (A), race (3).and sex (C), from the time of diagnosis to lupus-related
death. Statistical analysis was done using the Kaplan-Meier method.
coirnplete the followup, the more reliable the results.
Sackett and others have suggested that lost-tofollowup rates >20% make interpretation of survival
data very difficult (8). While death as an event is
relatively easy to document, assigning the cause of
death is much more complex (9). Many survival stud-
Table 3. Cox proportional hazards analysis of independent effects of
demographic factors on survival in systemic lupus erythematosus
Race (white, nonwhite)
Sex (male, female)
Age ( 4 5 , 2 5 5 )
Socioeconomic status (low, high)
Coefficient + SEt
- 2.0974
* See Patients and Methods for description.
t Coefficient divided by SE corresponds to the Z-statistic; see
Results for explanation.
ies have included patients who have died from all
causes (whether lupus-related or not) (3,15), while
other studies have been restricted to patients whose
deaths were directly attributable to SLE (6,7). In our
study, we have defined explicit criteria for lupusrelated death.
Multivariate techniques are important in the
determination of the independent effects of multiple
factors in complex groups and have only been used in
a few recent studies (16). We have utilized a Cox
proportional hazards model to analyze the effects of
demographic factors on survival. This approach is
feasible, efficient, and has methodologic advantages
over population matching.
Our analysis suggests significant and independent effects of both race and economic status on
survival in SLE. Nationwide mortality rates from
SLE, as determined from death certificates, were
higher for blacks than whites in 2 analyses: 1 on deaths
between 1968 and 1972 and 1 on deaths between 1972
and 1976 (7). These studies, which were not influenced
by geographic variation, suffered from lack of a denominator incidence rate, as well as from the known
weaknesses of death certificate data. In the earlier
study (6), Kaslow and Masi attempted to correct for
socioeconomic status by adjusting for overall raceand sex-specific mortality rates. Significantly higher
mortality for black patients persisted.
Siege1 and coworkers, in an early populationbased study of new cases in New York City, had a
very low lost-to-followup rate (13). They found high
rates of incidence and mortality in young black males,
although survivorship at 5 years was not significantly
different from that of white females. In another early
study of 150 patients, Estes and Christian (17) found
no statistically significant difference in survival, by
race. 'This investigation was not restricted to new
cases and the lost-to-followup rate was high. In a large
series with excellent followup rates, Wallace and
collaborators found similar survival rates in whites and
nonwhites (3). This study, which was not restricted to
new cases, included only 6% blacks (Hispanics, orientals, and blacks were pooled into a nonwhite category
making up 21% of the total group.)
The Lupus Survival Study Group investigated
survival rates in 1,103 patients observed at 9 university
centers (15). Although that study had great strengths in
sample size and center diversity, the findings had a
very high lost-to-followup rate (28%) (18), and included patients who had been diagnosed as having
SLE many years prior to referral. The investigators
found evidence of poorer survival in indigent patients,
many of whom were black. A stratified analysis indicated that this appeared to be an effect of economic
status, rather than of race. A recent abstract by
Reveille and Alarcon described survival in 388 patients
with SLE (19). An excellent followup rate was
achieved, but no information was provided about the
inclusion of only new cases or about statistical analysis. The authors stated that poorer survival was affected both by black race and lower socioeconomic
Given all the variations in the assembling of
patient series and outcome definitions, generalizations
have been difficult to make. An effect of socioeconomic status is likely and is not particularly surprising.
Our data also implicate an independent race effect on
outcome and survival. Renal disease occurred more
commonly in blacks than in whites in our study. A
previous investigation has implicated renal disease as
an independent predictor of survival (3). The higher
frequency of renal disease noted in our black lupus
patients may have contributed to the poorer survival
outcome in this group.
Our data support a trend of borderline significance toward poorer survival in males than females.
The studies involving national mortality statistics (6,7)
found death rates approximately 4 times higher in
females than in males. If a fema1e:male incidence rate
of 6: 1 existed at that time as it does now, according to
Hochberg (2), actual fatality rates would be higher in
males. Estes and Christian found no statistical difference in survival by sex (17). Wallace et al found a
statistically significant difference in survival by sex,
with males faring more poorly (3). The overall effect of
sex on survival in SLE remains unresolved, although
poorer survival in males seems possible.
Age has generally been implicated as an independent factor in survival. SLE has been perceived as
a benign disease in older individuals (5,14,20-22). The
studies based on national mortality statistics found
peak mortality rates in black women in the third
through fifth decades, in contrast to the gradual increase in lupus-specific mortality with increasing age
in men and white women (6,7). These effects are
unlikely to be attributable to an increase in incidence
with age (2). Estes and Christian found no difference in
survival by age through the fifth decade (17). Wallace
and coworkers (3) compared 5- and 10-year survival
rates in 3 age groups: <16 years, 16-50 years, and >50
years at diagnosis. Five-year mortality rates appeared
similar for all groups, except for those individuals with
chilldhood-onset lupus nephritis. In all lupus patients
with nephritis, 10-year survival with increasing age
groups was 58%, 68%, and 48%, respectively, and in
patients without nephritis it was loo%, 88%, and 81%,
reslpectively. Reveille and Alarcon (19) also defined 3
age groups: <20 years, 20-49 years, and 250 years;
5-year survival was 85.5%, 93.3%, and 83.1%, respectively. The authors suggested that the youngest group
had a poorer survival rate, but no comment was made
about the 2 5 0 subset. Our observations defined no
significant effect of age in a multivariate analysis,
although there was a trend toward poorer survival with
increasing age.
