The inflammatory bowel diseases (IBDs), Crohn s disease. Incidence and Risk Factors for Herpes Zoster Among Patients With Inflammatory Bowel Disease

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4: Incidence and Risk Factors for Herpes Zoster Among Patients With Inflammatory Bowel Disease GAUREE GUPTA,* EBBING LAUTENBACH,*,, and JAMES D. LEWIS*,, *Center for Clinical Epidemiology and Biostatistics, Centers for Education and Research on Therapeutics, and the Department of Internal Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania Background & Aims: An increased risk of herpes zoster in patients with inflammatory bowel disease (IBD) is hypothesized based on altered immune function, especially among patients receiving immunosuppressive medications. Methods: We performed a retrospective cohort study and a retrospective nested case-control study using data from the General Practice Research Database. In the cohort study, 7823 Crohn s disease (CD) and 11,930 ulcerative colitis (UC) patients were matched on age, sex, and primary care practice to 79,563 randomly selected controls without CD or UC. In the nested case-control study, 185 CD patients with zoster and 266 UC patients with zoster were matched on sex and year of birth to 1787 IBD patients without zoster. Results: In the cohort study, the incidence of zoster was higher in patients with CD and UC compared with their matched controls (UC incidence rate ratio, 1.21; 95% confidence interval [CI], ; CD incidence rate ratio, 1.61; 95% CI, ). In the nested case-control study, receipt of a prescription for corticosteroids (adjusted odds ratio, 1.5; 95% CI, ) or azathioprine/6- mercaptopurine (adjusted odds ratio, 3.1; 95% CI, ) were both associated with zoster. Conclusions: Patients with IBD, especially those on immunosuppressive medications, are at higher risk for herpes zoster compared with the general population. Future studies should clarify the relative risk associated with anti tumor necrosis factor therapies and determine the use of the new zoster vaccine for patients with IBD. The inflammatory bowel diseases (IBDs), Crohn s disease (CD) and ulcerative colitis (UC), are chronic conditions characterized by altered regulation of the immune system. Although immune dysregulation contributes to the progressive intestinal inflammation found in IBD, it may intrinsically predispose patients to other complications, such as an increased risk of infection. Medications commonly used to treat IBD also can increase susceptibility to infection by suppressing normal immune functions. 1,2 First-line therapy for IBD includes mesalamine medications that have little effect on the systemic immune system. In contrast, second- and third-line therapies, such as corticosteroids, azathioprine (AZA), 6-mercaptopurine (6-MP), methotrexate, cyclosporine, and anti tumor necrosis factor (TNF), therapies can cause systemic immunosuppression. Although the efficacy of these drugs has been established clearly, data on the absolute and relative risk of infection and other adverse events are more limited. As a result, understanding how to best balance the benefits of therapy with the risks of infection and other side effects is not yet evident. Much of the available data on infectious complications of IBD therapy has focused on the risk of postoperative complications. The use of corticosteroids increases the risk of postoperative infections, most of which are bacterial. 1,3 5 The use of AZA or 6-MP in the same setting appears to have less risk of infection. Infliximab use is associated with disseminated tuberculosis, but data suggest little if any increased risk of bacterial infection after surgery. 6 Fewer data are available on the risk of viral infections with IBD therapies. However, several patients treated for Crohn s disease and multiple sclerosis with natalizumab developed progressive multifocal leukoencephalopathy (a lethal complication of reactivation of human papovavirus JC virus). 7 9 Because treatment of IBD often involves combination therapy with multiple medications, each with the potential for immunosuppression, it is important to understand the risk of infection and the potential contribution that different medications have on that risk. Herpes zoster, also called shingles, is a common nonfatal infection characterized by a unilateral, painful vesicular rash in a dermatomal distribution. 