Hospital admissions for vitamin D related conditions and subsequent immune-mediated disease: record-linkage studies

Background Previous studies have suggested that there may be an association between vitamin D deficiency and the risk of developing immune-mediated diseases. Methods We analyzed a database of linked statistical records of hospital admissions and death registrations for the whole of England (from 1999 to 2011). Rate ratios for immune-mediated disease were determined, comparing vitamin D deficient cohorts (individuals admitted for vitamin D deficiency or markers of vitamin D deficiency) with comparison cohorts. Results After hospital admission for either vitamin D deficiency, osteomalacia or rickets, there were significantly elevated rates of Addison’s disease, ankylosing spondylitis, autoimmune hemolytic anemia, chronic active hepatitis, celiac disease, Crohn’s disease, diabetes mellitus, pemphigoid, pernicious anemia, primary biliary cirrhosis, rheumatoid arthritis, Sjogren’s syndrome, systemic lupus erythematosus, thyrotoxicosis, and significantly reduced risks for asthma and myxoedema. Conclusions This study shows that patients with vitamin D deficiency may have an increased risk of developing some immune-mediated diseases, although we cannot rule out reverse causality or confounding. Further study of these associations is warranted and these data may aid further public health studies.


Background
Immune-mediated diseases cumulatively represent one of the most common chronic disease groups in medicine today, affecting approximately 10% of first world populations [1]. There is substantial evidence in support of these disorders being determined by both genetic and environmental factors. One candidate environmental risk factor implicated in immune-mediated disease susceptibility is vitamin D deficiency [2][3][4].
Historically, vitamin D was thought to play a restricted role in calcium homeostasis; however, a wealth of studies now suggests that it exerts more widespread effects [5,6]. Functional laboratory studies have shown that vitamin D can modulate the immune response [5,6] and a number of epidemiological findings have implicated the involvement of vitamin D deficiency in the risk of developing immune-mediated diseases [2][3][4]. These include the ecological findings of the prevalence of diseases, such as multiple sclerosis (MS), type 1 diabetes, inflammatory bowel disease, rheumatoid arthritis and Sjogren's syndrome, positively correlating with latitude and reduced ultraviolet radiation exposure (the primary determinant of vitamin D levels) [1]. Further support comes from studies showing low serum vitamin D levels in patients suffering from a wide range of immune disorders, including MS [7], type 1 diabetes [8], systemic lupus erythematosus (SLE) [9] and rheumatoid arthritis [10]. These findings may, however, be biased by reverse causation [11].
To investigate further any association between vitamin D and immune-mediated disease, we undertook record linkage studies to determine the risk of immune-mediated disease in individuals admitted for vitamin D deficiency or a marker of vitamin D deficiency (rickets or osteomalacia) using an English national linked Hospital Episode Statistics (HES) dataset.

Population and data
We used a linked English national dataset of hospital admissions (Hospital Episode Statistics (HES)) and mortality. HES data are records of hospital care that are compiled for every episode of day case care or hospital admission in all English National Health Service (NHS) hospitals, and were supplied by the English national Information Centre for Health and Social Care. The mortality data were derived from death certificates and were supplied by the Office for National Statistics. The linked dataset used in this study, in which successive records for each individual were linked together, was constructed by the Oxford record linkage group.
The methods of analysis were the same for all vitamin D-related admissions and immune-mediated diseases; we describe the methods for rickets and Crohn's disease as the example. A cohort of people with rickets was constructed for those with a diagnosis of rickets as a reason for hospital care, by identifying the first episode of day case care, or admission, for rickets during the study period. A reference cohort was constructed by identifying the first admission for each individual with various other, mainly minor medical and surgical, conditions (listed in the Table 1 legend), as in previous studies of disease associations [12]. Standard epidemiological practice was followed by selecting a diverse range of conditions rather than relying on a limited range (in case the latter are themselves atypical in their risk of immune-mediated disease) [12]. As a check, we studied the risk of immune-mediated disease in the control conditions within the reference cohort to ensure that the reference cohort did not include control conditions that have atypically high or low immune mediated-disease rates. For some immune-mediated diseases (for example, rheumatoid arthritis), we removed some control conditions (for example, hip/knee replacement) from the reference cohort where we considered that this may otherwise have skewed the findings. For the fairly small number of associations affected by this, we show the 'adjusted' associations (that is, with some control conditions removed) for these diseases in the main paper; for completeness and comparison, the unadjusted associations are also provided (see below). Anyone with both an 'exposure' disease, for example, Crohn's disease, and a reference cohort condition was included in the exposure cohort and excluded from the reference cohort.
People were included in the rickets or reference cohort if they did not have an admission for an immunemediated disease either before or at the same time as the admission for rickets or the reference condition. The database was then investigated for any subsequent NHS hospital care for, or death from, Crohn's disease in these cohorts. We considered that rates of Crohn's disease in the reference cohort would approximate those in the general population while allowing for migration as data on migration of individuals were not available.
The analysis was performed using a suite of programs developed 'in house' using SAS 9 software (SAS Institute, Cary, NC, USA).

Ethical approval
The construction and analysis of the datasets were undertaken with the approval of the Central and South Bristol Research Ethics Committee (REC, reference 04/ Q2006/175).

