Risk for Active Tuberculosis in Inflammatory Bowel Disease Patients

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:1070 –1075

Risk for Active Tuberculosis in Inflammatory Bowel Disease Patients FATEN N. ABERRA,*,§ NICOLAS STETTLER,‡,§ COLLEEN BRENSINGER,§ GARY R. LICHTENSTEIN,* and JAMES D. LEWIS*,§ *Division of Gastroenterology, University of Pennsylvania; ‡Division of Gastroenterology, Hepatology, and Nutrition, Children’s Hospital of Philadelphia; and §Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, Philadelphia, Pennsylvania

Background & Aims: The primary aim of this study was to determine the incidence of active tuberculosis in an inflammatory bowel disease population compared with the general population before the availability of infliximab. Methods: We performed a retrospective cohort study with the General Practice Research Database from January 1988 – October 1997. Ulcerative colitis and Crohn’s disease subjects with a minimum of 1 year of follow-up were matched to randomly selected subjects from the remaining population on year of birth (ⴞ5 years), sex, and primary care practice at ratio of 1:4. Active tuberculosis was determined by tuberculosis diagnostic codes. The incidence of active tuberculosis in the inflammatory bowel disease population and the relative risk for active tuberculosis in inflammatory bowel disease population compared with the general population were calculated. Multivariate logistic regression analysis was performed to adjust for confounders. Results: There were 16,213 inflammatory bowel disease subjects and 66,512 control subjects. The annual incidence of active tuberculosis was 20/100,000 in inflammatory bowel disease subjects compared with 9/100,000 in control subjects, yielding an unadjusted relative risk for active tuberculosis of 2.36 (95% confidence interval, 1.17– 4.74). Adjusting for confounders, corticosteroid use and smoking, the odds ratio of inflammatory bowel disease for active tuberculosis was 1.88 (95% confidence interval, 0.68 –5.20). Conclusions: During the pre-infliximab era, inflammatory bowel disease subjects appeared to be at higher risk for active tuberculosis than the general population, with immunosuppressant medications likely the main reason for this increased risk.

U

lcerative colitis (UC) and Crohn’s disease (CD) are chronic inflammatory bowel diseases (IBDs). For decades, it has been hypothesized that IBD might result from abnormal host response environmental pathogens, resulting in excessive activation of the enteric mucosal immune system. To control this excessive immune response, immunosuppressant medications are routinely prescribed. As the armamentarium of immunosuppressive medications has expanded, the potential hazards of such therapy have become more apparent. The introduction of anti–tumor necrosis factor–alpha (TNF-␣) therapies resulted in many cases of active tuberculosis (TB) among patients with CD and rheumatoid arthritis.1 More recently, 3 patients treated with natalizumab, one with CD, were diagnosed with progressive multifocal leukoencephalopathy, a rare disease caused by reactivation of JC virus infection.2 Although these events have refocused the attention of physicians who treat patients with IBD on the potential complications of novel

therapies, data on the risk of infection with first and second line therapies for IBD are incomplete. In this study, we have examined the risk of active TB among patients treated for IBD during the era before use of anti–TNF-␣ therapies. The primary objectives were to determine the incidence of active TB in IBD patients, and whether IBD patients were at increased risk of active TB compared with non-IBD patients before the availability of infliximab by using a population database.

Methods Study Design We conducted a retrospective cohort study by using the General Practice Research Database (GPRD), a United Kingdom (UK) outpatient clinical database encompassing England, Wales, Northern Ireland, and Scotland. Established in 1988, as of 1997 the system includes data on more than 8,000,000 patients in total, representing 6% of the UK population.3 Of note, approximately 98% of the UK population are registered with a general practitioner who is responsible for most of the medical care of the patient.4 Several studies have shown that the clinical information in the computer record is sufficiently accurate for use in epidemiologic studies, including studies of IBD.5,6

Study Population The cohort included subjects enrolled in the GPRD from January 1988 –October 1997, before the availability of infliximab. This cohort has been used in several prior studies and is described in detail elsewhere.7–9 All eligible patients in this cohort are those who have been in the GPRD for a minimum of 1 year after a patient is registered by a GPRD physician, and the data are “up to standard.” General practitioners must undergo formal training, must show competency in entering data into the electronic medical record, and maintain adequate recording of follow-up visits verified by monthly audits for data in the GPRD to be considered “up to standard.”7

