Prevalence of adult systemic lupus erythematosus in California and Pennsylvania in 2000: Estimates obtained using hospitalization data

Authors


The prevalence of systemic lupus erythematosus (SLE) in the US is not known. The estimates that are available have been made based on studies of populations in small geographic areas with unique demographic characteristics, or on telephone surveys (1). These estimates range widely, from 5.8 to 1,000 per 100,000 persons (1–3). The availability of current, nationally representative estimates of the prevalence of SLE would increase our understanding of the burden of disease and help in health care planning.

We applied a new method that combines data on the frequency with which patients are hospitalized with data from state inpatient hospital databases to estimate the prevalence of SLE in adults (ages 18 years and older) in California and Pennsylvania (4). California, through its Office of Statewide Health Planning and Development, and Pennsylvania, through the Health Care Cost Containment Council, mandate that hospitals report discharge abstracts for each hospitalization annually. Abstracted data include patient age, sex, race, and discharge diagnoses (up to 25 in California; up to 8 in Pennsylvania), as well as encrypted unique patient identifiers to identify individuals with multiple hospitalizations.

From these databases, we identified the number of adults in each state who had a hospitalization in 2000 that included SLE among the discharge diagnoses. Dividing these numbers by the proportion of patients with SLE who are hospitalized each year provides an estimate of the number of patients with SLE in each state. Prevalence was then calculated by dividing this number by the state populations in the 2000 US Census. Prevalence estimates were age-, sex-, and race/ethnicity-adjusted to the structure of the population in the 2000 US Census, by direct standardization. Prevalence was calculated separately for each state.

Data on the proportion of patients with SLE hospitalized annually were obtained by medical record review of all patients with SLE (n = 415) followed up routinely at 1 of 2 medical centers in northern California: Stanford University Hospital and Clinics and Santa Clara Valley Medical Center, a public medical center serving the county of Santa Clara. In addition, annual hospitalization data were obtained by self-report of 117 patients with SLE in Pennsylvania enrolled in an observational study of health care use from 1996 through 2001 (5, 6). All patients met the American College of Rheumatology criteria for the classification of SLE (7). Data on sex and age (18–44 years, 45–64 years, and ≥65 years) and ethnicity were collected for each patient. Emergency department visits that did not result in a hospital admission were excluded since they are not captured in the inpatient hospital databases.

Overall hospitalization frequencies were 28% in California and 22% in Pennsylvania. Among California patients, hospitalization frequency was significantly higher among women age ≥65 years (63%) than for all other demographic subgroups (28%). The smaller number of Pennsylvania patients did not permit analysis of hospitalization frequencies in subgroups by age, sex, and ethnicity.

The overall age-, sex-, and race/ethnicity-adjusted SLE prevalence per 100,000 persons age 18 years or older was 107.6 (95% confidence interval [95% CI] 106.1–109.2) in California and 149.5 (95% CI 146.9–152.2) in Pennsylvania. Prevalence estimates for each state by sex, race/ethnicity, and age group are shown in Table 1. Women had a higher prevalence than men, with a ratio of ∼7:1. Because of the small numbers of men with SLE, rates in demographic subsets were not examined. Among women, African Americans had a higher prevalence of lupus by 2.5–3.4-fold compared with whites. Asian/Pacific Islander women appeared to have the lowest prevalence of SLE.

Table 1. Prevalence estimates of adult SLE per 100,000 persons*
GroupCaliforniaPennsylvania
Prevalence95% CIPrevalence95% CI
  • *

    SLE = systemic lupus erythematosys; 95% CI = 95% confidence interval.

  • Adjusted for age, sex, and ethnicity.

  • Adjusted for age and ethnicity.

  • §

    Adjusted for age.

  • Adjusted for ethnicity.