2 It results from reactivation of latent varicella zoster virus within dorsal root ganglia. 10 The incidence and severity of zoster increase over one s lifetime, in association with an age-related decrease in cellular immunity. A new vaccine appears to reduce the incidence of zoster among immunocompetent patients, although its efficacy in patients receiving immunosuppressive medications is unknown. 11,12 An increased risk of zoster in patients with IBD is hypothesized based on altered immune function, particularly among patients receiving immunosuppressant medications. 1,13 16 There are limited data on risk factors for zoster, particularly the use of steroids and immunomodulators, among patients with IBD. The specific aim of this study was to determine whether patients with IBD have a higher incidence of zoster than the general population. Our secondary aims were to determine among patients with IBD whether (1) use of corticosteroids is associated with zoster infection, (2) use of AZA/6-MP is associated with zoster infection, and (3) whether patients with CD have a higher risk of zoster than patients with UC. Abbreviations used in this paper: AZA, azathioprine; CD, Crohn s disease; CI, confidence interval; GPRD, General Practice Research Database; IBD, inflammatory bowel disease; IRR, incidence rate ratio; OR, odds ratio; 6-MP, 6-mercaptopurine; TNF, tumor necrosis factor; UC, ulcerative colitis by the AGA Institute /06/$32.00 doi: /j.cgh

2 1484 GUPTA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 12 Materials and Methods Data Source This study was conducted using data from the General Practice Research Database (GPRD) from 1988 to This database of primary care medical records from the United Kingdom has been described in detail elsewhere Briefly, the GPRD contains medical records of approximately 8,000,000 patients in total, representing 6% of the UK population. 21 Each medical practice must show competency at entering data into the electronic database before their data are considered up to standard. Subsequently, monthly audits ensure that the data quality remained up to standard. Data recorded in the GPRD include demographic information, prescription information, clinical events and diagnoses, preventive care, hospital admissions, cause of death, and free text. In addition, significant diagnoses occurring before the initiation of the electronic medical record are recorded retrospectively. Diagnoses are recorded using Oxford Medical Indexing System codes. 22,23 Prescribed medications are recorded using codes issued by the Prescription Pricing Authority of the National Health Service. 22,23 Several studies have shown that the clinical information in the computer record is sufficiently accurate for use in epidemiologic studies, including studies of IBD. 20,24,25 Patients This study was conducted using a previously described cohort. 17,26,27 The IBD patients include all patients with a diagnosis of UC or CD and at least 1 year of follow-up time. Patients diagnosed only as IBD not otherwise specified or with codes for both CD and UC were excluded. Each IBD patient had previously been matched to up to 4 control subjects on year of birth 5 years, sex, and primary care practice from the remaining population of patients without a diagnosis of IBD or potentially related conditions (eg, proctitis) who were cared for by GPRD physicians. We excluded patients with missing registration dates and any patients with a diagnosis of zoster before or within the first 3 months of the start of follow-up evaluation. 19 In addition, we excluded IBD patients for whom no matched controls remained after applying these exclusion criteria (n 25). Definition of Follow-Up Time and Outcome Measures in the Cohort Study The primary outcome for the cohort study was a first diagnosis of zoster, defined as herpes zoster, shingles, or a zoster-related diagnosis. Zoster-related diagnoses included ophthalmic herpes zoster, ophthalmic shingles, herpetic neuralgia, postherpetic pain, and Ramsay Hunt syndrome. Herpes zoster geniculate or herpes zoster auricularis also were included in the definition but no patients had these diagnoses. We have shown previously that diagnoses of acute conditions and infections, including zoster, can be overreported in GPRD if time periods shortly after registration are not excluded. 