Exposure of Interest Exposed subjects were those with a coded diagnosis of UC or CD. UC and CD diagnostic codes were previously valiAbbreviations used in this study: CD, Crohn’s disease; CI, confidence interval; GPRD, General Practice Research Database; IBD, inflammatory bowel disease; IQR, interquartile ratio; RR, relative risk; TB, tuberculosis; TNF, tumor necrosis factor; UC, ulcerative colitis; UK, United Kingdom. © 2007 by the AGA Institute 1542-3565/07/$32.00 doi:10.1016/j.cgh.2007.04.007

September 2007

dated.7,9,10 Subjects with less than 1 year of follow-up after a diagnosis of UC or CD were excluded in this study cohort to reduce the possibility of a misdiagnosis and reduce the possibility of prevalent cases of active TB.11 The completion of follow-up was defined as a diagnosis of active TB, exit from the electronic database, practice no longer up to standard for data entry, or any cause of death.

Selection of Controls The unexposed group was composed of randomly selected subjects from the remaining population with no diagnosis of IBD as well as other possible related diagnoses such as proctitis. The unexposed subjects were matched to the exposed subjects by year of birth (⫾5 years), sex, and primary care practice with a ratio of exposed to controls of 1:4. The completion of follow-up was defined as a diagnosis of active TB, exit from the electronic database, practice no longer up to standard for data entry, or any cause of death.

Definition of Clinical Outcome of Interest The first recorded diagnosis of TB after the start of follow-up was used to define the date of diagnosis of active TB (Supplementary Table 1; Supplementary material available online at www.cghjournal.org.). Codes referring to exposure to TB and/or evaluation for TB were not included. Patients were excluded if they had a diagnosis of human immunodeficiency virus infection. Subjects with gastrointestinal TB were also excluded, because this could have been confused with IBD or vice versa. In addition to exposure and outcome data, information was also recorded on possible effect modifiers or confounding factors such as age, sex, general practice location, the length of follow-up, diabetes mellitus, end-stage renal disease, history of gastrectomy, immigrant status, smoking history (current smokers vs nonsmokers), weight, height, oral immunosuppressant medications, and oral immunomodulator medications. General practice location is designated by a code and was used as a surrogate for adjusting for socioeconomic status. Smoking history was categorized as current smoking vs current nonsmoking. Smoking status has been previously validated in the GPRD, with the current smoking status as the most valid descriptor with a sensitivity and positive predictive value of 78% (95% confidence interval [CI], 52–94) and 70% (95% CI, 46 – 88), respectively.12 The oral immunosuppressant and immunomodulator medications recorded were prednisone, cyclosporine, azathioprine, 6-mercaptopurine, and methotrexate. Data were collected on exposure to these medications 3, 6, 9, and 12 months before the date of the outcome or end of followup. Body mass index (kg/m2) was calculated from weight and height data and used to determine nutritional status in adults. Underweight has been defined as a body mass index ⱕ18.5 kg/m2 in agreement with the World Health Organization criteria.13

Statistical Methods For the initial descriptive analysis, continuous variables were described with mean and standard deviation or median and interquartile (IQR) range if the data were not normally distributed. Categorical variables were described by using proportions and 95% CIs.

RISK FOR ACTIVE TB IN IBD PATIENTS

1071

To determine the incidence of active TB in a cohort of IBD and non-IBD subjects, incidence density was calculated (cases/ patient year) with 95% CI. To determine the relative rate (relative risk [RR]) of active TB in IBD subjects compared with non-IBD subjects, Poisson regression was used. We qualitatively compared the rate of TB in the control population with reported rates in the UK as a measure of completeness of reporting. Potential confounding variables were assessed with Poisson regression, and these variables were diabetes, end-stage renal disease, history of gastrectomy, immigrant status (immigrant vs nonimmigrant), smoking history (current smoking vs current nonsmoking), underweight (body mass index ⱕ18.5 kg/m2), and the immunosuppressant/immunomodulator medications: corticosteroids, cyclosporine, azathioprine, 6-mercaptopurine, and methotrexate. Interaction was assessed for all variables individually, comparing the relative rate of active TB for each stratum by using the test for homogeneity. If interaction was not present, then any variable changing the crude RR ⬎10% was included in the final multivariate regression model. The analyses were repeated separately for UC and CD. The analyses were also completely separate in children, defined as ⬍18 years of age, and adults ⱖ18 years of age.

