вход по аккаунту


The independence and stability of socioeconomic predictors of morbidity in systemic lupus erythematosus.

код для вставкиСкачать
Number 2, February 1995, pp 267-273
0 1995, American College of Rheumatology
Objective. We studied the relationship between
systemic lupus erythematosus (SLE) morbidity and
socioeconomic status (SES) at 5 centers.
Methods. Ninety-nine patients who met American
College of Rheumatology criteria for SLE were randomly sampled at each center, balancing by race and
insurance status. Subjects were interviewed for current
and past SES factors, such as insurance, occupation,
employment, education, and income. SLE disease activity was measured by the SLE Activity Measure (SLAM).
Result. Higher education, private insurance/
Medicare, and higher income were associated with less
disease activity at diagnosis. Controlling for SES, race,
and center, the best predictors of less active disease at
diagnosis were private insuranceMedicare (P = 0.002)
and higher education (P = 0.007). From the time of
diagnosis to the study visit (mean 3.5 years), insurance,
income, and employment status changed for a signifiSupported in part by NIH grants AR-36308 and AR-39921.
Dr. Karlson’s work was supported by NIH institutional training
program no. AI-07306. Dr. Grosflam’s work was supported by NIH
institutional training program no. AR-07530.
Elizabeth W. Karlson, MD, Robert A. Lew, PhD, Elizabeth A. Wright, PhD, Alison J. Partridge, LICSW, Jodi M. Grosflam, MD, Matthew H. Liang, MD, MPH: Robert B. Brigham
Multipurpose Arthritis and Musculoskeletal Diseases Center,
Brigham and Women’s Hospital, Harvard Medical School, Boston,
Massachusetts; Lawren H. Daltroy, DrPH: Robert B. Brigham
Multipurpose Arthritis and Musculoskeletal Diseases Center,
Brigham and Women’s Hospital, Harvard Medical School, and the
Harvard School of Public Health; W. Neal Roberts, MD: Medical
College of Virginia, Richmond; Steven H. Stem, MD: University of
Louisville, Louisville, Kentucky; Karin V. Straaton, MD: University of Alabama, Birmingham; Mary C. Wacholtz, MD: University
of Texas Southwestern Medical Center at Dallas.
Address reprint requests to Elizabeth W. Karlson, MD,
Department of Rheumatology/Immunology,Brigham and Women’s
Hospital, 75 Francis Street, Boston, MA 02115.
Submitted for publication March 11, 1994; accepted September 1, 1994.
cant number of subjects (37%, 16%, and 21% , respectively).
Conclusion. Private insurance or Medicare and
higher education are associated with less active disease
at diagnosis of SLE. Health insurance, income, and
employment status are unstable measures of socioeconomic status and may explain the variability in
conclusions of previous studies on the role of SES in
The purpose of this study was to examine the
relative usefulness of socioeconomic factors for predicting morbidity in a chronic disease, systemic lupus
erythematosus (SLE). Lower socioeconomic status
(SES) and black race are associated with excess
mortality and morbidity in many chronic diseases
(1-3). This has been demonstrated by several studies
of SLE, but the relationship between SES and race in
SLE is controversial (4-14). Three studies showed that
non-white race was significantly associated with decreased survival or more severe disease (5,9,11). This
association was confounded by SES in 2 of the studies
(5,ll). Two studies were unable to show a significant
relationship between SES, race, and disease severity
(12,13). There are several potential explanations for
these discrepancies. For example, patients were studied at different points in their disease course, indirect
measures of SES, such as health insurance status,
were used, and SES may not be constant over time. In
addition, none of the previous studies has evaluated
the uninsured, who now total 37 million people in the
US (15).
SES encompasses a cluster of related variables,
including education, occupational prestige, and income, as they relate to a person’s social status. Some
of these measures have changed over time in terms of
their importance or predictive power in medical and
sociologic research. For example, over the last 20
years, more people have completed college; thus,
more people in a younger cohort might be expected to
have higher educational status compared with people
in an older cohort (16). Most standard indices of SES
combine factors because no single factor adequately
measures all dimensions of the construct. However,
the optimal weight of these factors has not been
worked out, and any index of socioeconomic factors
may not be the best combination for predicting health
outcome and may mask some causal pathways. Indeed, Liberatos et a1 have pointed out weaknesses in
the use of surrogates or indirect measures of SES (16).
Although correlated, the components of SES, when
considered separately, may determine different aspects of morbidity and mortality.