Most of the studies of late-onset disease and
survival have pooled all older-onset patients in the
analysis (3,14,20). Ballou et al(23) attempted to stratify
their findings by age and race, but found that the small
number of older black patients prevented meaningful
analysis. Our findings suggest that racial differences in
age at onset and mortality could have created artificial
evidence of improved outcome with age. If younger
g r w p s were heavily weighted with black individuals
and older groups were weighted with whites, age would
appear, incorrectly, to be related to improved survival
in ]pooled groups. Only multivariate analytic techniques
could dissect these confounding effects.
In summary, our analysis has shown that race
anld socioeconomic status have independent negative
effects on survival. No evidence of a significant differenlce in outcome by sex o r age was found. Previously
reported findings of improved survival with age may
have been artificial due to differences in racial predominance among the groups.
1. Fessel WJ: Systemic lupus erythematosus in the community. Arch Intern Med 134:1027-1035, 1974
2. Hochberg MC: The incidence of systemic lupus erythematosus in Baltimore, Maryland, 1970-1977. Arthritis
Rheum 28:80-86, 1985
3. Wallace DJ, Podell T, Weiner J, Klinenberg JR, Forouzesh S, Dubois EL: Systemic lupus erythematosus:
survival patterns. JAMA 245:934-938, 1981
4. Masi AT, Kaslow RA: Sex effects in systemic lupus
erythematosus: a clue to pathogenesis. Arthritis Rheum
21:480-484, 1978
5. Baker SB, Rovira JR, Campion EW, Mills JA: Late
onset systemic lupus erythematosus. Am J Med
66:727-73 1, 1979
6. Kaslow RA, Masi AT: Age, sex, and race effects on
mortality from systemic lupus erythematosus in the
United States. Arthritis Rheum 21:473-479, 1978
7. Gordon MF, Stolley PD, Schinnar R: Trends in recent
systemic lupus erythematosus mortality rates. Arthritis
Rheum 24:762-769, 1981
8. Sackett DL, Haynes RB, Tugwell P: Clinical Epidemiology. Boston, Little, Brown, 1981, pp 159-169
9. Feinstein AR: Clinical Epidemiology. Philadelphia, WB
Saunders, 1985, pp 581-587
10. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ,
Rothfield NF, Schaller JG, Tala1 N, Winchester RJ: The
1982 revised criteria for the classification of systemic
lupus erythematosus. Arthritis Rheum 25: 1271-1277,
1 1 . SAS User’s Guide: Basics. Version 5. Cary, NC, SAS
Institute, Inc., 1985
12. BMDP Statistical Software. Berkeley, University of
California Press, 1983
13. Siege1 M, Gwon N , Lee SL, Rivero I, Wong W: Survivorship in systemic lupus erythematosus: relationship to
race and pregnancy. Arthritis Rheum 12:117-125, 1969
14. Wilson HA, Hamilton ME, Spyker DA, Brunner CM,
The authors thank Karen Cash and Judy Reich for
their secretarial support and Alice Castillo and Mary Lou
Bembe for laboratory assistance.
O’Brien WM, Davis JS IV, Winfield JB: Age influences
the clinical and serological expression of systemic lupus
erythematosus. Arthritis Rheum 24: 1230-1235, 1981
Ginzler EM, Diamond HS, Weiner M, Schlesinger M,
Fries JF, Wasner C, Medsger TA Jr, Ziegler G, Klippel
JH, Hadler NM, Albert DA, Hess EV, Spencer-Green
G, Grayzel A, Worth D, Hahn BH, Barnett EV: A
multicenter study of outcome in systemic lupus erythematosus. I. Entry variables as predictors of prognosis.
Arthritis Rheum 25:601-611, 1982
Masi AT: Clinical epidemiologicperspective of systemic
lupus erythematosus, Epidemiology of the Rheumatic
Diseases. Edited by RC Lawrence, LE Shulman. New
York, Gower Medical Publishing, 1984, pp 145-163
Estes D, Christian CL: The natural history of systemic
lupus erythematosus by prospective analysis. Medicine
(Baltimore) 50:85-95, 1971
Larson MG, Liang MH: The multicenter study of outcome in systemic lupus erythematosus: a critique
(letter). Arthritis Rheum 26570-572, 1983
19. Reveille JD, Alarc6n GS: Survival in systemic lupus
erythematosus: demographic, socioeconomic, clinical
factors, and causes of death (abstract). Arthritis Rheum
(suppl) 29(4):S27, 1986
20. Dimant J, Ginzler EM, Schlesinger M, Diamond HS,
Kaplan D: Systemic lupus erythematosus in the older
age group: computer analysis. J Am Geriatr SOC275861, 1979
21. Foad BSI, Sheon RP, Kirsner AB: Systemic lupus
erythematosus in the elderly. Arch Intern Med 130:743746, 1972
22. Catoggio LF, Skinner RP, Smith G, Maddison PS:
Systemic lupus erythematosus in the elderly: clinical
and serological characteristics. J Rheumatol 11:175-181,
23. Ballou SP, Khan MA, Kushner I: Clinical features of
systemic lupus erythematosus: differences related to
race and age of onset. Arthritis Rheum 2555-60, 1982
Без категории
Размер файла
577 Кб
factors, lupus, multivariate, survival, systemic, erythematosus, analysis, demographic
Пожаловаться на содержимое документа