19 Therefore, the follow-up period began with the latest of the following: 3 months after registration with the physician s practice, the start of up-to-standard data collection, and, for patients with IBD, the date of the first diagnosis of IBD. Follow-up evaluation ended with earliest of the date of the first diagnosis with zoster, death, transfer out of the practice, or the end of the data. Only data through 1997 (ie, before the introduction of anti-tnf medications to the market) were included in this study because use of anti-tnf therapies generally is not recorded in the database. Nested Case-Control Study To understand the relationship between IBD therapies and the risk of zoster, we conducted a case-control study nested within the cohort of IBD patients. Case and control subjects were those IBD patients with or without zoster, respectively. The date of first diagnosis of zoster was used as the case subject s index date. Each case patient (IBD patient with zoster) was matched on sex, year of birth, and duration of follow-up evaluation in the cohort to up to 4 IBD patients who did not have zoster during the follow-up period using incidence density sampling. 28 This method requires that the control patient not have been diagnosed with zoster before the index date. However, as is customary with incidence density sampling, if a control subject was later diagnosed with zoster, he or she could serve as a case subject as well. 28 Forty-two patients were included as both case and control subjects. Organ transplantation and infection with the human immunodeficiency virus both result in impaired cellular immunity and increased risk of viral infections. Because organ transplantation and human immunodeficiency virus infection are relatively rare, we excluded patients with these diagnoses from the nested case-control study. In total, we excluded 76 patients with documented organ/tissue transplantation and 22 patients with human immunodeficiency virus infection or acquired immune deficiency syndrome. Among these excluded patients, 3 had been diagnosed with zoster in the cohort study. All 3 had a history of transplant, 1 of whom also had a history of human immunodeficiency virus/acquired immune deficiency syndrome. Exposure Definitions for the Nested Case-Control Study Exposures were measured from the index date, which was the date of the first zoster diagnosis among the case patients and the same date for the matched control patients. The primary exposures of interest were the underlying IBD (CD vs UC) and use of immunosuppressant medications (corticosteroids, AZA/6-MP, methotrexate, or cyclosporine A/tacrolimus) in the period immediately before the index date. Mesalamine, which is not believed to cause systemic immunosuppression, was used as a negative control exposure. Of note, there was no use of cyclosporine A or tacrolimus by either case patients or matched control patients within 30 days of the index date. However, 1 control patient was noted to have received a prescription of cyclosporine A or tacrolimus within days of the index date. Because prescriptions of corticosteroids frequently include instructions to take as directed, we did not attempt to determine the prescribed duration of each prescription. Rather, we defined exposure on the basis of the dates that the prescriptions were written. We decided a priori that our primary exposure definition would include any prescription received within 30 days before the index date. We also examined several alternative definitions of exposure in secondary analyses to better understand the time course between exposure to immunosuppressant medications and risk of zoster. Current use was defined as a prescription within 30 days of the index date, and previous use was defined as having a