We used data on the first 99 subjects enrolled in
a multicenter study of modifiable risk factors in SLE to
assess the most common techniques of classifying
socioeconomic factors such as education, income,
occupation, employment status, and health insurance
status. Our hypothesis was that some of these socioeconomic factors change over time as the result of
disease morbidity. We therefore investigated these
factors at the time of diagnosis of SLE and at a
followup study visit a mean of 3.5 years later, to assess
their stability and their individual contributions to
SLE morbidity.
Study population. All SLE patients (n = 1,467) seen
at 5 centers since January 1, 1986 were identified from
rheumatology registries and hospital databases. The academic centers have investigative programs in SLE and large
numbers of well-characterized, well-studied black and white
SLE patients of varying socioeconomic class. To ensure
heterogeneity in the variables of interest, the patient population at each center was divided initially into 4 strata by race
(blacWwhite) and insurance status (private/public). Patients
meeting American College of Rheumatology criteria for SLE
(17) were randomly sampled for eligibility using centerspecific sampling fractions of 20/n, where n = the stratum
size. If a center had 20 or fewer persons in a specific stratum,
we sampled 100% in order to maintain as much balance as
Those subjects who passed the initial screen (n =
236, 16%) were invited by letter to participate in our study.
Black and white female interviewers were given extensive
training in research interviewing to ensure interrater and
intrarater reliability and sensitivity to potential biases introduced by the subjects’ race, income, and education. In-
formed consent was obtained. Travel expenses were reimbursed and an honorarium was given to each subject.
The study visits occurred an average of 3.5 years
after diagnosis. The interviewer conducted a structured
interview that included questions on age, race, disease
duration, education, income, occupation, employment status, insurance status, dietary practices, and psychosocial
indicators. Socioeconomic factors were assessed at the time
of the study visit as well as at the time of SLE diagnosis. At
the study visit, each subject was seen by a rheumatologist
who performed an SLE Activity Measure (SLAM). The
SLAM score at diagnosis was assessed by medical record
review by a rheumatologist blinded to SES information.
Study measures. Disease activity was measured by
the SLAM, a physician-rated, reliable index of clinical and
laboratory parameters that has been validated against patient
and clinician judgment of SLE disease activity (18,19). A
SLAM score can range from 0 (no disease activity) to 84
(maximum disease activity). The SLAM score was analyzed
as an ordered variable (mild 4,moderate 9-16, severe >16)
for Spearman and Pearson correlations and as a continuous
variable for all other analyses.
Socioeconomic factors. Education was measured as
the number of years of education and was analyzed as an
ordered variable (failed to complete high school, completed
high school, education beyond high school). To avoid misclassifying subjects who were still in high school at diagnosis, we assigned education levels achieved at study visit.
None of the subjects who were in high school at diagnosis
failed to complete high school by the time of study visit.
Income was measured as total yearly household
income at diagnosis (by patient report) and at the study visit,
in 7 categories (<$S,OOO, $5,000-9,999, $10,000-14,999,
$15,000-19,999, $20,00&29,999, $30,ooCrS0,000,
Other income measures used in the analysis included the
midpoint income of the category and the midpoint income of
the category adjusted for the number of persons in the
household (expressed in 1991 dollars). A dichotomous income variable was defined as income above or below the
national poverty level, adjusted for year and family size (15).
Occupational prestige at diagnosis and study visit
was scored using the National Opinion Research Center
(NORC) occupational prestige scale, which was developed
at the University of Chicago to cover the occupational codes
used in the 1980 US census. This scale is derived from a
nationally representative sample of 1,166 adults who were
asked to rank occupations by prestige, from 0-100, and
results in a range of 11-87 (20). Occupation at diagnosis and
study visit was defined as the usual occupation of the subject
if the subject was single, divorced, separated, or widowed at
the time. If married/cohabitating, the NORC score of the
spouse/partner (for usual occupation) was used, if higher. If
retired, the usual lifetime occupation was scored, again with
the higher score selected in couples. If the subject was a
student and not employed, occupation was scored as missing.
Employment status, employed/unemployed, was
scored separately from occupational prestige.
Insurance status at diagnosis and at study visit was
defined by the following categorical variables: private insurance and/or Medicare, Medicaid or free care, and uninsured
or self-pay. Medicare is the Federal Government’s health
insurance program for the aged (65 years and over) and
certain categories of the disabled regardless of income.
