Incidence and Prevalence of Systemic Lupus Erythematosus Among Arab and Chaldean Americans in Southeastern Michigan: The Michigan Lupus Epidemiology and Surveillance Program

Abstract

Objectives. We assessed the burden of systemic lupus erythematosus (SLE) among Arab and Chaldean Americans residing in southeast Michigan.

Methods. For those meeting SLE criteria from the Michigan Lupus Epidemiology and Surveillance Registry, we determined Arab or Chaldean ethnicity by links with demographic data from birth certificates and with a database of Arab and Chaldean names. We compared prevalence and incidence of SLE for Arab and Chaldean Americans with estimates for non-Arab and non-Chaldean American Whites and Blacks.

Results. We classified 54 individuals with SLE as Arab and Chaldean Americans. The age-adjusted incidence and prevalence estimates for Arab and Chaldean Americans were 7.6 and 62.6 per 100 000, respectively. Arab and Chaldean Americans had a 2.1-fold excess SLE incidence compared with non-Arab and non-Chaldean American Whites. Arab and Chaldean American women had both significantly higher incidence rates (5.0-fold increase) and prevalence estimates (7.4-fold increase) than did Arab and Chaldean American men.

Conclusions. Recognizing that Arab and Chaldean Americans experience different disease burdens from Whites is a first step toward earlier diagnosis and designing targeted interventions. Better methods of assigning ethnicity would improve research in this population.

Population-based registries of systemic lupus erythematosus (SLE) cases have expanded the epidemiological knowledge about SLE. In addition to producing overall incidence and prevalence estimates, registries allow further investigation of the disease burden among various racial and ethnic groups.^1–3 The Michigan Lupus Epidemiology and Surveillance (MILES) program, a collaboration between the Centers for Disease Control and Prevention, the Michigan Department of Community Health, and the University of Michigan, developed a population-based registry for 2 southeastern Michigan counties—Wayne and Washtenaw—comprising approximately a quarter of the Michigan population and including the cities of Detroit and Ann Arbor in the catchment area.¹

Michigan has the second largest Arab and Chaldean American population in the United States after California.⁴ Chaldeans are Eastern Rite Catholics from Iraq; they speak a form of Aramaic rather than Arabic.⁵ Although many Arab and Chaldean immigrants first came to Michigan in search of better economic opportunities, their numbers continue to rise with the influx of refugees,⁵ and individuals from these ethnic groups now represent a significant portion of the Michigan population. The 2000 US Census estimated that approximately 151 493 Arab and Chaldean Americans reside in Michigan⁶; however, because Arab and Chaldean Americans are not federally recognized minority groups, obtaining accurate population estimates proves difficult, often leading to underestimates. Within the state of Michigan, Wayne County has the highest percentage of Arab and Chaldean Americans, and Washtenaw County has the fourth highest.⁴ Although Arab and Chaldean Americans constitute noteworthy minority groups in Michigan, only limited health data specific to these populations exist.⁷

Previous research addressing SLE among Arabs or Chaldeans has focused primarily on clinical and immunological manifestations of the disease for those living in Arab countries.^8–13 To date, no research has looked at the burden of rheumatic or autoimmune diseases in Arab or Chaldean Americans.¹⁴ We used the MILES registry, a large population-based public health surveillance registry, and a novel methodology to assess incidence and prevalence estimates of SLE among Arab and Chaldean Americans living in the United States.

METHODS

The MILES registry includes the data of individuals meeting American College of Rheumatology classification criteria for SLE^15,16 who had a primary residence in Wayne County or Washtenaw County during 2002–2005. The MILES program is described in more detail elsewhere.¹ For our analysis, data from an additional year of surveillance (2005) were available. In brief, we screened potential SLE cases from multiple sources, including hospitals, private physician offices, laboratories, Medicaid, and end-stage renal disease databases. Highly trained abstractors extracted sociodemographic (including race and ethnicity), clinical, and laboratory data from medical records.

Study rheumatologists subspecializing in SLE then reviewed the abstracted data as part of ongoing quality control and determined the likelihood of lupus diagnosis on the basis of expert opinion. This analysis was limited to 3 subsets of the MILES registry population: Arab and Chaldean Americans, non-Arab and non-Chaldean American Whites, and non-Arab and non-Chaldean American Blacks.