Sample Size A sample size of 64,239 person-years exposed and 266,238 person-years unexposed was available from January 1988 –October 1997. Assuming a power of 0.80, an alpha error of 0.05, and a baseline incidence of TB of 0.0001, the minimal RR detectable is 2.5.14,15

Results There were 16,213 IBD subjects (9964 subjects with UC and 6249 with CD) and 66,512 controls. Characteristics of the population are described in Table 1. The median age of the control and IBD group was 46 years (IQR, 1–95 and 4 –93, respectively), and 47% were male in both groups. Two percent of the control group and IBD group were ⬍18 years of age. IBD and control subjects were from 2419 different practices. Median follow-up times of IBD subjects and control subjects were 3.4 years (IQR, 0.01–9.34) and 4.2 years (IQR, 0.01–9.34), respectively. The prevalence of diabetes, end-stage renal disease, and history of gastrectomy was low in both IBD subjects and controls, ranging from 0.2%–3% and 0.1%–2.2%, respectively. There were no recorded immigrants in the IBD group, and 4 immigrants were recorded in the control group. Among the control population, 23 subjects were diagnosed with TB. The annual incidence rate was 9/100,000. This is comparable to the annual rate of TB reported in the UK during the 1990s of 10.9 –11.8/100,000, which also includes subjects infected with human immunodeficiency virus.14,16 Twelve IBD subjects were diagnosed with active TB (4 CD and 8 UC subjects) for an incidence rate of 20 · 100,000⫺1 · year⫺1 vs 23 control subjects with active TB for an incidence rate of 9 · 100,000⫺1 · year⫺1. All TB cases were in adults with a median age of 63 years and range from 19 – 86 years. In IBD patients with TB the median age was 55 years (range, 25–75 years), and in control subjects with TB the median age was 69 years (range, 19 – 86 years). The unadjusted RRs for active TB in subjects with IBD, CD, and UC compared with controls were 2.36 (95% CI, 1.17– 4.74), 2.09 (95% CI, 0.72– 6.04), and 2.52 (95% CI, 1.13– 5.63), respectively.

1072

ABERRA ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 9

Table 1. Characteristics of Patients in the Study

Agea (y), IQR Sex (n ⫽ male/%) Follow-up timea,b (y), IQR Diabetesb (n/%) End-stage renal diseaseb (n/%) Gastrectomy (n/%) Adult body mass index (kg/m2), IQRb,c ⱕ18.5 kg/m2 (n/%) Smoking status (n/%) Nonsmoker Former Current Unknown

IBD (UC and CD)

UC

CD

Control group

46 (4–93) 7571/47 3.4 (0.01–9.34) 487/3 51/0.3 32/0.2 24 (14–45) 291/2 7426/51 1201/8 2418/16 3489/24

49 (5–93) 4971/50 3.5 (0.01–9.01) 363/3.6 30/0.3 12/0.2 25 (16–42) 78/0.9 4675/54 763/9 1103/13 2144/25

42 (5–91) 2600/42 3.3 (0.01–8.92) 124/2.0 21/0.3 13/0.2 23 (15–41) 213/3.6 2751/47 438/8 1315/23 1345/23

46 (1–95) 30,268/47 4.17 (0.01–9.34) 1402/2.2 85/0.1 106/0.2 25 (16–50) 850/1.3 29,285/44 5115/8 13,046/20 19,066/29

aMedian. bP

value ⬍ .001 comparing IBD with control group. of subjects missing data for body mass index calculation.

c61%

Because only 1 case of active TB occurred among IBD subjects on immunosuppressants or immunomodulators within the prior 3 months, it was not possible to accurately estimate the association of IBD and active TB among subjects taking immunosuppressants or immunomodulators during this window of exposure (Table 2). With a 6-month exposure window, 8 of 35 subjects with active TB were exposed to corticosteroids, 4 of whom were IBD subjects. With a 12-month exposure window, 9 of 35 subjects with active TB were exposed to corticosteroids, and 5 of these subjects had IBD (Table 3). None of the TB subjects were exposed to 6-mercaptopurine/azathioprine, cyclosporine, or methotrexate with 3-, 6-, and 12-month exposure windows. With a 12-month window, use of any immunosuppressant was a significant confounder for TB, with the unadjusted RR of IBD for TB changing from 2.36 (95% CI, 1.17– 4.74) to an adjusted RR of 1.60 (95% CI, 0.75–3.45). Ninety-four percent of immunosuppressant usage was corticosteroids. Corticosteroids were independently a significant confounder with the 6- and 12-month window of exposure. Adjusting for corticosteroids, the RR for active TB in IBD subjects was 1.65 (95% CI, 0.78 –3.51) and 1.61 (95% CI, 0.75–3.44) for 6- and 12-month window of exposure, respectively, compared with the unadjusted RR of 2.36 (95% CI, 1.17– 4.74). The variables diabetes mellitus, end-stage renal disease, and history of gastrectomy were not confounders on the basis of methods described in the statistical methods section and were not entered into the multivariate regression model. On the