Medicaid is a medical assistance program that provides
hospital and medical services free of charge to persons who
have total assets of less than $2,000 apart from house and car
and meet state-specific poverty guidelines.
Statistical analysis. We evaluated changes in socioeconomic factors over time by comparing the value of each
factor at diagnosis to the value at study visit. For selected
proportions, we calculated 95% confidence intervals. Sociodemographic factors in insurance groups were tested using
chi-square tests. We performed Spearman correlations
among socioeconomic factors at diagnosis and among socioeconomic factors at study visit, with insurance status dichotomized as high = PrivatelMedicare and low = Medicaid/
uninsured. For all other analyses, insurance status was
analyzed in 3 categories. Pearson and nonparametric Spearman correlations were used to assess bivariate associations
among socioeconomic and demographic factors and SLAM
score at diagnosis. Multiple linear regression and general
linear modeling were used to predict SLAM score at diagnosis by socioeconomic factors, demographic factors, and
center. Residuals and influence points were examined to
determine the sensitivity of the model. A P value of less than
0.05 was considered significant.
Characteristics of the study subjects. The participation rate in eligible subjects was 82% (range 74100% in the 5 centers). We analyzed the first 99
Table 1. Demographics and socioeconomic factors in 99 subjects
At diagnosis
Black, no. (%)
Female, no. (%)
Mean age, years (SD)
Mean NORC score (SD)*
Unemployed, no. (%)
Mean dollars (SD)
Adjusted mean dollars
Below poverty level
At study visit
58 (59)
93 (94)
34 (13.1)
45.0 (14.2)
7 (7)
37.5 (13.0)
46.7 (13.3)
22 (22)
19,400 (17,200)
29,400 (27,400)
19,700 (17,600)
25,500 (21,500)
Less than high school
High school (%)
Beyond high school (%)
Private/Medicare, no.
Medicaid, no. (%)
Uninsured, no. (%)
62 (63)
56 (57)
13 (13)
24 (24)
39 (39)
4 (4)
* National Opinion Research Center (NORC) occupational prestige
scale (see Patients and Methods for details).
t Adjusted to 1991 dollars for family size of 4.
$ Adjusted for year and family size.
Table 2. Changes in socioeconomic factors from diagnosis to
study visit
% with a change
% with no
(95% confidence
37 (3042)
16 (9-23)
21 (13-29)
* Defined as above or below the national poverty level (adjusted for
year and number of people in household).
subjects enrolled in the study. There were 93 females
and 6 males. Fifty-eight were black and 41 were white
(Table 1). The mean age at diagnosis of SLE was 34
years with mean age at study visit of 37.5 years.
Twenty-six subjects had less than a high school education, 35 were high school graduates, and 38 had
education beyond high school. At diagnosis of SLE, 62
subjects had private or Medicare insurance, 13 had
Medicaid insurance, and 24 were uninsured. At the
study visit, 56 subjects had private or Medicare insurance, 39 had Medicaid, and 4 were uninsured.
The unadjusted mean income was $19,400 at
diagnosis and $19,700 at study visit. Mean income
adjusted by family size expressed in 1991 dollars was
$29,400 at diagnosis and $25,500 at the study visit.
Twenty-eight percent were below the poverty level
(adjusted for year and family size) at diagnosis and
34% were below the poverty level at study visit. Mean
NORC occupational prestige scores were 45.0 at diagnosis and 46.7 at study visit. Six subjects who were
students at diagnosis were coded as missing a NORC
score but had NORC scores at the study visit. Seven
patients were unemployed at diagnosis and 22 patients
were unemployed at study visit.
Stability of socioeconomic factors. We studied
the stability of insurance status, income, and employment status over time, using cross tabulations. Insurance, income, and employment status changed in 37%,
16%, and 21% of subjects, respectively, and this was
statistically significant (Table 2). Thirty-seven subjects
reported being without health insurance at some time
since the SLE diagnosis, with a mean of 21 months
without insurance. Lack of insurance was attributed to
SLE-related job loss by 15 subjects and to low income
by 13 subjects.
Of the 24 subjects who were uninsured at
diagnosis, 17 had obtained Medicaid insurance (13
through Supplemental Security Income and 4 through
Aid to Families with Dependent Children), 3 had
Table 3.
