In this investigation, the theory of antigenic challenge in the development of autoimmune disease was supported for both infection and transfusion. The results suggest that chronic exposure to infectious agents and allogeneic blood components promotes immunomodulation. The nexus between inflammation, infection and immunomodulation (both reactivity and suppression) has been well documented [4
]. This is consistent with previous investigations implicating infection and inflammation as contributors to autoimmune disease [8
]. The significance of both transfusion and infection in the final models suggests that each of these factors may be important contributors to the onset of autoimmune disease in women. While red blood cell transfusion has been shown to increase the incidence of infection in randomized controlled trials [23
], our results suggest that transfusion may also affect the immune system via non-infectious mechanisms. CMS records do not capture every infection that may have occurred in the participants but our results may indicate that an infection serious enough to warrant medical attention is necessary for triggering autoimmune disease. It is also possible that an infection requiring medical services is a marker for more frequent infections in general. The results show that hospital stays per se
, or other factors associated with hospitalization, were not associated with the development of autoimmune disease; the significant predictors were specifically medical visits/stays in which an infection or a transfusion was recorded.
Previous studies of blood transfusion and rheumatoid arthritis reported contradictory results [30
]. In these studies, the ascertainment of blood transfusion was via recall of information. Such methods may be inadequate, since blood is often transfused during surgery when the patient is under anesthesia or during the post-surgical period when the patient is medicated and therefore, may be unaware of whether or not a transfusion occurred. In the United States, it is not required that hospitalized patients be specifically informed if they received a transfusion (and >80% of transfusions occur in the hospital [32
]). In our study, blood transfusion was documented in the medical record at the time when the subject was hospitalized, seen in the emergency department, visited an outpatient facility/clinic, in a skilled nursing facility or received home health care. Since transfusion is only administered by clinicians, dates of occurrence of the stay or visit were known.
Parity was not associated with autoimmune disease in older women. Guthrie and colleagues conducted a population-based study of parity and gravidity on the risk of rheumatoid arthritis [33
]. They found that parous women exhibited a reduction in the risk of rheumatoid arthritis within 1–15 years after delivery but there was no association between parity and rheumatoid arthritis more than 15 years post-partum. The underlying hypothesis was that fetal microchimerism may affect later risk in the mother [33
]. In our study, there was no association between parity and autoimmune disease in general, or rheumatoid arthritis. Since the mean age of onset of the autoimmune disease was 78 years, parous women in our cohort were more than 15 years post-partum and therefore, our results concur with the previous findings. A limitation was our inability to assess number of pregnancies in addition to parity. Fetal microchimerism is possible with each pregnancy and therefore, our assessment using parity alone could be less sensitive than a true measure of gravidity.
We found that younger women were at greater risk of developing autoimmune disease than older women. There are changes to the immune system with age, including reports of increased regulatory T cell levels with older age [34
]. Animal studies indicate both an increased number of regulatory T cells and an increased level of suppression per cell as age rises [35
]. Regulatory T cells function as suppressors of the immune response and their deficiency or defects in their function have been associated with several autoimmune diseases in humans [36
]. Therefore, the inverse association between age and autoimmune disease in our study could be partially explained by the age-related increase in regulatory T cells.
The prevalence of rheumatoid arthritis has been reported as 1.37% in women from Minnesota [38
] which is similar to our finding of 1.09% in a representative sample of older American women nationally. Worldwide, the prevalence of rheumatoid arthritis has varied from 0.3% to 2.0% in different populations of women [39
]. Data from the Third National Health and Nutrition Examination Study indicated that the prevalence of systemic lupus erythematosus was 100 per 100,000 women in year 2000 [40
]; results from our study indicate the prevalence to be 135 per 100,000 older women. In total, our investigation showed that 1.42% of older women in the United States had one of the eight autoimmune diseases under investigation. We chose not to include type 1 diabetes mellitus in this cohort because of the lower incidence in older women and the inability to adequately distinguish insulin-dependent type 2 diabetes mellitus from type 1 diabetes in the records. In addition, some autoimmune diseases, such as hemolytic anemia and immune thrombocytopenic purpura, were too rare to investigate in this cohort. Therefore, the 1.4% prevalence figure is likely to be less than the overall prevalence of autoimmune disease in older women. Of note, Cooper and colleagues summarized the evidence regarding prevalence of all autoimmune diseases (combined) with a correction for under-ascertainment and reported the prevalence as 7.6 to 9.4% for all age and gender groups combined [41
Unfortunately, we did not have genetic information on the subjects. Many autoimmune diseases have demonstrated a genetic component [42
], although some studies of concordance in twins indicate that the genetic factors do not completely explain the incidence of several autoimmune diseases [44
]. It is possible that the increased risk of autoimmune disease in Jewish women was due to genetic differences although undefined environmental exposures cannot be ruled out. However, this finding was incidental to the main hypotheses of interest and therefore, would benefit from replication in another study.
The association between heart disease and autoimmune diseases is consistent with evidence implicating an underlying inflammatory response [46
]. Prospective studies have shown an association between atherosclerosis and both inflammation and chronic infection [48
]. Moreover, this association appears general, in that infections of the respiratory tract, urinary tract and other types of infection have been associated with atherogenesis, with heterogeneity among organisms [48
]. Immune dysfunction has also been demonstrated in patients with end-stage renal disease [50
]. The degree of alteration in immune function in end-stage renal disease is quite extensive, resulting in increased rates of infection with chronic immunosuppression and increased chronic inflammation due to immunoactivation [50
]. We modeled end-stage renal disease and heart disease such that any history of these diseases was captured; the intent was to recognize the pro-inflammatory states that may be evident for many years prior to a physician’s diagnosis. Therefore, in our investigation, heart disease and end-stage renal disease were associated with autoimmune diseases, but without directionality; the extent to which one precedes the other was not known. It is notable, however, that our findings regarding infection and transfusion remained significant after adjustment for heart disease and end-stage renal disease. We did not have the statistical power necessary to assess interactions among these factors but this may be of interest in future studies.
While the underlying mechanisms for the elevated risk of autoimmune disease from infection and transfusion could not be fully explored in this investigation, it is important to note that both of these risk factors are modifiable and thus, to some extent, preventable. While the numbers of possible cases prevented with a 10% reduction in exposure were modest, the long-term disability and healthcare costs associated with autoimmune diseases can be extensive [51
]. There has been greater emphasis on reducing healthcare-associated infections in hospitals nationwide [52
]. Moreover, conservative approaches to the use of red blood cell transfusion (i.e., lower hemoglobin triggers) have been found to provide similar clinical benefit or less harm than traditional approaches [23
], prompting initiatives in blood management [53
]. Whether these efforts succeed in lowering the prevalence of such antigenic challenges and whether this translates into lower rates of autoimmune disease is yet to be demonstrated.