other hand, smoking was found to be a confounder and was included in the multivariate model. Adjusting for smoking, the RR of active TB for IBD compared with non-IBD subjects was 2.70 (95% CI, 1.06 – 6.89) compared with the unadjusted RR of 2.36 (95% CI, 1.17– 4.74). An increased RR was still observed after adjusting for smoking in CD subjects (RR, 2.68; 95% CI, 0.76 –9.51) and UC subjects (RR, 2.67; 95% CI, 0.85– 8.44). The final multivariate model included the variables IBD, smoking, and corticosteroids with a 12-month window of exposure before the end of follow-up (Table 4). The adjusted odds ratio of IBD for active TB was 1.88 (95% CI, 0.68 –5.20). In this model, both smoking and corticosteroids were significant risk factors for active TB, 2.77 (95% CI, 1.12– 6.85) and 4.19 (95% CI, 1.38 –12.72), respectively. Separate multivariate models for CD and UC were completed, adjusting for smoking and corticosteroids with a 12-month window of exposure before the end of follow-up. The adjusted odds ratios for active TB in CD and UC subjects were 1.47 (95% CI, 0.36 –5.95) and 2.10 (95% CI, 0.63– 7.01), respectively. Limiting the dataset to the adult population, the multivariate adjusted odds ratio of IBD for active TB was 1.88 (95% CI, 0.68 –5.18).

Discussion The results of this study revealed that IBD independently might not be associated with an increased risk for active TB, but IBD in combination with the immunosuppressants used to treat

Table 2. Distribution of Patients on Immunosuppressants at 3, 6, and 12 Months Before Date of Outcome or End of Follow-up 3 Monthsa

Prednisoneb (%) Azathioprine/6-mercaptopurineb (%) Methotrexateb (%) Cyclosporine1 (%) aBefore bP

6 Months

12 Months

IBD

Control

IBD

Control

IBD

Control

11.3 3.9 0.15 0.10

2.0 0.08 0.07 0.04

14.3 4.3 0.17 0.14

2.4 0.1 0.08 0.04

18.3 5.0 0.20 0.17

3.2 0.11 0.09 0.04

date of outcome or end of follow-up. value ⱕ .001 for comparison of IBD subjects to controls.

September 2007

RISK FOR ACTIVE TB IN IBD PATIENTS

Table 4. Unadjusted and Adjusted RRs of IBD, UC, and CD for Active TB

IBD UC CD aModel

Unadjusted RR (95% CI)

Adjusteda RR (95% CI)

2.36 (95% CI, 1.17–4.74) 2.52 (95% CI, 1.13–5.63) 2.09 (95% CI, 0.72–6.04)

1.89 (95% CI, 0.68–5.20) 2.10 (95% CI, 0.63–7.01) 1.47 (95% CI, 0.36–5.95)

included the variables corticosteroid use and smoking.

