Case Definition

How cases are defined is essential to a study’s interpretation and comparability to other studies. The “gold standard” for diagnosing SLE is by clinical assessment from an experienced clinician (i.e., a rheumatologist), which is often impractical for population-based studies. However, relatively smaller geographical areas with a unified health care system are particularly amenable to this type of approach.

Classification criteria are designed to provide consistency across study populations and are appropriate for epidemiologic purposes. Several exist for SLE.³ The most universally accepted criteria have been those endorsed by the American College of Rheumatology (ACR). The former American Rheumatism Association, now known as the ACR, established the first classification criteria for SLE in 1971. These criteria, which included the LE cell test, allowed for standardized classification of patients. The earliest studies adhered generally to the 1971 criteria but did not strictly apply them.⁴ In 1982, the criteria were revised to include further advances in serologic testing—for antinuclear antibody (ANA) and anti–double-stranded DNA (anti-dsDNA)—as well as improved biostatistical techniques. In 1997, the Diagnostic and Therapeutic Criteria Committee of the ACR reviewed the 1982 criteria and recommended that a positive result of LE cell preparation be removed and replaced with the finding of antiphospholipid antibodies. These updates were based on committee consensus but were never subjected to rigorous validation testing.

Although the use of ACR criteria enhances the comparability of research studies, there are also drawbacks. Notably, the sensitivity of the 1982 criteria has been shown to be only 83% in an external population compared with 96% in the test population. Additionally, the criteria tend to be skewed toward limited detection of mild cases of SLE, or incident cases at early stages of their prodrome. Not only would the population size be underestimated with the criteria but the cases would also be biased toward those of longer disease duration and greater severity. Four of the 11 ACR criteria are overly biased toward cutaneous manifestations of SLE, even though every other organ system has one. The Systemic Lupus International Collaborating Clinics (SLICC) group has revised the ACR SLE classification criteria and validated alternative criteria in order to improve clinical relevance, meet more stringent methodology requirements, and incorporate new knowledge in SLE immunology since 1982.^4a It will be important to externally compare these and any new criteria with the existing ACR criteria. Cutaneous lupus does not have classification criteria. Biopsies of the skin may not be available and are not specific.

Each scientific advance that leads to improved laboratory techniques and/or any greater public awareness of the disease can make a profound impact on our ability to define and ascertain cases and, therefore, the rates of disease. Comparisons of earlier studies with later ones are difficult. Therefore, epidemiologic studies utilizing case definitions prior to the 1982 ACR criteria are presented separately (see Table 2-1).

TABLE 2-1 Early Population-Based Studies of Incidence and Prevalence of SLE

Case Ascertainment

Patients with SLE interact with the health care system at a variety of different points. So that the full spectrum of disease can be assessed, ascertainment of cases should come from a range of sources. A unified health care system with centralized health information has a potential advantage in that several different sources across a health system can be queried with relative efficiency, thereby improving case ascertainment. Furthermore, because lupus manifestations change over time for a particular patient or certain types of damage predominate, patients with SLE may be more likely to be captured at different types of facilities.

Administrative data are often used to find patients with potential lupus. Using such information can be an efficient way to ascertain cases throughout large health systems and different levels of care. However, the validity and accuracy of the data need to be better determined in different health systems. This can be achieved by utilizing multiple sources and adjusting for error in each.⁵ Self-reported physician diagnosis of SLE has been used but was found to be unreliable after a review of a sample of available medical records or asking whether the patients take lupus medications.^6,7 With the advent of electronic medical records in certain countries, there may be ways to query large systems with improved accuracy.

Rheumatologists and hospitals have been the common sources for many studies. However, nephrologists and dermatologists may potentially see high numbers of cases. Unless a study takes advantage of a relatively closed health system or administrative data, there is little consistency with regard to ascertaining cases from these physician groups. Cutaneous manifestations of SLE are quite common and may be addressed by dermatologists. Other specialists, such as hematologists, cardiologists, and obstetricians, may also encounter patients with SLE. But it is unclear how many new cases or additional clinical information could be derived from these specialists. University databases, though an important source of cases, can be biased toward patients with more severe disease. Access to clinical and pathology (particularly skin and renal) laboratories would be a high-yield source of locating additional patients. Those with end-stage renal disease are an important group but pose a unique challenge in epidemiologic studies. These patients often have less active systemic disease activity and can spend several years having other important comorbidities addressed. Therefore, their lupus-related information may be less likely to be found at a rheumatologist’s or outpatient nephrologist’s office, particularly as time goes by. Identifying presumably milder cases of SLE that are located in the primary care arena or that have not yet been diagnosed is challenging to address on a population-based level and requires different methodology.