3 December 2006 IBD AND HERPES ZOSTER 1485 prescription during the period days before the index date but not within 30 days of the index date. Persistent use was defined as receipt of a prescription both in the period days before the index date and in the period 1 30 days before the index date. New use was defined as a prescription for the medication within 1 30 days of the index date and no prescriptions in the period days before the index date. Potential Confounders Based on previous reports of potential association, patients smoking status, alcohol abuse, cancer, and depression were evaluated as potential confounder variables in the diagnosis of zoster Cancer was defined as leukemia, lymphoma, and malignant tumors other than squamous or basal cell skin cancer. In the case-control study, exposure was defined as a diagnosis of cancer or depression within 1 year, or alcoholism within 5 years, of the index date. Statistical Analyses Continuous variables are reported as means with SDs. Categoric variables are reported as counts and percentages. In the cohort study, we calculated incidence rate ratios (IRRs) and corresponding 95% confidence intervals (CIs) to compare the incidence of zoster among IBD patients and controls. In the nested case-control study, we used conditional logistic regression to calculate odds ratios and 95% CIs. We examined the associations for evidence of confounding by including all potential confounder variables in a single multivariable model. We retained in the final model any variable that could not be removed without resulting in a 10% or greater change in the odds ratio (OR) for either the association of corticosteroids or AZA/6-MP with zoster. Ultimately, the only confounders of these associations were the other medication group (ie, corticosteroid use was a confounder of the association of AZA/6-MP with zoster and vice versa). All analyses were performed using STATA version 9.0 (STATA Corp, College Station, TX). All analyses used 2-sided statistical tests. Ethics Board Approval The study protocol was approved by the Scientific and Ethical Advisory Group of the GPRD and the University of Pennsylvania s institutional review board. Results Cohort Study The final study population for the cohort study included 7823 patients with CD (58.5% women; mean age at start of follow-up period, y) and 11,930 patients with UC (50.2% women; mean age at start of follow-up period, y) (Table 1). All IBD patients had at least 1 matched control and more than 95% had 4 matched controls, resulting in a total of 79,563 control patients (53.5% women; mean age at start of follow-up period, y). To examine the completeness of the recording of zoster in the GPRD, we first examined the incidence rates of zoster among the patient groups. During the follow-up period, 186 patients with CD and 268 patients with UC were diagnosed with zoster compared with 534 patients and 982 patients in the matched control groups, respectively. Among both the IBD patients and the controls, incidence rates of zoster increased with increasing age (Figure 1). A prior population-based study performed within the United Kingdom reported age-specific Table 1. Characteristics of Patients in the Cohort Study CD (n 7823) UC (n 11,930) Controls (n 79,563) Sex (% women in total population) 4573 (58.5) 5989 (50.2) 42,561 (53.5) Median number of years of follow-up evaluation, y (interquartile range) 4.7 ( ) 5.0 ( ) 4.4 ( ) Mean age at start of follow-up evaluation, y (SD) 42.5 (18.3) 48.8 (17.7) 46.3 (18.3) Smoking status Smoker (%) 2467 (31.5%) 2207 (18.5%) 20,010 (25.2%) Nonsmoker (%) 3234 (41.3%) 6658 (55.8%) 33,693 (42.4%) Unknown (%) 2122 (27.1%) 3065 (25.7%) 25,860 (32.5%) Number of patients with zoster Ages 0 4, y (%) 0 (0.0%) 0 (0.0%) 0 (0.0%) Ages 5 14, y (%) 3 (0.0%) 1 (0.0%) 16 (0.0%) Ages 15 44, y (%) 66 (0.8%) 60 (0.5%) 367 (0.5%) Ages 45 64, y (%) 69 (0.9%) 106 (0.9%) 537 (0.7%) Ages 65, y (%) 48 (0.6%) 101 (0.8%) 537 (0.7%) Diagnoses of herpes zoster, n (%) Herpes zoster 61 (0.8%) 75 (0.6%) 446 (0.6%) Shingles 131 (1.7%) 198 (1.7%) 1120 (1.4%) Herpes zoster or shingles 181 (2.3%) 262 (2.2%) 1506 (1.9%) Ophthalmic herpes zoster 3 (0.0%) 5 (0.0%) 21 (0.0%) Ophthalmic shingles 1 (0.0%) 3 (0.0%) 15 (0.0%) Herpes zoster geniculate 0 (0.0%) 0 (0.0%) 0 (0.0%) Herpes zoster auricularis 0 (0.0%) 0 (0.0%) 0 (0.0%) Herpetic neuralgia 6 (0.1%) 3 (0.0%) 33 (0.0%) Postherpetic pain 9 (0.1%) 16 (0.1%) 62 (0.1%) Ramsay Hunt syndrome 1 (0.0%) 2 (0.0%) 2 (0.0%) NOTE. Some patients had more than 1 zoster-related diagnosis on the index date.