IBD subjects was associated with an increased risk for active TB compared with non-IBD subjects. Our data suggested that the IBD population was at an increased risk for active TB before infliximab availability, largely as a result of immunosuppressant use and corticosteroids specifically. Corticosteroids are commonly used to treat IBD. Population-based studies have suggested that at any given point in time, 10% of CD patients are on corticosteroids, 42% within the first 3 years of disease and 39%–56% during the course of their disease in patients with CD.17–22 This association of IBD subjects for risk for active TB has not yet been reported, and it is not clear what the potential mechanism for this risk would be aside from immunosuppressant use. Perhaps immune dysregulation that occurs in IBD is associated with genetic factors that increase susceptibility to active TB. In a study by Bahr et al,23 rheumatoid arthritis patients with HLA-DR2 and HLA-DR7 were shown to have a decreased antibody response to tuberculous antigen. Genetic factors, without an overt clinical syndrome or disease state such as human haplotypes HLA-DR2, HLA-DRB1-1501, and HLADRB1-1502, have been shown to increase risk for active TB.24 Prior epidemiologic studies have shown that corticosteroids likely increase risk for active tuberculosis, but the studies did not reach statistical significance, possibly as a result of a small sample size.25,26 Possible mechanisms for corticosteroids increasing risk for active TB are impairment of macrophage function by way of inhibition of lysosomal enzyme release, an important defense in fighting off tubercle bacilli, as well as inhibiting release of kinins, and neutrophil chemotaxis.27 Corticosteroids have also been shown to alter the presentation of active TB from pulmonary to extrapulmonary and disseminated disease and to decrease the occurrence of fever, night sweats, malaise, and weight loss.27–29 There have been a few reported cases of active TB in IBD patients, and all were on some form of immunosuppressant aside from anti-TNF therapy.30 –33 In one case report a 37-yearold man with CD for 17 years developed tuberculous pleurisy after receiving treatment with azathioprine.31 In another case report, skin tuberculosis developed in a 27-year-old patient with UC treated with sulfasalazine and prednisone.32 In a large Indian case series of 102 patients with UC, 7 patients had TB (3 pulmonary and 4 small bowel), and all were on steroids at the time of presentation.30 In this study we also observed an increased risk of TB among smokers. The finding of an increased risk of active TB as a result of smoking has previously been suggested.34,35 For example, in a study of patients from southern India, current smokers had an odds ratio of 2.48 (95% CI, 1.42– 4.37) compared with nonsmokers for tuberculosis.34 Further solidifying this association, the odds ratios increased with increasing number of cigarettes smoked. The odds ratios for mild (1–10 cigarettes/day), moderate (11–20/day), and heavy (⬎20/day) smokers were 1.75,

1073

3.17, and 3.68, respectively (P ⬍ .0001, test for linear trend). A proposed mechanism for this risk is through the nicotinic acetylcholine receptor. Acetylcholine has been shown to inhibit the release of macrophage TNF-␤ by way of the nicotinic acetylcholine receptor, and nicotine is a potent agonist of this receptor. There are several potential limitations of this study. Unlike the diagnosis of IBD, the diagnosis of TB has not been previously validated within GPRD. In addition, it would not be possible to validate these diagnoses today because the coding system currently used within the GPRD practices is different from that used at the time that these data were collected. However, there are several reasons to believe that misclassification of the outcome measure has not significantly biased our results. (1) TB is an important diagnosis that almost certainly would be recorded in the primary care medical record. (2) We have excluded the time period shortly after registration with a new physician and shortly after the diagnosis of IBD. This minimizes the risk of bias from patients seeking medical attention for the care of either IBD or other symptoms and having a resultant diagnosis of TB as part of the work-up of these symptoms.11 (3) The incidence of TB observed in the control cohort was comparable to that expected in the general UK population during the 1990s, suggesting complete recording of TB and minimal recording of a positive purified protein derivative skin test as active TB.16 (4) The GPRD contains specific codes for a positive PPD test, which we did not include in our outcome definition. (5) The magnitude of the associations we observed for smoking and tuberculosis was nearly identical to that previously reported in another population. (6) To the extent that misclassification has occurred, if it was nondifferential, the bias would be toward the null, thus not negating our primary conclusion that patients treated with corticosteroids are at an increased risk for active TB. If misclassification were differential, nearly all of the cases of TB among steroid-exposed subjects would need to have been wrong diagnoses to account for the results. Overall, 26% of the TB subjects (42% of those with IBD and 17% of the non-IBD patients) had been exposed to corticosteroids during the preceding year. On the basis of the distribution of TB cases between the 2 study cohorts and the rates of corticosteroid use in the cohorts, only 3 of the 35 TB subjects (9%) would have been expected to have corticosteroid exposure during the previous 12 months. Thus, even if one third of the TB diagnoses were wrong among the steroidtreated subjects, the proportion of TB subjects with corticosteroid exposure during the previous year would be nearly double that expected (17% vs 9%). Taken as a whole, these data suggest that misclassification of the TB diagnosis was unlikely to explain our conclusion that corticosteroid exposure during the

Table 3. Corticosteroid Use Among the Study Cohort Before the Date of TB Diagnosis or End of Follow-up Non-IBD IBD subjects Non-IBD subjects IBD cohort with TB cohort with TB (N ⫽ 16,213) (N ⫽ 12) (N ⫽ 66,512) (N ⫽ 23) 3 Months 6 Months 12 Months