Although a number of different sources may be utilized to find cases, the final result is likely to be an underestimate. Capture-recapture methodology of data analysis aims to correct for this by taking advantage of duplicate cases to mathematically determine the degree of overlap between the sources. The result is an alternative estimate that includes potentially missed cases and should be incorporated whenever possible.⁸

Although much of the epidemiologic data have been from the more developed Western nations, relatively little has been known about other countries (Tables 2-2 to 2-6). An enduring epidemiologic question is how African, Asian, and Hispanic ancestry influences the risk for SLE. There are no reports of significant rates of SLE in rural Africa or Asia, a situation that could, in part, be the result of underreporting due to a lack of health resources and expertise as well as other, more prevalent competing health issues. Further confirmation of the underlying indigenous rates of disease in these ethnic groups is needed. The strikingly increased rate in Americans of African ancestry has suggested that an SLE “prevalence gradient” exists,⁹ whereby genetic admixture and environmental factors are thought to raise the risk of SLE in people of African descent living in industrialized nations.

TABLE 2-3 Population-Based Studies of the Incidence and Prevalence of SLE in South America

TABLE 2-4 Population-Based Studies of the Incidence and Prevalence of SLE in Asia

TABLE 2-5 Population-Based Studies of the Incidence and Prevalence of SLE in Australia

The Community Oriented Program for the Control of Rheumatic Diseases (COPCORD) strategy was designed to be cost effective and has been implemented in various regions throughout the world in the past three decades. This validated method utilizes multistage random probability sampling to select participants, who are visited in their households by trained staff and administered a survey. Rheumatologists then examine and confirm the diagnosis in participants with suspected rheumatic disease. This strategy has the advantage of being able to compare prevalence rates in various parts of the country where it has been performed. It also engages participation of community lay workers and brings rheumatologists out in the field to experience the burden of disease at the community level.

Population at Risk

The population from which incidence and prevalence rates are being determined and the geographic area they live in should be well defined. Significant numbers of people at risk for the disease should be captured. It would not be appropriate for countries with a heterogeneous mix of ethnicities, like the United States, to extrapolate national rates on the basis of a single study, particularly those with few cases and large confidence intervals. Multiple studies are needed in areas that contain different high-risk ethnic populations. This “sampling” of rates could be used to determine more accurate national estimates.

Special attention should be given to different ethnic groups with appropriate stratified rates. Data summarized in the mid-2000s showed that some of the lowest rates are seen among Caucasian Americans, Canadians, and Spaniards with incident rates of 1.4, 1.6, and 2.2 cases per 100,000 people, respectively.¹⁰ In predominantly Caucasian populations, who tend to have less severe and longer duration of disease, prevalence rates may be underestimated if milder cases or patients in remission are not ascertained. The data from Northern European countries are excellent resources for understanding the burden of SLE in this group. Incident and prevalence rates are consistently higher among those of African, Hispanic, or Asian descent in studies from different countries. In England, for example, the annual incidence rate in people of Afro-Caribbean ethnicity has been reported to be 31.9/100,000 for both genders in Nottingham, and 25.8/100,000 for females in Birmingham, whereas whites showed rates of 3.4/100,000 and 4.3/100,000, respectively.^11,12

Recent migration patterns may also introduce biases that are rarely accounted for in population-based studies. In most European countries, where the net migration rate is usually ±1% per year, the overall prevalence estimates may not appear to be affected significantly by migration. However, they may be susceptible to the “healthy migrant effect.”¹³ An area of south London showed much higher prevalence of SLE in recent immigrants from West Africa (110/100,000) than in European women (35/100,000), although the prevalence in West African women was lower than in Afro-Caribbean women living in the same area (177/100,000). Because most West Africans migrated as adults, individuals with SLE are thought to have been less likely to leave their countries. On the other hand, Afro-Caribbeans were predominantly either born in the United Kingdom or had migrated as children and so their migration patterns were less likely to have been influenced by the disease. Continued epidemiologic evaluation and surveillance of the prevalence rate of SLE in the second generation of West African immigrants could help confirm this migration bias. In the United States, the undocumented Hispanic population is mobile and may move depending on a variety of different factors (economic, legislative, etc.). Immigrants with SLE may cluster in areas with better access to medical care. On the other hand, their numbers may be underestimated because they have more barriers to health care in general than the native population. Evaluation of migration rates should be considered and, when possible, adjusted for.