Education level and mortality in systemic lupus erythematosus (SLE): Evidence of underascertainment of deaths due to SLE in ethnic minorities with low education levels

Authors

  • Michael M. Ward

    Corresponding author
    1. National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Maryland
    • NIH/NIAMS/IRP, Building 10, Room 9S205, 10 Center Drive, MSC 1828, Bethesda, MD 20892-1828===

    Search for more papers by this author

Abstract

Objective

To determine if socioeconomic status, as measured by education level, is associated with mortality due to systemic lupus erythematosus (SLE), and to determine if these associations differ among ethnic groups.

Methods

Sex- and race-specific mortality rates due to SLE by education level were computed for persons age 25–64 years using US Multiple Causes of Death data from 1994 to 1997. SLE-specific mortality rates were compared with all-cause mortality rates in 1997 to determine if the association between education level and mortality in SLE was similar to that in other causes of death.

Results

Among whites, the risk of death due to SLE was significantly higher among those with lower levels of education, and the risk gradient closely paralleled the 1997 all-cause mortality risks by education level. However, in African American women and men and Asian/Pacific Islander women, the risk of death due to SLE was lower among those with lower education levels, contrary to the associations between education level and all-cause mortality in these groups. Comparing the distribution of education levels among deaths due to SLE and all deaths in 1997, persons with lower education levels were underrepresented among deaths due to SLE in African Americans and Asian/Pacific Islanders.

Conclusion

Among whites, higher education levels are associated with lower mortality due to SLE. These associations were not present in ethnic minorities, likely due to underascertainment of deaths due to SLE in less-well educated persons. This underascertainment may be due to underreporting of SLE on death certificates, but may also represent underdiagnosis of SLE in ethnic minorities with low education levels.

INTRODUCTION

In epidemiologic studies, higher socioeconomic status has been consistently associated with lower all-cause mortality and with lower mortality due to cardiovascular and cerebrovascular disease (1–15). Similar associations have been found for both whites and African Americans when race-specific mortality has been examined (2–4, 6–10, 12, 14, 15). In studies that used education level as the measure of socioeconomic status, mortality rates among those with ≤12 years of education were often 1.5–2.0 times higher than those of college graduates (1, 4, 5, 8, 13). Similar associations have been reported for patients with rheumatoid arthritis (16–20).

Few studies have examined the association of socioeconomic status with mortality in systemic lupus erythematosus (SLE) (21–25). Most studies examined clinic-based cohorts, which may be susceptible to selection bias and have limited generalizability. Four of the 5 studies used medical insurance status as the measure of socioeconomic status, or inferred patients' socioeconomic status from the economic characteristics of their area of residence, rather than using more direct personal measures of socioeconomic status, such as income or education level (21–24). An ecologic analysis of national data from the United States found higher mortality rates due to SLE in counties with higher poverty rates (26). Previous studies also reported higher mortality rates due to SLE among African Americans and Asian Americans, but these studies did not examine mortality by socioeconomic status (27–29). Knowing if socioeconomic status is associated with mortality in patients with SLE would indicate whether reports of higher mortality among ethnic minorities were confounded by differences in socioeconomic status. This question is important because although the health outcomes of patients with SLE have improved over time, 10–20% of patients with SLE die within 10 years of its onset (30).

In this study, I examined the hypothesis that socioeconomic status, measured by education level, was inversely associated with mortality due to SLE in whites, African Americans, and Asian/Pacific Islanders in the United States, using national data on deaths from 1994 to 1997. Trends in SLE mortality by education level were also compared with similar trends in all-cause mortality.

METHODS

Data on the number of deaths occurring in the United States in persons with SLE were abstracted from Multiple Causes of Death files of the National Center for Health Statistics (31). These files include information on all recorded deaths occurring in the United States each year, based on death certificates filed in each state. Records for each death include the underlying cause of death and up to 20 additional conditions listed as associated or contributing causes of death, by International Classification of Diseases Ninth Revision (ICD-9) codes (32). Each record also includes the decedents' age, sex, race (white, black, Native American, Asian/Pacific Islander), Hispanic ethnicity, and place of residence, as well as information on the time and place of death.

