Abstract
Objective. To estimate incidence and prevalence rates of rheumatoid arthritis (RA) in the city of Buenos Aires (CABA), Argentina, using data from a university hospital–based health management organization.
Methods. Global, age-specific, and sex-specific incidence and prevalence rates were calculated for members of the Hospital Italiano Medical Care Program (HIMCP), age ≥ 18 years. Incidence study followed members with continuous affiliation ≥ 1 year from January 2000 to January 2015 until he/she voluntarily left the HIMCP, RA was diagnosed, death, or study finalization. Cases from the Rheumatology Section database, electronic medical records, laboratory database, and pharmacy database were filtered with the 2010 American College of Rheumatology/European League Against Rheumatism criteria. Prevalence was calculated on January 1, 2015, and standardized for CABA. Capture-recapture (C-RC) analysis estimated true population sizes.
Results. In the study period, incidence rates (cases per 100,000 person-yrs) were 18.5 (95% CI 16.7–20.4) overall, 25.2 (95% CI 22.4–28.0) for women, and 8.8 (95% CI 6.8–10.8) for men. Prevalence rates (percentage of RA cases in the sample population) were 0.329 (95% CI 0.298–0.359) overall, 0.464 (95% CI 0.417–0.510) for women, and 0.123 (95% CI 0.093–0.152) for men. Standardized CABA prevalence rate was 0.300 (95% CI 0.292–0.307). C-RC adjusted rates were almost the same as unadjusted rates.
Conclusion. This study’s incidence and prevalence rates are in the lower range of the rates found around the world. Our female to male prevalence ratio was 4:1. Our peak incidence age was in the sixth and seventh decades for both sexes.
Rheumatoid arthritis (RA) is a chronic autoimmune disorder in which the body’s immune system attacks joints, leading to their inflammation, erosion, and deformity. RA has incidence and prevalence rates with significant regional variation. Epidemiological data demonstrating regional variation contributes to our understanding of how genetic and environmental factors affect the development of RA in patients.
RA epidemiological data is lacking in Argentina and Latin America. Two previous studies reported only RA prevalence in Argentina: a 1999 study reported a prevalence rate of 0.197% for northwest Argentina1, and a 2010 study reported a prevalence rate of 0.94% for a pool of patients derived from the RA registry and a telephone survey as a secondary source, in a small city in the Buenos Aires province2.
To our knowledge, this paper is the first study of RA incidence in Argentina with inclusion of the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) criteria. Our aim was to determine the most recent incidence and prevalence of RA in a healthcare organization from the city of Buenos Aires (CABA), the largest populated metropolis in Argentina, using a methodology similar to those used in previous studies for other diseases3,4,5,6,7.
MATERIALS AND METHODS
Setting
The population studied was the membership of the Hospital Italiano Medical Care Program (HIMCP), a prepaid health maintenance organization (HMO) in CABA, Argentina. The HIMCP provides comprehensive medical and health services through 2 main hospitals and 24 peripheral outpatient centers to around 140,000 members primarily located in the urban areas of CABA. The city covers an area of 202 km2 and has a subtropical climate. It is located on the western bank of the Rio de la Plata and has a population of 2,890,151 inhabitants (2010 census). In all, 92% of the population is white and of European descent, and the remaining is a mixture of natives and other ethnicities8.
Argentina has a segmented health system consisting of 3 large sectors: public, private, and social security (the last 2 covering a population of nearly 18.3 million people distributed among close to 300 entities of varying scope and size)9. Beneficiaries of the private system can freely choose their HMO.
In this context, the HIMCP provides a private health system insurance. About 5%–7% of the population in these geographic areas is affiliated with private health system insurance.
Population
For incidence calculation, the population contributing time at risk was all HIMCP members over 18 years old, with continuous affiliation for at least 1 year from January 2000 to January 2015. Each person was followed up until he/she voluntarily left the HIMCP, a diagnosis of RA was established, death, or finalization of the study. For prevalence calculation, the denominator population was the number of active members over 18 years old on January 1, 2015.