11% 14% 18%

8% 33% 42%

2% 2% 3%

17% 17% 17%

1074

ABERRA ET AL

previous year predisposes subjects to develop active TB infection. Because of a low prevalence of exposure to azathiorine/6mercaptopurine, methotrexate, and cyclosporine, we could not adequately assess these individual drugs as an independent risk factor for TB. Although there were no cases of TB in subjects exposed to these agents, an increased risk might have not been detected as a result of few exposures. Although we did not observe a significant increased risk of TB among IBD subjects after accounting for corticosteroid use and smoking, it is possible that our results were biased toward the null as a result of lack of complete data on socioeconomic status and immigrant status. Immigrant status is not a required field in the database, and the low rate of reported immigrant status suggests that this variable is likely under-recorded. Immigrants and persons of lower socioeconomic status are at an increased risk of TB. In general, IBD is more common among higher socioeconomic groups.36 Immigrants to developed countries appear to have an incidence of IBD comparable to that of their new home and higher than that of their native land. As such, the proportion of the population of IBD patients in the GPRD who are immigrants is likely similar or lower than the proportion of immigrants in the remainder of the GPRD population. Accounting for this, we cannot definitively state that IBD does not in and of itself have a small increased risk for TB, perhaps comparable to that observed in patients with rheumatoid arthritis. However, because under-representation of immigrants in the IBD population would serve to bias away from a positive association between IBD and TB, incomplete data on immigrant status would not be expected to alter our conclusion that IBD patients treated with immunosuppressant medications are at an increased risk of TB. In conclusion, subjects with IBD might have a small increased risk of TB that is largely attributable to exposure to immunosuppressant medications. Because of the high rate of use of corticosteroids, immunomodulators, and anti–TNF-␣ therapies, consideration should be given for testing all IBD patients for TB exposure early in the course of their disease and before the initiation of immunosuppressant medications if possible. Further investigation is warranted to determine whether such a strategy would prove cost-effective.

Supplementary Data Note: to access the supplementary materials accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org. References 1. Keane J, Gershon S, Wise R, et al. Tuberculosis associated with infliximab, a tumor necrosis factor alpha neutralizing agent. N Engl J Med 2001;345:1098 –1104. 2. Van Assche G, Van Ranst M, Sciot R, et al. Progressive multifocal leukoencephalopathy after natalizumab therapy for Crohn’s disease. N Engl J Med 2005;353:362–368. 3. Anonymous. The General Practice Research Database: a guide for researchers. London: EPIC, Regeneration House, 2001. 4. Lis Y, Mann RD. The VAMP Research multi-purpose database in the U.K. J Clin Epidemiol 1995;48:431– 443. 5. Garcia Rodriguez LA, Perez Gutthann S. Use of the UK General Practice Research Database for pharmacoepidemiology. Br J Clin Pharmacol 1998;45:419 – 425.