Information on the decedents' education level was first included on the Standard Certificate of Death in 1989, but this item was not immediately adopted by all states. In 1993, information on education level was not available for deaths in New York City, Georgia, Oklahoma, Rhode Island, and South Dakota, was missing for >20% of deaths in Kentucky and West Virginia (and therefore considered unreliable), and was missing for 9% of deaths in other states. Beginning in 1994, information on education level was available for deaths in 46 states, New York City, and the District of Columbia. Georgia, Oklahoma, Rhode Island, and South Dakota did not include education level on their death certificates. Data on education level were missing for >20% of deaths in Kentucky. In the remaining states, data on education level were missing for ≤4.2% of deaths since 1994.

For this study, all deaths occurring in 45 states and the District of Columbia (excluding Georgia, Oklahoma, Rhode Island, South Dakota, and Kentucky) from 1994 to 1997 in which SLE was included as either the underlying cause, immediate cause, or contributing cause of death were included. The study was limited to persons age 25–64 years because SLE most often occurs in young adulthood or middle age, because most people have completed their formal education by age 25, and because death certificate data on education level may be less accurate among persons age 65 and older (33). For each death, data were abstracted on age, sex, race, years of education, and place of death.

Nephritis and renal failure are markers of severe SLE and major risk factors for death (23, 34–37), and likely identify a subset of persons in whom the diagnosis of SLE is made with greater specificity. Therefore, decedents whose records included acute or chronic glomerulonephritis (ICD-9 codes 580–583), acute renal failure (ICD-9 code 584), chronic renal failure (ICD-9 code 585), or renal failure, unspecified (ICD-9 code 586) as one of the multiple causes of death along with SLE were also identified.

Mortality rates by education level for each sex and race subgroup were computed using census counts as the denominators. Counts by age (in 5-year age groups), sex, race, education level, and state from the 1990 census were projected to 1994–1997, and the projections were added to provide the population at risk of death in these years (38). Counts for Georgia, Oklahoma, Rhode Island, South Dakota, and Kentucky were omitted from these totals. Although information on deaths by both race and Hispanic ethnicity was available, mortality by Hispanic ethnicity could not be analyzed because census data on education level by both race and Hispanic ethnicity were not available for each state. Therefore, Hispanics were not analyzed as a separate category, but were included in either the white, African American, or Native American racial groups. For white and African American women, mortality rates were computed for education levels of 0–8 years, 9–11 years, 12 years, 13–15 years, and 16 or more years. Because fewer deaths due to SLE occurred in men and Asian/Pacific Islander women, mortality rates for these groups were computed for education levels of 0–11 years, 12 years, and ≥13 years. Because validation studies of death certificate information have indicated that ethnic minorities with <12 years of education were more likely to be recorded as being high school graduates than recorded with lower levels of educational attainment, comparisons were also performed using only 2 categories of education (≤12 versus ≥13 years) (33).

Sex- and race-specific mortality rates were age adjusted using direct standardization, with the 1990 United States population as the standard. For strata with <100 deaths, 95% confidence intervals (95% CIs) for the mortality rates were computed based on Poisson distribution. For strata with ≥100 deaths, 95% CIs for the mortality rates were computed based on z distribution (39). Poisson regression models were used to estimate rate ratios for mortality due to SLE among education groups, with the highest education category as the reference group. P values ≤ 0.05 were considered statistically significant. Analyses were performed separately for each sex and race, and for persons with SLE included as any category of cause of death (underlying, immediate, or contributing), for those with SLE as the underlying cause of death, and for those with SLE who had renal disease included among the causes of death. Analyses were performed using SAS programs (version 8; SAS Institute, Cary, NC).