Case ascertainment
Multiple methods for case finding were used to ensure complete ascertainment: (1) patients noted in the Rheumatology Section database; (2) patients found to be diagnosed with “arthritis” using searches of the following keywords registered in the list of problems in the electronic medical records: arthritis, rheumatoid arthritis, seronegative arthritis, seropositive arthritis, polyarthritis, seropositive rheumatoid polyarthritis, seronegative rheumatoid polyarthritis, polyarticular rheumatoid arthritis; (3) patients with positive results for rheumatoid factor (RF) and/or cyclic citrullinated peptide (CCP) antibodies in the laboratory database; and (4) patients purchasing medication commonly used in the treatment of RA as noted in the HIMCP pharmacy database: methotrexate, sulfasalazine, leflunomide, or biologic drugs (anti-tumor necrosis factor-α group, rituximab, abatacept, tocilizumab, tofacitinib). Medical records of all patients found were manually reviewed, and only patients older than 18 years and fulfilling the 2010 ACR/EULAR criteria for RA were included. Clinical data for disease description were obtained from this review of the medical records.
Statistical analysis
Global, age-specific, and sex-specific incidence and prevalence rates were calculated with 95% CI. For incidence calculation, the date of diagnosis was considered as that when the diagnosis first appeared in the clinical records, and cases had to have been enrolled with the HIMCP for at least 1 year before this date. Prevalence was estimated on January 1, 2015, and the denominator population was the number of HIMCP active members older than 18 years on that date (n = 135,750).
In addition, we used the direct method to estimate age- and sex-standardized prevalence rates of RA, with the CABA population from the 2010 national census as the reference population.
Capture-recapture (C-RC) methods estimated the completeness of case ascertainment when using multiple information sources. Log-linear models were used to estimate the true RA population size by evaluating the degree of overlap among 3 data sources: the Rheumatology Section database, the database of patients identified by keyword searches in the electronic medical records, and the pharmacy database. Note that all patients in the laboratory database were included in the other databases; therefore, it was excluded for our analysis. Although independence of data sources could not be completely assured, there is no direct relationship among the 3 databases used in our analysis.
Modeling was performed separately for incidence and prevalence data. The log-linear model was used to estimate the number of persons who were missed in the population. Eight hierarchical log-linear models were fit to the data: 1 model assuming independence among the 3 data sources, 3 models of pairwise interaction, 3 models of 2 pairwise interactions, and the so-called saturated model that adjusted for interaction between all 3 pairs of sources. The best-fitting model was determined by the Akaike information criterion and Bayesian information criterion. Based on the estimated undercount, revised C-RC estimates of incidence and prevalence were calculated.
RESULTS
Incidence
In the study period for the HIMCP, 349,775 persons contributed a total of 2,073,438 person-years. A total of 384 patients developed RA between January 2000 and January 2015. The incidence rate was measured as cases per 100,000 person-years. The overall incidence rate was 18.5 (95% CI 16.7–20.4). In more detail, 310 cases of new-onset RA were women (81%), with an incidence rate of 25.2 (95% CI 22.4–28.0). The incidence rate for men was 8.8 (95% CI 6.8–10.8). Age-specific incidence rates in both female and male patients peaked in the sixth and seventh decades of life (Figure 1); however, in the 61–80 age group, the incidence rate for women was almost double that for men. Further, while the incidence rate was 5.4 (95% CI 2.4–8.5) for the female 18–30 age group, no incident cases were found for the male 18–30 age group. Details for demographic and clinical features of incident cases are shown in Table 1.
Crude rheumatoid arthritis incidence rate by sex and age group for the Hospital Italiano Medical Care Program, 2000–2015. Error bars represent 95% CI.
Demographic and clinical features of rheumatoid arthritis cases at diagnosis in the Hospital Italiano Medical Care Program, Buenos Aires, Argentina, 2000–2015.
When we considered patients with available laboratory information, 66% of incident cases were seropositive (when only the patient’s level of RF was considered). However, 85% of incident cases were seropositive when we considered patients with either elevated RF or CCP, or both. The majority (44%) of incident cases overall had involvement of 4–10 small joints.
Prevalence
On January 1, 2015, 446 prevalent cases were identified from a denominator population of 135,750 HIMCP members. Prevalence rate was the total number of RA cases observed in the sample population, indicated as a percentage. Our study resulted in a prevalence rate of 0.329 (95% CI 0.298–0.359) overall. The female prevalence rate of 0.464 (95% CI 0.417–0.510) was almost 4 times the male rate of 0.123 (95% CI 0.093–0.152). In terms of age, RA was most prevalent in the 50s for women and the 70s for men (Figure 2). As the population ages, the prevalence rates for both sexes move closer together. Demographic and clinical features of prevalent cases are similar to those of incident cases, as shown in Table 1.
Crude rheumatoid arthritis prevalence rate by sex and age group for the Hospital Italiano Medical Care Program. Error bars represent 95% CI.