Health insurance status at diagnosis and at study visit
Insurance at
SLE diagnosis
Insurance at study visit
Medicaid Uninsured PrivatelMedicare Total
Table 5. Spearman correlation coefficients among socioeconomic
factors at diagnosis
Educatiodemploy ment
obtained private insurance, 2 had obtained Medicare,
and 2 remained uninsured by the time of the study
visit. Of the 4 uninsured subjects at study visit, 2 had
lost private insurance since the diagnosis of SLE
(Table 3). Subjects who were uninsured or had Medicaid at diagnosis were significantly less likely to have
education beyond high school than those with private
insurance/Medicare (P < 0.05) and more likely to be
below national poverty levels (P < 0.05) (Table 4).
Correlations among socioeconomic factors. Correlations among socioeconomic factors assessed at
diagnosis, including education, income, occupational
prestige, and insurance status, ranged from 0.22 to
0.50 (all P c 0.05) and all were in the directions
expected (Table 5 ) . Correlations among socioeconomic factors determined at the time of the study visit
ranged from 0.27 to 0.53 (all P < 0.05) (Table 6).
Occupational prestige had a relatively low correlation
with insurance status at study visit (r = 0.27) compared with other correlations among the factors. Employment status at diagnosis and study visit correlated
poorly with other factors (r = 0.10422).
Individual socioeconomic factors and demographic factors at diagnosis were studied for associations with SLE disease activity (SLAM score) at
diagnosis, using Pearson correlations (Table 7). Higher
education, private insurancelMedicare, and higher income were significantly associated with less disease
activity at diagnosis. Medicaid and no insurance were
associated with more disease activity. Occupational
prestige, employment status, age, sex, and race were
not significantly associated with disease activity. The
analysis was repeated with Spearman correlations and
gave virtually identical results.
Regression models. In a multivariate linear
model that adjusted for individual socioeconomic factors at diagnosis, the factors which best predicted less
disease activity at diagnosis were higher education
(P = 0.03) and private insurance/Medicare status (P =
0.003). When demographic factors such as race, sex,
and study center were included in the model, the
associations for higher education and insurance status
remained significant (P = 0.007 and P = 0.002, respectively). Twenty-seven percent of the total variance
was explained by this model. Race, Medicaid, uninsured, income defined in 4 different ways (see Patients
and Methods), occupational prestige, and employment
status were not significant correlates of disease activity at diagnosis. No interactions between the significant variables in the multivariate models were found.
In all models, education and insurance remained significant.
Demographics of subjects at diagnosis, by insurance
Of the components of socioeconomic status, we
found that education and insurance status were signif-
Table 4.
Income below
Education beyond
high school
(n = 13)
(n = 24)
(n = 62)
6 (46)t
12 (SO)$
10 (16)
32 (52)
Insurance/employ ment
1 (8)t
9 (69)
2 (15)
5 (211%
17 (71)
2 (8)
32 (52)
3 (5)
* Values are the number (%) of patients. P values determined by
chi-square test.
t P 4 0.05 for Medicaid versus private insurancehiedicare.
$ P 4 0.05 for uninsured versus private insurance/Medicare.
Table 6. Spearman correlation coefficients among socioeconomic
factors at study visit
Table 7. Pearson correlations of socioeconomic and demographic
factors with disease activity (SLAM score) at diagnosis*
Male sex
Black race
* SLAM = Systematic Lupus Erythematosus Activity Measure.
icantly associated with baseline morbidity in SLE. We
present evidence that of the possible ways of characterizing SES, education is the most stable and, thus,
preferred. Insurance status changed in 37% of the
subjects from the time of diagnosis to the time of
sampling. Thirty-seven percent of subjects reported
being without insurance at some time after SLE diagnosis. This lack of stability suggests that insurance
status is a poor indirect measure of SES. Previous
cross-sectional studies showing significant SES and
racial associations with disease morbidity and mortality using insurance status assessed at the time of the
study may have shown different results if insurance
status had been assessed at the diagnosis of SLE,
before it could be affected by disease morbidity.
SES has been repeatedly demonstrated to be a
powerful explanatory variable in disease outcome (21).
Its mechanism, however, is complex. Research on
what features related to lower SES, such as poor
nutrition, unwise health behaviors, and lack of access
to health care, is urgently needed. Access alone does
not explain all variation in outcome. In countries such
as England, where there is universal access to health
care, a health and SES gradient still exists (21).
SLE is a specific example of the general mortality differences between races (1). Blacks with SLE
appear to have earlier onset, more severe manifestations, and higher mortality than whites (4-14).