To determine if the association between education level and mortality due to SLE was similar to that of other causes of death, these analyses were repeated using all deaths in 1997. Consistency in the education-related risks of mortality between deaths due to SLE and all-cause mortality would support the validity of this association, whereas inconsistency would suggest either that the education-mortality association was modified in the setting of SLE, or that there were deficiencies in the identification of deaths due to SLE.

RESULTS

From 1994 to 1997, there were 4,779 deaths among persons age 25–64 years for which SLE was recorded as the underlying, immediate, or contributing cause. Data on education level were missing for 160 persons (3.3%), and these people were excluded from analysis. These persons did not differ from the 4,619 persons who had information on education level in age, sex, race, or the likelihood that SLE was the underlying cause of death. Persons with missing data on education level were more likely to have died in 1994 (4.2%) than in 1997 (2.2%). Of the 4,619 persons studied, 3,939 (85.3%) were women, 2,649 (57.3%) were white, 1,755 were African American (38.0%), 187 (4.0%) were Asian/Pacific Islander, and 28 (0.6%) were Native American. Three hundred forty-seven persons (7.5%) completed ≤8 years of education, 583 (12.6%) completed 9–11 years, 2,003 (43.4%) completed 12 years, 981 (21.2%) completed 13–15 years, and 705 (15.3%) completed ≥16 years. The mean age of death was 46.1 years. SLE was the underlying cause of death for 2,852 persons (61.7%). Renal disease was listed as one of the multiple causes of death for 1,049 persons (22.7%), with the most common diagnoses being unspecified renal failure (n = 564) and chronic renal failure (n = 293). Table 1 shows the distribution of deaths due to SLE in 1994–1997 and all deaths in 1997 by education level (≤12 years versus >12 years). There were too few deaths due to SLE among Asian/Pacific Islander men (n = 24) and Native Americans for meaningful analysis, and these groups were not analyzed further.

Table 1. Number of deaths due to SLE in 1994–1997 and all deaths in 1997 by sex-race subgroup and years of education (≤12 years versus >12 years)*
 White womenWhite menAfrican American womenAfrican American menAsian/Pacific Islander women
  • *

    Data presented as no (%). SLE = systemic lupus erythematosus.

SLE deaths     
 ≤12 years' education1,456 (65.9)290 (66.1)944 (61.3)137 (64.0)73 (44.8)
 >12 years' education754 (34.1)149 (33.9)597 (38.7)77 (36.0)90 (55.2)
All deaths     
 ≤12 years' education91,138 (68.2)150,474 (67.1)25,885 (73.3)42,446 (79.6)2,007 (58.5)
 >12 years' education42,461 (31.8)73,690 (32.9)9,408 (26.7)10,837 (20.4)1,426 (41.5)

Among white women, the age-adjusted mortality rate due to SLE (as an underlying, immediate, or contributing cause of death) decreased from 20.9/million in those with 0–8 years of education to 7.4/million in those with ≥16 years of education (Table 2). White women with 0–8 years of education were twice as likely to have died of SLE than those with ≥16 years of education, and the risks decreased progressively with increasing education level. Results were similar for analyses of SLE as the underlying cause of death and for the subgroup with renal disease, with somewhat stronger risk gradients for these outcomes. Findings in white men were similar to those in women, with mortality rates due to SLE as any cause of death or as the underlying cause significantly higher among those with ≤12 years of education (Table 3). For both white women and white men, the education gradient in risk of mortality due to SLE closely paralleled the education gradient in risk in all-cause mortality (Figure 1).

Table 2. Mortality due to SLE among white women, by education level*
 Education
0–8 years9–11 years12 years13–15 years16 years or more
  • *

    Rates are deaths per million. SLE = systemic lupus erythematosus; RR = rate ratio; 95% CI = 95% confidence interval.