Standardization
Because of differences in sex and age demographics, we standardized our HIMCP results for CABA. For CABA, the overall prevalence rate was 0.300 (95% CI 0.292–0.307). The female prevalence rate was 0.463 (95% CI 0.451–0.475). The male prevalence rate was 0.099 (95% CI 0.093–0.105).
Capture-recapture
For incidence, C-RC analysis was done for the total population and yielded an estimated 1 missing patient with RA, resulting in an incidence C-RC–adjusted rate of 18.57 cases (95% CI 18.52–18.86) per 100,000 persons-years. For prevalence, C-RC yielded an estimated 1 missing prevalent patient with RA, resulting in a prevalent C-RC–adjusted rate of 0.329 (95% CI 0.3285–0.334). Figure 3 shows the overlaps in the number of cases identified by the different methods of case finding.
Overlaps of the number of rheumatoid arthritis cases identified by different methods of case finding in the Hospital Italiano Medical Care Program. PD: pharmacy database; EMR: electronic medical records with keyword searches; RD: Rheumatology Section database.
DISCUSSION
To our knowledge, ours is the first study of RA incidence in Argentina using the 2010 ACR/EULAR criteria for RA diagnosis by analysis of patient data from the HIMCP, an HMO. Results for the HIMCP are standardized for CABA because of the demographic differences between the 2 populations. The CABA prevalence rate is about 10% lower than that of the HIMCP, probably because of the greater percentage of elderly patients in the HIMCP. Moreover, while the female prevalence rate for both populations is similar, the CABA male prevalence rate is about 20% lower than the HIMCP male prevalence rate. The likely reason is that the ratio of young to elderly men is much greater in the CABA population than in the HIMCP population. However, the results of our study might not be applied to the rest of Argentina because of the demographic differences.
The RA incidence and prevalence rates found for the HIMCP are in the lower range of the rates found around the world (Table 210–19,20,21 and Table 31,2,10,12,13,16,18,20,22,23,24,25,26). Data are insufficient in Latin America to conclusively compare our rates to those of other Latin American countries. However, our RA prevalence of 0.329 is lower than the RA prevalence rate of 0.46 found in Brazil24, 1.24 in Cuba27, and 1.6 in Mexico23. Regional variability between countries and within Argentina could be because of the differences in genetics, infectious agents, medication, alcohol and tobacco consumption, education, methodology, case ascertainment, and population used for standardization. Further, the results of our study deviated from those of Soriano, et al’s study16, which also took place in the HIMCP. The study reported an RA incidence rate of 0.24/1000 person-years and a prevalence rate of 0.24%. This deviation could be because (1) the 2003 study analyzed a denominator patient population that was half the size of our study, (2) electronic medical records at the HIMCP and its associated databases were not established until 2001, and (3) the 2010 ACR/EULAR criteria were not used in that study period.
Incidence of rheumatoid arthritis around the world.
Prevalence of rheumatoid arthritis around the world.
Peak incidence for the HIMCP population was in the 61–70 and 71–80 age groups for both sexes. These data are important because they confirm information from previous studies that show RA to be a disease frequently occurring in the elderly, therefore challenging the concept that “elderly RA” is a different disease18,21.
Our study has some strengths. It was based on a large population that was followed for a long period of time. Multiple sources were used to assure complete case ascertainment. In fact, C-RC methods showed very similar results, suggesting that few cases were missed. Finally, our results were standardized for CABA using census data.
As in many other epidemiologic studies, our study has some limitations. It was an analysis based on clinical encounters with the medical system, so unless patients sought care, they would not be identified. The HIMCP data, however, showed that almost 90% of patients enrolled have at least 1 medical encounter per year (data not published). In addition, because of the fact that HIMCP is a healthcare plan in which “ill” patients are not incorporated, our prevalent cases may be reduced because patients with existing disease would not be allowed to enter. However, this limitation would not affect the incident cases because these would be defined by patients developing disease during the observation period. Another limitation for incidence estimation: the 2010 ACR/EULAR criteria include anti-CCP antibodies that were not commonly used for diagnosis in the earlier years of our study. This could result in an underestimation of global incidence.
We showed that RA incidence and prevalence rates in CABA, one of the largest populated cities in Latin America, are in the lower range of the rates found around the world. Our female to male prevalence ratio was 4:1. Our peak incidence age was in the sixth and seventh decades for both sexes. RA epidemiological data are much needed in Latin America for global comparison and study.
- Accepted for publication March 3, 2016.