Data Links

We linked the MILES registry to Michigan birth certificate data from the Division for Vital Records and Health Statistics at the Michigan Department of Community Health. We linked SLE cases to electronic birth records from 1950 through 2010; birth records did not exist in electronic format before 1950. SLE cases could match as either the child, parent, or both on Michigan birth certificates. By linking to birth certificates, we obtained parent-reported race, ancestry, and country of birth information for matched cases. If the SLE case matched as a child on a birth certificate, we also extracted information on the parent’s race, ancestry, and country of birth. We matched approximately 66% of MILES registry cases to birth certificates. Because race and ethnicity data obtained from medical records are often missing or inaccurate and may not be self-reported, we considered the self-reported birth certificate data for race and ethnicity as the gold standard compared with these data from the MILES registry.

We also used a validated Arab and Chaldean surname database developed by the Department of Family Medicine at Wayne State University (WSU) in Detroit, Michigan to identify individuals of Arab and Chaldean American ethnicities.^17,18 This surname database was originally developed from statewide vital statistics data for 2000–2008 and was further validated to include data from 2009.¹⁸ We considered individuals Arab or Chaldean Americans if there was an unequivocal surname match or if there was an equivocal surname match and first name match.¹⁷

Unequivocal surnames indicated Arab or Chaldean ethnicity with greater certainty, whereas equivocal surnames (e.g., Joseph or Simon) were those that may indicate Arab and Chaldean American or non-Arab and non-Chaldean American ethnicity. Using surnames and first names from both the MILES database and birth certificates, we matched SLE cases to the WSU database. When possible, we matched women’s maiden names instead of their married surnames. For individuals matching as children to the birth certificate, we also matched the names of their parents against the database.

We considered individuals Arab and Chaldean Americans if they matched the WSU name-matching algorithm, had self-reported Arab ancestry on the birth certificate, or were born in a Middle Eastern or North African country (data available as a supplement to this article at http://www.ajph.org). We also considered names, ancestry, and birthplaces of parents when those with SLE matched as a child on the birth certificate. To minimize misclassification of names on the basis of religion rather than ethnicity, we did not consider individuals listing their country of birth as an East Asian country Arab and Chaldean Americans, consistent with previously published research that used the WSU name database.^17,19

We considered individuals non-Arab and non-Chaldean American Whites if either the MILES registry or birth certificate indicated race as White with no evidence for Arab or Chaldean ethnicity. Similarly, we considered individuals non-Arab and non-Chaldean American Black if either the MILES registry or birth certificate indicated race as Black with no evidence for Arab or Chaldean ethnicity. If we found a discrepancy, we again considered self-reported birth certificate data the gold standard.

Statistical Analyses

We estimated crude incidence and prevalence estimates as either new cases or cumulative cases divided by person–years, respectively. We calculated the corresponding binomial 95% confidence intervals (CIs). For the numerator, incident cases contributed person–years for the year of diagnosis, and prevalent cases contributed total person–years for each year starting from diagnosis. For the denominator, we used the 2000 US Census data (5% sample) with appropriate survey weights from the Integrated Public Microdata Series to compute race-, gender-, and age-specific denominators for Wayne and Washtenaw counties.²⁰

To calculate the denominators for Arab and Chaldean Americans, we included individuals who listed their race as White and also spoke Arabic at home or indicated an ancestry or birthplace in a country where Arabic is the official language. The Integrated Public Microdata Series data also provide demographic information for other persons in the household; we also included father’s or mother’s ancestry, birthplace, and languages spoken to determine population estimates for Arab and Chaldean Americans. We calculated denominator estimates for non-Arab and non-Chaldean American Whites from those indicating their race as White on the census minus the number of Arab or Chaldean Americans.

We calculated denominator estimates for non-Arab and non-Chaldean American Blacks from those indicating their race as Black on the census. To calculate age-standardized rates and associated exact 95% CIs, we used the direct method to adjust for age and applied 10-year age band weights from the 2000 US Standard Population.²¹ We computed median age at diagnosis and interquartile ranges among incident cases. We calculated age-specific prevalence estimates for Arab and Chaldean American women on the basis of their age at the middle of the surveillance period (December 31, 2003); small sample sizes restricted similar calculations for Arab and Chaldean American men. We assessed differences by gender and race/ethnicity through logistic regression models; an α level of < .05 denoted statistical significance. We performed all analyses using SAS version 9.2 (SAS Institute, Cary, NC).

RESULTS

Figure 1 illustrates the flow diagram of included cases. Of the 4012 fully abstracted potential SLE cases in the MILES registry, 2502 met the American College of Rheumatology definition for SLE. From January 2002 to December 2005, there were 549 unduplicated incident cases and 2379 unduplicated prevalent cases. After further restricting on race and ethnicity study criteria, the final study population included 2296 prevalent cases and 517 incident cases. We classified 54 (2%) SLE prevalent cases as Arab and Chaldean Americans: 8 men and 46 women. Of the 54 cases, 36 matched only to the WSU name lists, 6 matched with only birth certificate or medical record information, and 12 matched with both the name list and supplemental information. There were 886 (39%) non-Arab and non-Chaldean American Whites and 1356 (59%) non-Arab and non-Chaldean American Blacks (Figure 1).