SLE as underlying, immediate, or contributing cause of death     
 n1912601,005433321
 Age-adjusted rate (95% CI)20.9 (17.5–24.4)14.1 (12.2–16.0)12.1 (11.3–12.8)7.7 (6.9–8.4)7.4 (6.4–8.4)
 RR (95% CI)2.05 (1.40–3.00)1.64 (1.16–2.31)1.57 (1.20–2.04)1.03 (0.72–1.33)1.00
 P0.00020.0050.00080.90
SLE as underlying cause of death     
 n122155595269196
 Age-adjusted rate (95% CI)14.0 (11.1–16.9)8.9 (7.4–10.4)7.2 (6.6–7.8)4.7 (4.1–5.3)4.7 (4.0–5.4)
 RR (95% CI)2.37 (1.61–3.47)1.72 (1.21–2.45)1.59 (1.21–2.09)1.00 (0.74–1.37)1.00
 P<0.00010.0030.00070.97
SLE with renal disease among causes of death     
 n43512018360
 Age-adjusted rate (95% CI)4.7 (3.1–6.2)3.0 (2.1–3.9)2.5 (2.1–2.8)1.5 (1.1–1.8)1.6 (1.2–2.1)
 RR (95% CI)2.63 (1.31–5.26)1.79 (0.92–3.46)1.72 (1.04–2.86)0.97 (0.56–1.80)1.00
 P0.0070.090.040.98
Table 3. Mortality due to SLE among white men, by education level*
 Education
0–11 years12 years13 years or more
  • *

    Rates are deaths per million. SLE = systemic lupus erythematosus; RR = rate ratio; 95% CI = 95% confidence interval.

SLE as underlying, immediate, or contributing cause of death   
 n102188149
 Age-adjusted rate (95% CI)3.2 (2.4–3.8)2.9 (2.4–3.3)1.4 (1.1–1.7)
 RR (95% CI)2.13 (1.42–3.19)2.11 (1.51–2.96)1.00
 P0.0002<0.0001
SLE as underlying cause of death   
 n549682
 Age-adjusted rate (95% CI)1.8 (1.2–2.3)1.5 (1.1–1.8)0.7 (0.6–0.9)
 RR (95% CI)2.31 (1.40–3.81)2.03 (1.33–3.09)1.00
 P0.00090.0009
SLE with renal disease among causes of death   
 n254245
 Age-adjusted rate (95% CI)0.8 (0.5–1.2)0.6 (0.4–0.8)0.4 (0.3–0.6)
 RR (95% CI)1.81 (0.83–3.95)1.59 (0.82–3.07)1.00
 P0.140.17
Figure 1.

Relative odds of mortality due to systemic lupus erythematosus as the underlying, immediate, or contributing cause of death (hatched bars) and all-cause mortality (open bars) in persons age 25–64 years, by education level, in each sex–race subgroup. Odds ratios were adjusted for age. Persons in the highest education category were the reference group.

Among African American women, the age-adjusted mortality rates due to SLE were similar among those with 12 years, 13–15 years, and ≥16 years of education (Table 4). However, the mortality rates due to SLE were lower among those with 0–8 years or 9–11 years of education. African American women with 0–8 years of education were only 48% as likely to die of SLE as those with a college education, and those with 9–11 years of education were only 76% as likely to die of SLE as those with a college education. Results were similar in analyses of SLE as the underlying cause of death and for the subgroup with renal disease as a cause of death. These decreased risks of mortality due to SLE contrasted with the increased risk of all-cause mortality among less-well educated African American women (Figure 1).

Table 4. Mortality due to SLE among African American women, by education level*
 Education
0–8 years9–11 years12 years13–15 years16 years or more
  • *

    Rates are deaths per million. SLE = systemic lupus erythematosus; RR = rate ratio; 95% CI = 95% confidence interval.