Whether mortality and morbidity differences in SLE
patient groups result from racial or from SES differences is controversial and of major policy relevance.
In a landmark study, Ginzler et a1 (5) found that after
controlling for SES (classified as health insurance
status), race did not significantly influence survival.
Studies by Dubois et a1 (7), Wallace et a1 (8), Studenski
et a1 (9), Reveille et a1 (lo), and Ward and Studenski
27 1
(11) also suggest the independent effect of SES on
survival and outcome when health insurance status is
used as a proxy for SES. In contrast, Studenski et a1
(9) found independent effects of both race and SES on
survival. In Canada, where at least financial barriers
do not prevent access to health care, SES seems to
have less effect (12). Unlike most of these studies
which have examined low SES black patients and high
SES white patients, we balanced racial groups on SES
to mitigate confounding. When this is done, race is not
significantly associated with disease activity, even
when other socioeconomic factors are taken into account.
In studies of SES in general and in the ones
cited above, high SES and low SES have been defined
in different ways and may explain some of the apparent discrepancies. Esdaile et a1 (12) assessed SES by
occupational prestige and educational level. Ginzler et
al(5) used private insurance and Medicare as high SES
and public funding (Medicaid) as low SES. Ward and
Studenski (11) and Studenski et a1 (9) used private
insurance as high SES but classified Medicare alone
with Medicaid as low SES. Means-testing is used to
determine eligibility for Medicaid, which is defined by
state law and, thus, easily classified as low SES.
However, patients with Medicare or no insurance
cannot be easily classified into SES categories without
information about other socioeconomic factors such as
In our study, education, income, and occupational prestige were examined separately and in combinations, but not as an index. Indices do not assess
the stability of the component factors over time or
discriminate between the patients with very low incomes. For example, the Nam-Powers SES Index,
used in one study of SLE morbidity (13), is based on
occupational status scores, which include education
and income factors. To score the index, education and
income factors are averaged with the occupational
status score, thus using education and income twice
(16). Such subtle combinations tend to obscure the
mechanism by which individual socioeconomic factors
influence disease outcome.
No feature of SES is completely reliable or
stable over time. However, compared with other features, health insurance status is the least stable. Income and employment status also vary over time. The
most stable measure of SES appears to be the level of
education, because most patients have completed
schooling prior to the onset of disease and the level
cannot decline. In addition, education is highly correlated with other socioeconomic factors. If an indirect
measure for SES is needed in epidemiologic studies of
the effect of the social environment on outcome in
chronic disease, education is a better choice than
health insurance status.
Patients who are uninsured for health services
present unique problems. They have not been evaluated as a separate group in previous studies of SLE.
Currently, approximately 37 million Americans fall
into this category (15). In this study, the uninsured
subjects were less likely to have higher education and
more likely to be poor than were those with insurance,
but were otherwise similar to the other study subjects.
Lack of insurance, analyzed in the multivariate models
as a separate category, was not significantly associated
with disease activity at diagnosis of SLE. It is important to assess whether being uninsured is a risk factor
for excess disease morbidity at a later time, however.
Some limitations of the study are acknowledged. Some of the data were retrospective. It is
conceivable that patient recall of certain SES factors
and assessment of disease activity by chart review are
not accurate, but we do not believe the results would
be altered. The subjects were drawn from 5 referral
centers and related practices, and the results may not
be generalizable to patients with SLE followed in the
community. The relative importance of individual socioeconomic factors in predicting chronic disease morbidity may change or differ between chronic diseases and
the findings may not be generalizable to other chronic
diseases. However, it should be noted that researchers
in cardiovascular diseases have also concluded that
education is the best single predictor of cardiovascular
morbidity and have recommended its use as a proxy
for SES in epidemiologic studies (22).
In conclusion, changes in insurance status accelerated by the morbidity of a chronic disease, increasingly higher standards of education, and variation
in the valuation of prestige all imply that we should
consider the effects of socioeconomic factors at different times in the patient’s history. The precise manner
in which SES is classified and when during the course
of disease socioeconomic factors are measured can
affect the conclusions of a study. Our study shows that
indirect measures of SES, such as health insurance
status, are unstable in SLE. Education is the more
stable measure of SES and is significantly associated
with baseline disease morbidity.