FIGURE 1—

Flow diagram of systemic lupus erythematosus cases: Michigan Lupus Epidemiology and Surveillance Program (MIILES), Southeastern Michigan, 2002–2005.

Note. ACR =  American College of Rheumatology; SLE = systemic lupus erythematosus. Incident cases are a subset of the prevalent cases.

Incidence

Crude and age-standardized incidence rates by gender and race/ethnicity are presented in Table 1. Among the Arab and Chaldean American population, we identified 20 incident cases of SLE during the study period: 5 cases were Arab and Chaldean American men and 15 cases were Arab and Chaldean American women (Table 1). Overall, the age-standardized incidence rate for Arab and Chaldean Americans was 7.6 per 100 000 person–years (95% CI = 4.1, 11.2), which was similar to the rate for non-Arab and non-Chaldean American Blacks (8.8; 95% CI = 7.8, 9.8) and significantly higher than was the rate for non-Arab and non-Chaldean American Whites (3.7; 95% CI = 3.1, 4.2; P = .02).

TABLE 1—

Median Age at Diagnosis for Incident SLE Cases and Estimated SLE Incidence Rates by Race/Ethnicity and Gender: Michigan Lupus Epidemiology and Surveillance Program, Southeastern Michigan, 2002–2005

	No. of Unduplicated Cases	Median Age at Diagnosis, Years (IQR)	Crude Incidence per 100 000 (95% CI)	Age-Standardized Incidence per 100 000 (95% CI)
Arab and Chaldean American	20	34.2 (24.7–43.5)	6.5 (4.2, 10.0)	7.6 (4.1, 11.2)
Male	5	30.4 (10.3–34.7)	3.0 (1.3, 7.3)	2.7 (0.2, 5.2)
Female	15	34.2 (30.4–49.1)	10.4 (6.3, 17.3)	13.6 (6.4, 20.9)
Non-Arab and Non-Chaldean American White	190	41.8 (31.9–55.9)	3.8 (3.3, 3.4)	3.7 (3.1, 4.2)
Male	29	50.2 (28.5–58.9)	1.2 (0.8, 1.7)	1.1 (0.7, 1.6)
Female	161	41.3 (32.1–55.9)	6.3 (5.4, 7.4)	6.2 (5.2, 7.1)
Non-Arab and Non-Chaldean American Black	307	35.4 (25.7–47.3)	8.5 (7.6, 9.5)	8.8 (7.8, 9.8)
Male	36	32.0 (22.2–47.3)	2.2 (1.6, 3.0)	2.3 (1.5, 3.0)
Female	271	35.9 (26.5–47.2)	14.0 (12.4, 15.8)	14.2 (12.5, 15.9)

Note. CI = confidence interval; IQR = interquartile range; SLE = systemic lupus erythematous.

Incidence rates for Arab and Chaldean American men, non-Arab and non-Chaldean American White men, and non-Arab and non-Chaldean American Black men were 2.7 (95% CI = 0.2, 5.2), 1.1 (95% CI = 0.7, 1.6), and 2.3 (95% CI = 1.5, 3.0), respectively. Incidence rates among women were 13.6 (95% CI = 6.4, 20.9) for Arab and Chaldean American women, 14.2 (95% CI = 12.5, 15.9) for non-Arab and non-Chaldean American Black women, and 6.2 (95% CI = 5.2, 7.1) for non-Arab and non-Chaldean American White women. The age-adjusted incidence rate for Arab and Chaldean American women was 5.0 times higher than was that for Arab and Chaldean American men (13.6 vs 2.7; P = .02).

Similarly, incidence rates were 5.6 times higher for non-Arab and non-Chaldean American White women than for men and 6.2 times higher for non-Arab and non-Chaldean American Black women than for non-Arab and non-Chaldean American Black men. Median age at diagnosis (Table 1) for Arab and Chaldean Americans was 30.4 years for men (interquartile range = 10.3–34.7) and 34.2 years for women (interquartile range = 30.4–49.1).