SLE as underlying, immediate, or contributing cause of death     
 n68221655380217
 Age-adjusted rate (95% CI)33.4 (23.0–43.8)46.5 (40.0–53.1)59.2 (54.6–63.8)46.5 (41.5–51.5)59.6 (50.9–68.3)
 RR (95% CI)0.48 (0.31–0.73)0.76 (0.57–1.02)1.04 (0.82–1.31)0.82 (0.63–1.06)1.00
 P0.00060.070.760.12
SLE as underlying cause of death     
 n37138450241140
 Age-adjusted rate (95% CI)19.1 (10.1–26.8)30.4 (25.0–35.7)40.7 (36.9–44.5)28.2 (24.4–31.9)39.2 (32.0–46.4)
 RR (95% CI)0.44 (0.24–0.81)0.78 (0.53–1.15)1.12 (0.82–1.53)0.80 (0.57–1.12)1.00
 P0.0080.210.460.19
SLE with renal disease among causes of death     
 n14591558960
 Age-adjusted rate (95% CI)4.9 (1.4–8.4)12.2 (8.8–15.5)13.9 (11.6–16.1)11.1 (8.6–13.6)16.0 (11.5–20.4)
 RR (95% CI)0.34 (0.19–0.62)0.72 (0.50–1.03)0.87 (0.65–1.18)0.68 (0.49–0.95)1.00
 P0.00030.080.390.03

Results for African American men and Asian/Pacific Islander women demonstrated a pattern similar to that of African American women. Among African American men, the risk of SLE-related mortality was 30% lower among those with 0–11 years of education than among those with ≥13 years of education (adjusted odds ratio [OR] 0.70), whereas all-cause mortality was approximately twice as high in those with 0–11 years of education (Table 5 and Figure 1). Among Asian/Pacific Islander women, the risk of SLE-related mortality among those with 0–11 years of education was only 72% of the risk of those with ≥13 years of education, whereas the risk of all-cause mortality was slightly increased among those with 0–11 years of education (Table 6 and Figure 1).

Table 5. Mortality due to SLE among African American men, by education level*
 Education
0–11 years12 years13 years or more
  • *

    Rates are deaths per million. SLE = systemic lupus erythematosus; RR = rate ratio; 95% CI = 95% confidence interval.

SLE as underlying, immediate, or contributing cause of death   
 n498877
 Age-adjusted rate (95% CI)5.9 (4.0–7.7)9.3 (7.3–11.3)8.8 (6.6–11.1)
 RR (95% CI)0.70 (0.55–0.90)1.07 (0.87–1.29)1.00
 P0.0040.53
SLE as underlying cause of death   
 n285949
 Age-adjusted rate (95% CI)3.6 (2.1–5.2)6.1 (4.5–7.7)5.4 (3.7–7.2)
 RR (95% CI)0.71 (0.48–1.05)1.16 (0.85–1.58)1.00
 P0.090.36
SLE with renal disease among causes of death   
 n142626
 Age-adjusted rate (95% CI)1.6 (0.6–2.7)2.7 (1.6–3.8)2.9 (1.6–4.1)
 RR (95% CI)0.60 (0.32–1.12)0.91 (0.55–1.50)1.00
 P0.110.71
Table 6. Mortality due to SLE among Asian/Pacific Islander women, by education level*
 Education
0–11 years12 years13 years or more
  • *

    Rates are deaths per million. SLE = systemic lupus erythematosus; RR = rate ratio; 95% CI = 95% confidence interval.

SLE as underlying, immediate, or contributing cause of death   
 n244990
 Age-adjusted rate (95% CI)14.4 (8.0–20.9)22.0 (15.5–28.4)20.0 (15.7–24.3)
 RR (95% CI)0.72 (0.43–1.20)1.02 (0.69–1.52)1.00
 P0.210.92
SLE as underlying cause of death   
 n133061
 Age-adjusted rate (95% CI)8.3 (3.1–13.4)13.7 (8.6–18.9)13.2 (9.7–16.6)
 RR (95% CI)0.64 (0.35–1.16)0.97 (0.62–1.50)1.00
 P0.150.90
SLE with renal disease among causes of death   
 n51026
 Age-adjusted rate (95% CI)2.6 (0–5.0)3.6 (1.3–5.9)5.1 (3.0–7.2)
 RR (95% CI)0.49 (0.23–1.04)0.69 (0.39–1.22)1.00
 P0.070.21