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 9

6. Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991;302:766 –768. 7. Lewis J, Brensinger C, Bilker W, et al. Validity and completeness of the General Practice Research Database for studies of inflammatory bowel disease. Pharmacoepidemiol Drug Saf 2002;11: 211–218. 8. Gupta G, Gelfand JM, Lewis JD. Increased risk for demyelinating diseases in patients with inflammatory bowel disease. Gastroenterology 2005;129:819 – 826. 9. Lewis J, Bilker W, Brensinger C, et al. Inflammatory bowel disease is not associated with an increased risk of lymphoma. Gastroenterology 2001;121:1080 –1087. 10. Van Staa T-P, Cooper C, Brusse LS, et al. Inflammatory bowel disease and the risk for fracture. Gastroenterology 2003;125: 1591–1597. 11. Lewis JD, Bilker WB, Weinstein RB, et al. The relationship between time since registration and measured incidence rates in the General Practice Research Database. Pharmacoepidemiol Drug Saf 2005;14:443– 451. 12. Lewis J, Brensinger C. Agreement between GPRD smoking data: a survey of general practitioners and a population-based survey. Pharmacoepidemiol Drug Saf 2004;epub Nov 26 2003. 13. World Health Organization. Obesity: preventing and managing the global epidemic. Geneva: WHO Technical Report Series, 2000. 14. Rose A, Watson J, Graham C. Tuberculosis at the end of the 20th century in England and Wales: results of a national survey in 1998. Thorax 2002;56:173–179. 15. Duffield JS, Adams WH, Anderson M, et al. Increasing incidence of tuberculosis in the young and the elderly in Scotland. Thorax 1996;51:140 –142. 16. Health Protection Agency, United Kingdom. Annual report on tuberculosis cases. Tuberculosis Section Health Protection Agency Centre for Infections. 2006. 17. Faubion WAJ, Loftus EVJ, Harmsen WS, et al. The natural history of corticosteroid therapy for inflammatory bowel disease: a population-based study. Gastroenterology 2001;121: 255–260. 18. Munkholm P, Langholz E, Davidsen M, et al. Frequency of glucocorticoid resistance and dependency in Crohn’s disease. Gut 1994;35:360 –362. 19. Silverstein M, Loftus E, Sanborn W, et al. Clinical course and costs of care for Crohn’s disease: Markow model analysis of a population-based cohort. Gastroenterology 1999;117:49 –57. 20. Munkholm PLE, Davidsen M, Binder V. Disease activity courses in a regional cohort of Crohn’s disease patients. Scand J Gastroenterol 1995;30:699 –706. 21. Lewis JLG, Brensinger C, Bilker W, et al. Characteristics of prior steroid use predicts duration of steroid therapy in inflammatory bowel disease (IBD). Gastroenterology 2001;120:A-269. 22. Faubion W, Loftus E, Harmsen W, et al. The natural history of corticosteroid therapy for inflammatory bowel disease: a population-based study. Gastroenterology 2001;121:255–260. 23. Bahr GM, Rook GA, Shahin A, et al. HLA-DR-associated isotypespecific regulation of antibody levels to mycobacteria in rheumatoid arthritis. Clin Exp Immunol 1988;72:26 –31. 24. Mandell GL, Bennett JE, Dolin R, et al. Principles and practice of infectious diseases. New York: Churchill Livingstone, Inc, 2000. 25. Cline J, Davis S. Risks of infection or reactivation of tuberculosis associated with chronic corticosteroid therapy. Ann Pharmacol 1997;31:775–776. 26. Bateman E, Wolmarans K, White N, et al. Corticosteroids treatment and the risk of active tuberculosis in patients with chronic intestinal lung disease (abstract). Am Rev Respir Dis 1991;143: A54. 27. Sahn S, Lakshminarayan S. Tuberculosis after corticosteroid therapy. Br J Dis Chest 1976;70:195–205.

September 2007

28. Allen M, Cooke N. Corticosteroids and tuberculosis. BMJ 1991; 303:871– 872. 29. Selwyn P, Hartel D, Lewis V, et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N Engl J Med 1989;320:545–550. 30. Kumar N, Govil A, Puri AS, et al. Tuberculosis in ulcerative colitis: bird in the bush. Trop Gastroenterol 1994;15:219 –221. 31. van Wijngaarden P, Meijssen MA. Tuberculous pleurisy: an unusual complication during treatment of Crohn disease with azathioprine. Scand J Gastroenterol 2001;36:1004 –1007. 32. Puiatti P, Salvai M, Alberico G, et al. Cutaneous tuberculosis: atypical skin lesions in immunodepressed patients. G Ital Dermatol Venereol 1990;125:445– 448. 33. Sategna Guidetti C, Pulitano R. Pulmonary tuberculoma associ-

RISK FOR ACTIVE TB IN IBD PATIENTS

1075

ated with Crohn’s disease. J Clin Gastroenterol 1991;13: 593–594. 34. Kolappan C, Gopi PG. Tobacco smoking and pulmonary tuberculosis. Thorax 2002;57:964 –966. 35. Maurya V, Vijayan VK, Shah A. Smoking and tuberculosis: an association overlooked. Int J Tuberc Lung Dis 2002;6:942–951. 36. Danese S, Sans M, Fiocchi C. Inflammatory bowel disease: the role of environmental factors. Autoimmun Rev 2004;3:394 – 400.

Address requests for reprints to: Faten N. Aberra, MD, University of Pennsylvania, Division of Gastroenterology, 3400 Spruce St, Philadelphia, PA 19104. e-mail: [email protected].

September 2007

RISK FOR ACTIVE TB IN IBD PATIENTS

Supplementary Table 1. Tuberculosis Codes Searched in the GPRD Medical key/ OXMIS code

Problem-text

010 A 010 C 010 M 11 011 A 011 B 011 C 011 D 011 F 011 P 0120P 121 0121A 0121B 0121C 122 0122A 123 0124A 129 0129A 0129B 0129BM 130 0130M 014 B 014 C 014 H 014 RP 0150A 0150AA 0150AB 0150B 0150C 0150D 0150E 0150F 0150G 151 152 0158A 0158B 0158C 0158D 159 0159A 0159B 0159C 0159CW 16 016 A 016 BL 016 EP 016 F 016 K 016 P 016 PA 016 PB 016 PN 016 TE 016 TH 0170A