Prevalence

Table 2 details crude and age-standardized prevalence estimates per 100 000 by gender and race/ethnicity. Overall prevalence estimates for Arab and Chaldean Americans (62.6; 95% CI = 51.9, 73.3) were similar to those for non-Arab and non-Chaldean American Whites (52.3; 95% CI = 50.4, 54.3) and lower than were those for non-Arab and non-Chaldean American Blacks (120.1; 95% CI = 116.3, 123.8). Among men, the prevalence estimate for Arab and Chaldean Americans was 16.2 (95% CI = 8.4, 24.0) compared with 10.0 (95% CI = 8.8, 11.2) for non-Arab and non-Chaldean American White men and 20.8 (95% CI = 18.5, 23.1) for non-Arab and non-Chaldean American Black men.

TABLE 2—

Prevalence Estimates by Race/Ethnicity and Gender for Systemic Lupus Erythematous Cases: Michigan Lupus Epidemiology and Surveillance Program, Southeastern Michigan, 2002–2005

	No. of Unduplicated Cases	Case–Years	Crude Prevalence per 100 000 (95% CI)	Age-Standardized Prevalence per 100 000 (95% CI)
Arab and Chaldean American	54	149	48.2 (36.1, 64.5)	62.6 (51.9, 73.3)
Male	8	19	11.5 (5.4, 24.8)	16.2 (8.4, 24.0)
Female	46	130	90.4 (66.0, 123.7)	120.5 (98.5, 142.6)
Non-Arab and Non-Chaldean American White	886	2748	54.6 (50.9, 58.6)	52.3 (50.4, 54.3)
Male	89	253	10.2 (8.1, 12.8)	10.0 (8.8, 11.2)
Female	797	2495	97.7 (90.7, 105.3)	93.8 (90.2, 97.5)
Non-Arab and Non-Chaldean American Black	1356	3984	110.5 (104.3, 17.1)	120.1 (116.3, 123.8)
Male	108	310	18.6 (15.1, 22.8)	20.8 (18.5, 23.1)
Female	1248	3674	189.7 (178.6, 201.5)	200.0 (193.5, 206.5)

Note. CI = confidence interval.

For women, estimates for Arab and Chaldean Americans, non-Arab and non-Chaldean American Whites, and non-Arab and non-Chaldean American Blacks were 120.5 (95% CI = 98.5, 142.6), 93.8 (95% CI = 90.2, 97.5), and 200.0 (95% CI = 193.5, 206.5), respectively. The age-adjusted prevalence estimate for Arab and Chaldean American women was 7.4 times higher than was that for Arab and Chaldean American men (120.5 vs 16.2; P < .001). For non-Arab and non-Chaldean American Whites, the rates for women were 9.4 times higher than were the rates for men, and for non-Arab and non-Chaldean American Blacks the women’s rates were 9.6 times higher than were the men’s rates. Figure 2 illustrates age-specific prevalence estimates by 10-year age bands for Arab and Chaldean American women. Estimates rose from early childhood, peaked during the 30s, and declined thereafter.

FIGURE 2—

Age-specific prevalence estimates by 10-year age bands for Arab and Chaldean American females: Michigan Lupus Epidemiology and Surveillance Program, Southeastern Michigan, 2002–2005.

Note. Symbol markers indicate point estimates. Bars indicate 95% confidence intervals.

DISCUSSION

Capturing Arab and Chaldean American ethnicity in population-based studies presents a challenge because these individuals are rarely categorized in medical records or population censuses as distinct from Whites. Through a novel combination of sociodemographic identifiers and a surname and first name list, we identified a cohort of Arab and Chaldean Americans within the population-based MILES registry and assessed their SLE burden.

Although health disparities in SLE burden between Blacks and Whites are well documented,^1,2,22,23 our results suggest that the pattern of SLE among Arab and Chaldean Americans is different and distinct from that of other racial or ethnic groups. Arab and Chaldean Americans had substantially higher (2.1-fold increase) incidence than did non-Arab and non-Chaldean American Whites. Arab and Chaldean American women also had a significantly higher prevalence of SLE than did non-Arab and non-Chaldean American White women. Gender-specific estimates for Arab and Chaldean Americans revealed that women had both significantly higher incidence (5.0-fold increased) and prevalence estimates (7.4-fold increased) than did men.