To limit the effects of any upgrading of education levels to high school graduate, the data were also analyzed by comparing those with >12 years of education with those reported as having ≤12 years of education (Table 7). White women with ≤12 years of education were 1.64 times more likely to die of SLE than white women with >12 years of education. This risk was similar to the all-cause mortality risk of 1.51 associated with lower education levels in white women. Among white men, the risk of mortality due to SLE in those with ≤12 years of education (OR 2.12) was almost identical to the all-cause mortality risk (OR 2.10). However, among African American women (OR 1.03), African American men (OR 0.92), and Asian/Pacific Islander women (OR 0.90), there was no association between lower education levels and mortality due to SLE, even though in each group those with lower education levels had significantly higher risks of all-cause mortality (ORs of 1.38, 1.76, and 1.22, respectively). Results were similar for the subgroups with mortality due to SLE as the underlying cause and mortality due to SLE with renal disease.

Table 7. Relative risks of mortality due to SLE and all-cause mortality in persons with ≤12 years of education, compared with those with >12 years of education, by sex-race subgroup*
 White womenWhite menAfrican American womenAfrican American menAsian/Pacific Islander women
  • *

    Data are presented as age-adjusted rate ratios (95% confidence intervals). SLE = systemic lupus erythematosus

  • P < 0.0001.

  • P < 0.001.

  • §

    P < 0.05.

All-cause mortality1.51 (1.24–1.83)2.10 (1.62–2.73)1.38 (1.20–1.57)1.76 (1.29–2.40)1.22 (1.10–1.40)§
SLE as underlying, immediate, or contributing cause1.64 (1.34–2.01)2.12 (1.52–2.96)1.03 (0.85–1.25)0.92 (0.78–1.08)0.90 (0.66–1.23)
SLE as underlying cause1.68 (1.36–2.07)2.11 (1.44–3.12)1.12 (0.89–1.40)0.98 (0.72–1.33)0.84 (0.56–1.25)
SLE with renal disease1.81 (1.12–2.90)§1.66 (0.91–3.01)0.99 (0.83–1.18)0.78 (0.49–1.25)0.60 (0.32–1.12)

A possible explanation for the divergent association between education level and mortality risk due to SLE and its association with all-cause mortality in ethnic minorities, and for the differences in this association between ethnic minorities and whites, may be found in the distribution of education levels in these groups. In white women and white men, two-thirds of deaths due to SLE and two-thirds of deaths due to any cause occurred in persons with ≤12 years of education (Table 1). However, among African American women, African American men, and Asian/Pacific Islander women, there were relatively fewer deaths due to SLE occurring in those with ≤12 years of education than deaths due to any cause. For example, among African American women, 73.3% of all deaths occurred in persons with ≤12 years of education, but only 61.3% of deaths due to SLE occurred in this education group, representing an absolute decrease of 12% and a relative decrease of 20%. This suggests a selective absence of deaths due to SLE recorded among ethnic minorities with ≤12 years of education.

DISCUSSION

This national, population-based study indicates that education level has an important association with mortality due to SLE among whites. Compared with those with ≥16 years of education, the risk of mortality due to SLE was twice as high among white women with ≤8 years of education. Mortality risks decreased with increasing education level so that risks were similar between those with 13–15 years and ≥16 years of education. Among white women, the mortality risk gradient with education level was higher among those with renal disease, as might be expected if renal disease was a marker of more severe SLE, and if socioeconomic status was associated with the severity of SLE. Among white men, risks of mortality due to SLE were twice as high among those with ≤12 years of education, compared with those with >12 years of education. The association between education level and mortality due to SLE in whites paralleled that found for all-cause mortality, and is consistent with many previous studies that demonstrate that persons of higher socioeconomic status have lower mortality rates (1–15). These associations support the validity of the analytic approach used here, which merged information on education level from death certificates with census data to compute mortality rates by education level.