TUBERCULOSIS WITH ASBESTOSIS SILICOTUBERCULOSIS TUBERCULOSIS WITH MINERS’ LUNG TUBERCULOSIS PULMONARY TUBERCULOSIS TUBERCULOUS ABSCESS LUNG TUBERCULOSIS MILIARY LUNG TUBERCULOUS BRONCHIECTASIS MULTIPLE FOCI TUBERCULOSIS PRIMARY PULMONARY T B PRIMARY RESPIRATORY (TUBERCULOSIS) TUBERCULOSIS WITH PLEURISY TUBERCULOUS ADHESIONS PLEURA TUBERCULOUS EMPYEMA TUBERCULOUS PLEURAL EFFUSION SEROFIBRINOUS PLEURISY PLEURAL EFFUSION TUBERCULOSIS NOT EXCLUD TUBERCULOUS LARYNGITIS TUBERCULOMA TUBERCULOUS LYMPHADENOPATHY TUBERCULOUS NASAL SINUS TUBERCULOUS LYMPH NODES HILAR TUBERCULOUS LYMPH NODES MEDIASTINAL TUBERCULOUS MENINGITIS TUBERCULOUS MENINGOMYELITIS TUBERCULOUS LYMPH NODES MESENTERIC TUBERCULOUS PERITONITIS TUBERCULOUS HEPATITIS TUBERCULOUS ABSCESS RETROPERITONEAL TUBERCULOSIS SPINE TUBERCULOUS ABSCESS JOINT VERTEBRAL TUBERCULOUS ABSCESS VERTEBRA TUBERCULOUS ABSCESS LUMBAR TUBERCULOUS ABSCESS ILIOPSOAS TUBERCULOUS ABSCESS SACRUM TUBERCULOUS KYPHOSIS TUBERCULOUS LORDOSIS TUBERCULOUS SPONDYLITIS TUBERCULOUS JOINT HIP TUBERCULOUS JOINT KNEE TUBERCULOUS JOINT TUBERCULOUS BONE TUBERCULOUS MASTOIDITIS TUBERCULOUS DACTYLITIS TUBERCULOSIS SKELETAL TUBERCULOUS OSTEOMYELITIS TUBERCULOUS ARTHRITIS TUBERCULOUS TENOSYNOVITIS T B ABSCESS CHEST WALL (TUBERCULOSIS) TUBERCULOSIS GENITO-URINARY SYSTEM TUBERCULOUS ORCHITIS TUBERCULOSIS URINARY BLADDER TUBERCULOSIS EPIDIDYMIS TUBERCULOSIS FALLOPIAN TUBE TUBERCULOSIS KIDNEY TUBERCULOSIS PROSTATE TUBERCULOSIS SEMINAL VESICLES TUBERCULOSIS PROSTATE & SEMINAL VESICLES TUBERCULOSIS PENIS TUBERCULOSIS URETER TUBERCULOSIS URETHRAL TUBERCULOUS ERYTHEMA NODOSUM

1075.e1

Supplementary Table 1. Continued Medical key/ OXMIS code 0170B 0170C 0170CN 0170CP 0170CT 0170LP 0170PA 171 0171A 0171NG 0172A 0172L 179 0179AR 0179B 0179D 0179DA 181 0189C 199 0199LG A1. . .00

Problem-text TUBERCULOSIS LUPUS VULGARIS TUBERCULOUS VERRUCA VERRUCA NECROGENICA (PRIMARY) WART(S) PROSECTOR’S VERRUCA TUBERCULOSA PURPURA LICHENOIDES ACNITIS (PRIMARY) TUBERCULOUS ADENITIS TUBERCULOSIS AXILLA GLAN TUBERCULOUS GLANDS NECK TUBERCULOUS CHOROIDITIS TUBERCULOSIS ACHROACYTOSIS LACHRYMAL GLA TUBERCULOUS PERICARDITIS T B RELAPSE (TUBERCULOSIS) TUBERCULOSIS BREAST TUBERCULOSIS ADRENAL GLANDS TUBERCULOUS ADDISON’S DISEASE TUBERCULOSIS MILIARY CHRONIC MILIARY TUBERCULOSIS TUBERCULOSIS LATE EFFECTS LATE EFFECT TUBERCULOSIS ACHROACYTOSIS TUBERCULOSIS

OXMIS, Oxford medical information system.