Similarly, we observed significant increases in prevalence and incidence estimates for women compared with men for non-Arab and non-Chaldean American Whites and non-Arab and non-Chaldean American Blacks. These results are consistent with previous findings of a higher SLE burden among women in all racial/ethnic groups than among men.^1,2,24 Finally, age-specific prevalence estimates for Arab and Chaldean American women were highest in premenopausal women (those younger than aged 53 years on average in the United States²⁵). Figure 2 indicates a steady increase in SLE burden with a peak in the 30s and a subsequent decline. This pattern is roughly similar to previous findings that age-specific prevalence for Black women and White women peak in the 40s.^1,2

Although age-specific incidence estimates for Arab and Chaldean Americans were unstable because of a small number of newly diagnosed cases, the median age of diagnosis of 34.2 years for Arab and Chaldean American women is similar to that of non-Arab and non-Chaldean American Black women in this population but substantially lower than that of White women in this study (41.3 years) and mean age for women reported in a large, United Kingdom–based SLE incidence study (47.3 years).²⁴ Our findings illustrate the importance of distinguishing Arab and Chaldean Americans from White Americans to better understand their specific patterns of disease onset and progression.

Recognizing that Arab and Chaldean Americans have a greater SLE burden than do non-Arab and non-Chaldean American Whites may aid in earlier disease diagnosis and prevention of typical SLE complications. One study conducted in the Middle East found that Arabs with SLE suffered from hypertension, renal failure, and cerebral vascular disease as typical complications from the disease, which the authors attributed to late disease presentation or delayed diagnosis.⁹ Similar research is necessary to determine whether the same complications are seen at the same frequency among Arab and Chaldean individuals with SLE living in the United States.

Limitations

Although we used caution in classifying cases and assembling population denominators, some limitations persisted, which may affect Arab and Chaldean American estimates. First, we could not match approximately one third of SLE cases in the MILES registry to a Michigan birth certificate, and thus no self-reported data on ancestry or country of birth were available. For cases not matched to a birth certificate, we used information extracted from the medical records into the MILES registry for classification. Our inability to fully capture ancestry or country of birth from birth certificates may underestimate the number of Arab and Chaldean Americans in the population with lupus and deflate incidence and prevalence estimates.

Second, in some cases, the maiden name for female SLE cases was unknown. We obtained all surnames associated with each of the SLE cases from vital records and medical records, and when possible, we used maiden names to determine Arab or Chaldean ethnicity. When maiden name was unknown, we matched a married surname to the WSU name list, which may have resulted in a misclassification of those women as Arab and Chaldean Americans and an overestimation of the number of Arab and Chaldean American cases and inflation of incidence and prevalence estimates.

Third, denominator estimates from the 2000 US Census may underestimate the population of Arab and Chaldean Americans because of immigration into Michigan during the study period (January 1, 2002–December 31, 2005) and misclassification. State-level estimates show the Arab and Chaldean American population growing by more than 20% between 2000 and 2008.⁴ Although annual American Community Survey data would have provided more accurate state-level estimates for the study period because of annual collection, American Community Survey data before 2005 do not capture county-level estimates. The 2000 US Census data, however, allowed county-level estimates in Wayne County and Washtenaw County on the basis of ancestry, birthplace, and language spoken. Underestimating the Arab and Chaldean American denominator population could inflate Arab and Chaldean American rate estimates and deflate non-Arab and non-Chaldean American White rate estimates.

Fourth, as with any population-based registry, case ascertainment may be incomplete and lead to underestimation of cases for all racial/ethnic groups; previous capture–recapture modeling determined an extremely low level of underascertainment of cases in the MILES data set.¹

Strengths

Our study also has several strengths. We used multiple sources of sociodemographic information—birth certificate data from vital records, medical record data from the MILES registry, and the WSU name lists—to identify Arab and Chaldean American SLE cases. By using vital records, we obtained self-reported data to supplement the medical record, which is often missing or misclassified. For individuals who matched as a child to a birth certificate, we also used parents’ information (name, race, birthplace, and ancestry) to classify cases as Arab and Chaldean Americans.

An additional strength was using the validated WSU name lists; we developed these lists on the basis of data for individuals residing in the metropolitan Detroit area. Because the MILES registry captured SLE cases from Wayne County and Washtenaw County, the geographic regions between the registry and the WSU name lists closely overlapped. Although name lists, such as those WSU developed, and other algorithms have previously identified Arab Americans in health statistics,^17,19,26 we expanded on that methodology to include other sources of sociodemographic information. We took advantage of MILES and a novel methodology to develop the first, to our knowledge, estimates for the burden of SLE among Arab and Chaldean Americans.

Conclusions

Our results highlight the value of classifying Arab and Chaldean Americans as distinct from Whites with regard to disease burden. Recognizing that Arab and Chaldean American individuals experience higher incidence and prevalence of certain diseases, in this case SLE, will allow earlier diagnosis and more effective targeted interventions to mitigate disease complications. Better capture of Arab and Chaldean American ethnicity in future studies and in potential denominators (census, electronic health records) will further clarify the impact of SLE in this population.