We are indebted to Nova1 Abraham, Kay Morgan,
Elain Davis, and Sarah Breitbach for coordinating the study
at the collaborating centers and for help with patient interviews, to Holly Fossel for data management, and to Jacqueline Mazzie and Nancy Tanner for secretarial support. We
also thank Drs. John Esdaile and Jeffrey Katz for their
expert advice.
1. Pappas G, Queen S, Hadden W, Fisher G: The increasing
disparity in mortality between socioeconomic groups in the
United States, 1960 and 1986. N Engl J Med 329: 103-109, 1993
2. United States Department of Health and Human Services:
Health in the United States. DHHS publication no. (PHS)
91-1232. Washington, DC, United States Government Printing
Office, 1990
3. Ayanian JZ, Kohler BA, Abe T, Epstein AM: The relationship
between health insurance coverage and clinical outcomes
among women with breast cancer. N Engl J Med 326-331, 1993
4. Gordon MF, Stolley PD, Schinnar R: Trends in recent systemic
lupus erythematosus mortality rates. Arthritis Rheum 24:762769, 1981
5. 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
systematic lupus erythematosus. I. Entry variables as predictors of prognosis. Arthritis Rheum 25:601-611, 1982
6. Larson MG, Liang MH: The multicenter study of outcome in
systemic lupus erythematosus: a critique (letter). Arthritis
Rheum 26:57&571, 1983
7. Dubois EL, Wierzchowiecki M, Cox MB, Weiner JM: Duration
and death in systemic lupus erythematosus: an analysis of 249
cases. JAMA 227:1399-1402, 1974
8. Wallace DJ, Podell T, Weiner J, Klinenberg JR, Forouzesh S,
Dubois EL: Systemic lupus erythematosus survival patterns:
experience with 609 patients. JAMA 245:934938, 1981
9. Studenski S, Allen NB, Caldwell DS, Rice JR, Polisson RP:
Survival in systemic lupus erythematosus: a multivariate analysis of demographic factors. Arthritis Rheum 30: 1326-1332, 1987
10. Reveille JD, Bartolucci A, Alarc6n GS: Prognosis in systemic
lupus erythematosus: negative impact of increasing age at onset,
black race, and thrombocytopenia, as well as causes of death.
Arthritis Rheum 33:3748, 1990
11. Ward MM, Studenski S: Clinical manifestations of systemic
lupus erythematosus: identification of racial and socioeconomic
influences. Arch Intern Med 150:849-853, 1990
12. Esdaile JM, Sampalis JS, Lacaille D, Danoff D: The relationship
of socioeconomic status to subsequent health status in systemic
lupus erythematosus. Arthritis Rheum 31:423427, 1988
13. Petri M, Perez-Gutthann S, Longenecker JC: Morbidity of
systemic lupus erythematosus: role of race and socioeconomic
status. Am J Med 91:345-353, 1991
14. Liang MH, Partridge AJ, Daltroy LH, Straaton KV, Galper SR,
Holman HR: Strategies for reducing excess morbidity in blacks
with systemic lupus erythematosus. Arthritis Rheum 34: 11871196, 1991
15. United States Bureau of the Census: Current Population Reports, series P-60(181): Poverty in the United States, 1991.
Washington, DC, United States Government Printing Office,
16. Liberatos P, Link BG, Kelsey JL: The measurement of social
class in epidemiology. Epidemiol Rev 1037-121, 1988
17. 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, 1982
18. Liang MH, Socher SA, Roberts WN, Esdaile JM: Measurement
of systemic lupus erythematosus activity in clinical research.
Arthritis Rheum 31:817-825, 1988
19. Liang MH, Socher SA, L a s o n MG, Schur PH: Reliability and
validity of six systems for the clinical assessment of disease
activity in systemic lupus erythematosus. Arthritis Rheum
32: 1 107-1 1 18, 1989
20. Nakao K, Treas J: Computing 1989 occupational prestige
scores. General Social Survey Methodological Report 70:1-27,
21. Adler NE, Boyce WT, Chesney MA, Folkman S, Syme L:
Socioeconomic inequalities in health: no easy solution. JAMA
269:314&3145, 1993
22. Winkleby MA, Jatulis DE, Frank E, Fortmann SP: Socioeconomic status and health: how education, income and occupation
contribute to risk factors for cardiovascular disease. Am J
Public Health 82:816-820, 1992
Без категории
Размер файла
665 Кб
lupus, systemic, erythematosus, morbidity, independence, socioeconomic, predictor, stability
Пожаловаться на содержимое документа