Statistical methods
Rates of Crohn's disease were calculated based on personyears. Date of entry into each cohort was the date of the first admission for rickets, or reference condition, and date of exit was the date of the first record of Crohn's disease, death or the end of data collection (28 February 2011), whichever was the earliest. We first calculated rates for Crohn's disease, stratified and then standardized by age (in five-year age groups), sex, calendar year of first recorded admission, region of residence, and quintile of patients' Index of Deprivation score (a standard English measure of socio-economic status). The indirect method of standardization was used, with the combined rickets and reference cohorts as the standard population. We applied the stratum-specific rates in the standard population to the number of people in each stratum in the rickets cohort and then, separately, to those in the same stratum in the reference cohort, to obtain the expected number of people with Crohn's disease in each stratum of the rickets and reference cohort. Observed and expected numbers were then summed across all strata to give totals for all strata combined. Rate ratios were calculated by taking the standardized rate of occurrence of Crohn's disease in the rickets cohort relative to the reference cohort using the where O and E are the observed and expected numbers of Crohn's disease cases in the rickets and reference cohorts, respectively. In each table, we only show diseases in which either the observed or expected number, or both were five or more. The confidence interval for the rate ratio of Crohn's and χ 2 statistics for its significance were calculated as described elsewhere [13].

Results
The number of people in the cohort with vitamin D deficiency was 13,260 (71.5% female), osteomalacia 5,191 (69.4% female), and rickets 1,228 (59.5% female). Age distributions are shown in Table 1. There were more than 8.6 million people in the reference cohort. There were significantly elevated risks of Addison's disease, ankylosing spondylitis, autoimmune hemolytic anemia, chronic active hepatitis, celiac disease, Crohn's disease, diabetes mellitus, pemphigoid, pernicious anemia, primary biliary cirrhosis, rheumatoid arthritis, Sjogren's syndrome, systemic lupus erythematosus, thyrotoxicosis; and significantly reduced risks for asthma and myxoedema after hospital admission for either vitamin D deficiency, osteomalacia or rickets ( Table 2).
There were significantly elevated risks of celiac disease, pernicious anemia and thyrotoxicosis after hospital admission for rickets (Table 3). Celiac disease had a substantially increased risk.
There were significantly elevated risks of Addison's disease, celiac disease, diabetes mellitus, Sjogren's syndrome and thyrotoxicosis; and significantly reduced risks for asthma and myxoedema after hospital admission for osteomalacia ( Table 4). The rate ratios for Addison's disease, and celiac disease were particularly high. There were significantly elevated rates of Addison's disease, celiac disease, Crohn's disease, diabetes mellitus, pemphigoid, pernicious anemia, primary biliary cirrhosis, rheumatoid arthritis, Sjogren's syndrome, SLE and thyrotoxicosis; and significantly reduced rates for asthma and myxoedema after hospital admission for vitamin D deficiency (Table 5). Again, the rate ratios for Addison's disease and celiac disease were particularly high, as was the rate ratio for SLE.
Unadjusted associations when done, see Methods, are shown in Table 6.

Discussion
We present evidence associating vitamin D deficiency (or proxies of vitamin D deficiency) with risks of or protection against developing subsequent immune-mediated disease. Previous studies have linked inherited forms of rickets with MS [14] and type 1 diabetes [15], but no study has attempted an analysis similar to the one we present here. Addison's disease and thyrotoxicosis were significantly associated in all analyses, and asthma was significantly low in all but one. The combined analysis found associations with a greater number of immunemediated diseases -a lack of significance in the individual rickets, osteomalacia or vitamin D deficiency cohorts may reflect the lower power of these individual analyses.
The associations described warrant further attention. Mechanistically, it is possible that vitamin D deficiency predisposes to the development of immune-mediated disease by perturbing the immune response [2]. Another explanation is reverse causality -that the disease (either sub-clinically or clinically present but not recorded with a hospital admission prior to the vitamin D deficiency Hip replacement and knee replacement were excluded from the reference cohort. 2 Nasal polyp and deflected septum were excluded from the reference cohort. 3 Hemorrhoids, appendectomy and cholelithiasis were excluded from the reference cohort. 4 Cataracts were excluded from the reference cohort. 5 Limb fractures, dislocations, superficial injury, squint and head injury were excluded from the reference cohort. Diseases studied, but with fewer than five observed cases: dermatomyositis, Goodpasture's syndrome, myasthenia gravis, pemphigus, polyarteritis nodosa. admission) reduces vitamin D levels as a result of an inflammatory state and/or causes the individual to spend less time outdoors [16]. The increasing awareness of vitamin D may also lead to an increase in requests for vitamin D testing, and subsequent diagnosis of vitamin D deficiency [17]. However, under either of these hypotheses the apparent protective effects of vitamin D deficiency on asthma and myxoedema are not easy to reconcile. There are reports, however, that state that higher levels of vitamin D in childhood may increase the risk of asthma [18,19]. For disorders such as celiac disease there may be an issue with regards to intake and/or malabsorption of vitamin D; with Addison's disease, glucocorticoid deficiency may lead to suppression of the parathyroid hormone-vitamin D axis [20]. The study has limitations. It is not a prospective cohort with follow-up from the date of "first ever" diagnosis, but just from "first recorded" diagnosis in a hospital day case or inpatient record. Data are not recorded on patients who move out of the area covered by data collection or who are treated in hospitals outside the area. The dataset is limited to people who were admitted to hospital, or who received day case specialist care, and thus there exists the potential for selection bias. There is very limited information on potential confounding factors, such as detailed socioeconomic characteristics, ethnicity and smoking. The effect of making multiple comparisons needs to be considered. It is possible that some of the associations that are significant may result from making multiple comparisons and the play of chance. To aid assessment of this, we have provided P-values so that readers can judge where P-values are extreme: for example, with P-values as high as <0.001 for the associations between the vitamin D related conditions, combined, and celiac and Crohn's disease, the associations are unlikely to be chance alone.