4 1486 GUPTA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 12 Figure 1. Annual incidence rate of zoster among patients with inflammatory bowel disease and their matched controls *P.01, # P.05. incidence rates of zoster as follows (rate per 100,000 personyears): ages 0 4 years (92), 5 14 years (219), years (212), years (712), and 65 years and older (932). 33 These rates are quite similar to those observed in our non-ibd control cohorts (Figure 1). Compared with their matched controls, the incidence rate of zoster was significantly higher for both patients with CD (IRR, 1.6; 95% CI, ) and UC (IRR, 1.2; 95% CI, ). Nearly identical results were obtained when we restricted the outcome to a diagnosis of herpes zoster or shingles (CD: IRR, 1.6; 95% CI, ; UC: IRR, 1.2; 95% CI, ). In subgroup analyses, the significant differences in incidence rates were limited to patients with CD in the age groups 15 44, 45 64, and 65 and older (Figure 1). Nested Case-Control Study The final study population in the nested case-control study included 451 cases (51.2% women; mean age at start of follow-up period, y) and 1787 controls (51.4% women; mean age at start of follow-up period, y) (Table 2). The first zoster diagnosis in the medical record was herpes zoster or shingles in more than 96% of the cases. Of the cases, 185 (41%) had CD and 266 (59%) had UC. Patients with CD were more likely to have been diagnosed with zoster than patients with UC, even after adjusting for use of medications and other potential confounders (adjusted OR, 1.3; 95% CI, ). All the potential confounders examined were more common in the cases than the controls, only reaching statistical significance for cancer. However, the prevalence of these risk factors was generally lower than that of steroid use and AZA/ 6-MP use among the cases. In both unadjusted and adjusted analyses, receipt of a prescription for corticosteroids (adjusted OR, 1.5; 95% CI, ) or AZA/6-MP (adjusted OR, 3.1; 95% CI, ) were both associated with the risk of zoster (Table 3). Fourteen percent of case subjects were exposed to at least 1 immunosuppressant medication (AZA/6-MP, steroids, or methotrexate) within 30 days of the index date (any immunosuppressant medication use: OR, 2.0; 95% CI, ). To further explore the association observed in the primary analyses, secondary analyses were conducted examining different time intervals between the index date and the most recent prescription and duration of medication use before the index date (Tables 3 and 4). Evaluating each drug alone, exposure to AZA/6-MP or oral steroids at various intervals within 90 days Table 2. Characteristics of Study Cohorts in Nested Case-Control Study Cases (n 451) Controls (n 1787) P value* Sex (% women in total population) 231 (51.2) 918 (51.4) N/A Mean age at start of followup 55.3 (17.4) 55.3 (17.3) N/A evaluation, y (SD) Median time at risk, y 2.2 ( ) 2.2 ( ) N/A (interquartile range) IBD diagnoses CD,n(%) 185 (41.0) 617 (34.5) Ref UC,n(%) 266 (59.0) 1170 (65.5).01 Smoking status Nonsmoker, n (%) 268 (59.4) 1025 (57.4) Ref Smoker, n (%) 101 (22.4) 385 (21.5).91 Unknown, n (%) 82 (18.2) 377 (21.1).15 Medication use (exposure within 30 days of index date) AZA/6-MP, n (%) 22 (4.9%) 26 (1.5%).00 Oral steroids, n (%) 48 (10.6%) 118 (6.6%).00 Oral mesalamine, n (%) 96 (21.3%) 374 (20.9%).83 Oral or injected 1 (0.2%) 1 (0.1%).33 methotrexate, n (%) Any immunosuppressant, 63 (14.0%) 133 (7.4%).00 n(%) a Medical diagnoses (exposure period) Cancer (1 y) 13 (2.9%) 22 (1.2%).02 Depression (1 y) 23 (5.1%) 68 (3.8%).21 Alcoholism (5 y) 3 (0.7%) 7 (0.4%).44 NOTE. P value shown represents unadjusted results. N/A, patients were matched on these variables. a Any Immunosuppressant includes use of at least 1 or more of the following medications: AZA/6-MP, oral steroids, or oral or injected methotrexate.