In contrast to the findings in whites, and in contrast to the associations between socioeconomic status and all-cause mortality in many prior studies, higher education levels were not associated with lower mortality due to SLE among African Americans or Asian/Pacific Islander women. In fact, lower education levels appeared protective of mortality due to SLE among African American women, and similar trends were evident among African American men and Asian/Pacific Islander women. This effect appeared to be specific for mortality due to SLE, because all-cause mortality in these groups was higher among those with lower education levels. There are several potential explanations for the difference. It is possible that SLE is more prevalent in ethnic minorities with higher education levels and more prevalent in whites with low education levels, and these differences in prevalence translated into the differences in mortality rates that were observed. However, this explanation seems unlikely, as there is no evidence that the prevalence of SLE varies by socioeconomic status differently in different ethnic groups. Postponement of mortality past age 65 may also influence these results, but for this explanation to account for the ethnic differences detected, one would need to postulate that high education levels lead to preferential postponement of mortality in whites with SLE, whereas low education levels lead to preferential postponement of mortality in ethnic minorities with SLE, which seems improbable. Similarly, the findings could be explained by high mortality before age 25 in ethnic minorities with <12 years of education, but low mortality before age 25 in whites with <12 years of education, in those with onset of SLE in adolescence or young adulthood. However, the generally low mortality in this age group makes this explanation unlikely (24, 40, 41).

It also is possible that the severity of SLE differs among ethnic groups. The interpretation that ethnic differences in severity account for the results would need to explain why low education levels would be associated with more severe SLE and higher mortality rates in whites but with less severe SLE and lower mortality rates in ethnic minorities. Similar associations were also present in the subgroup of patients with SLE and renal disease as a cause of death. The associations in this subgroup, which controls in part for the severity of SLE, suggest that differences in severity are unlikely to account for the findings. A differential association between education level and mortality by race would also conflict with the results of a cohort study that demonstrated similar socioeconomic effects on mortality in whites and African Americans with SLE (24).

A more likely explanation for the findings in African Americans and Asian/Pacific Islanders is underascertainment of deaths due to SLE in those with ≤12 years of education. This group was underrepresented among deaths due to SLE, with relative decreases of 20–25% compared with the proportion of deaths from all causes among persons of similar education levels in these ethnic groups. Such underrepresentation was not present among whites, for which the proportions with ≤12 years of education were identical in deaths due to SLE and deaths due to any cause. Because of this underascertainment, it may not be possible to report valid mortality rates due to SLE in African Americans and Asian/Pacific Islanders from this source.

Bias in the reporting of education level on death certificates results in more people with low education levels being classified as having completed high school (33). Although more common among decedents 65 years or older, this misclassification also occurs in younger age groups and is more common among African Americans. Although this upgrading may have impacted our examination of risk gradients by education level in ethnic minorities, racial differences in the associations between education level and SLE-related mortality on one hand, and between education level and all-cause mortality on the other hand, persisted when results were compared between those with some college and those with high school educations or less.

This study indicates that mortality due to SLE is higher among whites with lower socioeconomic status, as measured by educational attainment. It is difficult to draw conclusions about the association between education level and mortality due to SLE in other ethnic groups because of evidence suggesting possible underascertainment of deaths due to SLE in ethnic minorities with lower levels of education. This underascertainment may be due to more frequent misreporting of causes of death among ethnic minorities with low socioeconomic status, and specifically underreporting of SLE among these patients. However, of more concern is that this underascertainment could reflect the underdiagnosis of SLE in persons of low socioeconomic status, particularly ethnic minorities. Underdiagnosis would likely be associated with inadequate treatment. Population-based prospective cohort studies with active case findings are needed to investigate these possibilities, and to investigate the role of limited access to medical care in the diagnosis and treatment of SLE in ethnic minorities of lower socioeconomic status. These findings also suggest that SLE may be underrepresented in United States mortality statistics (29).

Ancillary