All107.6106.1–109.2149.5146.9–152.2
All men25.524.5–26.638.736.8–40.7
All women184.2181.4–187.0253.0248.3–257.7
White women§164.4161.6–167.3203.1198.8–207.2
African American women§406.3392.5–420.2693.7669.7–717.8
Hispanic women§138.7134.7–142.7244.5216.1–272.9
Asian/Pacific Islander women§92.787.9–97.7103.281.0–125.4
Women ages 18–44 years162.5158.9–166.2180.6174.7–186.5
Women ages 45–64 years272.2266.0–278.4346.2336.0–356.4
Women age ≥65 years107.6102.8–112.5305.5295.1–315.9

To examine the influence of differences in hospitalization frequencies on the differences in prevalence estimates between California and Pennsylvania, we repeated the analyses applying the hospitalization frequencies from California to the Pennsylvania hospital discharge data. This resulted in prevalence estimates in Pennsylvania of 104.3 per 100,000 (95% CI 102.1–106.6) for all adults and 173.3 per 100,000 (95% CI 169.3–177.2) for women, which were very similar to the estimates in California (107.6 per 100,000 adults and 184.2 per 100,000 adult women). Similarly, applying the hospitalization frequencies from Pennsylvania to the California hospital discharge data resulted in prevalence estimates of 153.9 per 100,000 adults (95% CI 152.1–155.7) and 266.9 per 100,000 women (95% CI 263.6–270.2), estimates that were very similar to the Pennsylvania estimates (149.5 per 100,000 adults and 253.0 per 100,000 adult women). Differences in the hospitalization frequencies led directly to differences in the prevalence estimates between the states, rather than true differences in the prevalence of SLE. The true prevalence likely lies somewhere between these calculated estimates.

Using this new method for estimating statewide prevalence of chronic diseases, which compares hospitalization frequency with state hospitalization databases, we obtained age-, sex-, and race/ethnicity specific estimates of the prevalence of SLE. Adjusted prevalence estimates obtained from these data range from 108 to 150 per 100,000 persons overall, and from roughly 1.8 to 2.5 per 1,000 women. These results fall within the upper bounds of estimates obtained from studies of local populations and within the range of estimates based on a nationwide study of self-report of physician-diagnosed SLE (1, 3). The prevalence estimates in this study are higher than those reported from Olmsted County, Minnesota (8) and rural Wisconsin (9), which might be expected given the known predisposition of ethnic minorities to SLE and the relative absence of ethnic minorities in those populations. The close approximation of prevalences estimated with this technique to those reported in the literature is similar to findings in other uncommon chronic diseases (4).

The use of state hospitalization databases does not allow for verification of diagnoses of SLE, but these estimates are based on physician (not necessarily rheumatologist) diagnosis. Undercounting in the state databases is unlikely given the large number of available fields for coding discharge diagnoses. More accurate estimates would be obtained using hospitalization frequencies from a population-based sample of patients with SLE, but such data are not available. However, we obtained similar estimates (22–28% of patients hospitalized annually) using 2 separate cohorts of patients (consecutive patients followed up at academic as well as community-based rheumatology clinics in northern California, and self-reported hospitalization data obtained from patients participating in a longitudinal study of SLE), suggesting that our estimates may approximate true annual hospitalization frequencies among adults with SLE.

It is important to emphasize that these prevalence estimates apply to persons age 18 years and older. If children had been included, these estimates would be ∼25% lower, based on the proportion of children in each state (27.3% in California and 23.8% in Pennsylvania). Given that our prevalence estimates reflect a broader demographic range than estimates obtained from previous studies and represent samples from 2 populous states, these rates may be considered current nationwide estimates for the prevalence of adult SLE.

AUTHOR CONTRIBUTIONS

Dr. Ward had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Ward.

Acquisition of data. Chakravarty, Bush, Manzi, Clarke, Ward.

Analysis and interpretation of data. Chakravarty, Bush, Clarke, Ward.

Manuscript preparation. Chakravarty, Bush, Manzi, Clarke, Ward.

Statistical analysis. Chakravarty, Ward.

Acknowledgements

Supported in part by the Fonds de la recherche en sante du Quebec and the Arthritis Society of Canada. Dr. Manzi's work was supported in part by NIH grants 5-R01-AR-4658805 and 5-K24-AR-00221305 and the Pennsylvania Department of Health. Dr. Ward's work was supported in part by the Intramural Research Program of the National Institute of Arthritis and Musculoskeletal and Skin Diseases. Dr. Clarke is a Quebec National Research Scholar.

Ancillary