5 December 2006 IBD AND HERPES ZOSTER 1487 Table 3. Association of IBD Medication Use and Risk of Zoster Exposure period Oral steroids AZA/6-MP Oral mesalamines Unadjusted 1 14 days 2.0 ( ) 3.8 ( ) 1.1 (.8 1.5) 1 30 days 1.7 ( ) 3.4 ( ) 1.0 (.8 1.3) Anytime 1.9 ( ) 2.5 ( ) 1.1 (.9 1.4) 1 60 days Anytime 1.9 ( ) 2.3 ( ) 1.0 (.8 1.3) 1 90 days Adjusted a 1 30 days 1.5 ( ) 3.1 ( ).9 (.7 1.2) a Adjusted for other classes of IBD medications. of the index date yielded a similarly increased risk of zoster (Table 3). In contrast, receipt of oral mesalamine was not associated with the risk of zoster (Table 3). The association of medication use with zoster was slightly stronger among those patients classified as new users of corticosteroids and AZA/6-MP than for those classified as persistent users (Table 4). When the definition of persistent user was narrowed further to only patients who had prescriptions from 1 to 30 days, from 31 to 60 days, and from 61 to 90 days, the association of AZA/6-MP use with zoster was similar to that observed in patients taking AZA/6-MP when persistent user was defined as 1 30 days and days (adjusted OR, 2.6; 95% CI, vs adjusted OR, 2.4; 95% CI, , respectively). Similarly, comparing the more narrow definition of persistent user with our primary definition of persistent user, the association with oral steroids was similar in magnitude, although it was no longer statistically significant (adjusted OR, 1.4; 95% CI, vs adjusted OR, 1.6; 95% CI, ). For mesalamines the associations were almost identical regardless of the definition of persistent user (adjusted OR, 0.9; 95% CI, vs adjusted OR, 0.9; 95% CI, ) (Table 4). When the most recent prescription occurred more than 30 days before the index date, there was no apparent association between AZA/6-MP use and the risk of zoster (adjusted OR, 0.6; 95% CI, ). In contrast, the association remained significant for corticosteroids (Table 4). Concurrent use of oral steroids and AZA/6-MP within 30 days of a diagnosis of zoster among IBD patients was examined. The odds ratio was smaller for the use of steroids alone (OR, 1.6; 95% CI, ) than for AZA/6-MP alone (OR, 4.1; 95% CI, ) or concurrent use of steroids and AZA/6-MP (OR, 3.0; 95% CI, ). Discussion With increasing use of medications targeting the immune system, infectious complications of IBD therapies are to be expected. Herpes zoster is a common viral infection in adults, with the potential for prolonged symptoms and decreased quality of life after the acute infection has resolved. In this study, we observed a small but significant increased incidence rate of zoster among patients with CD and UC compared with the general population. Within the population of patients with IBD, the risk of zoster was greater in patients with CD and those receiving therapy with immunomodulators and corticosteroids. Although there was a strong association with immunomodulator use and the risk of zoster, the vast majority of IBD-related zoster cases were on neither steroids nor immunomodulators. The increased risk observed among IBD patients compared with their matched controls is likely multifactorial. Patients with active IBD have dysregulation of the immune system that could increase the risk of zoster. However, the use of immunosuppressant medications is likely as important, or more important. Table 4. Timing of Prescriptions and Risk of Zoster Oral steroids AZA/6-MP Oral mesalamines Unadjusted Adjusted f Unadjusted Adjusted f Unadjusted Adjusted f Frequency of prescriptions before index date Never user a New user b 1.8 ( ) 1.8 ( ) 4.2 ( ) 4.2 ( ) 1.0 (0.6.8 ( ) 1.6) Persistent user c 1.7 ( ) 1.6 ( ) 3.2 ( ) 2.4 ( ) 1.1 ( ) 0.9 ( ) Time from most recent prescription to index date Never user a Current user d 1.8 ( ) 1.6 ( ) 3.4 ( ) 2.7 ( ) 1.0 ( ) 0.9 ( ) Previous user e 2.2 ( ) 2.2 ( ) 0.8 ( ) 0.6 ( ) 1.0 ( ) 0.9 ( ) a Never user defined as no prescriptions within 90 days. b New user defined as a prescription within 30 days of, but no prescriptions in the period days before the index date. c Persistent user defined as a prescription in the 30 days before, and in the period days before, the index date. d Current user defined as a prescription within 30 days of index date. e Previous user defined as a prescription anytime within 31 to 90 days of, but not within 30 days of, the index date. f Adjusted for other classes of IBD medications.

6 1488 GUPTA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 12 The mechanisms of action and pharmacokinetics of oral steroids and AZA/6-MP help to explain our observations. Corticosteroids act nonspecifically at the level of transcription to modulate genes involved in the priming of the immune response, resulting in the interruption of the network of inflammatory cytokines, reduced antigen-presenting cell chemotaxis and phagocytosis, and suppression of T-cell activation AZA and 6-MP inhibit proliferation of T and B lymphocytes and reduce the number of cytotoxic T cells and plasma cells. 37 Thus, both corticosteroids and AZA/6-MP are expected to result in a reduction in cell-mediated immunity, particularly increasing susceptibility to opportunistic and intracellular infections, such as zoster. 34 The risk for infection with corticosteroids increases with duration of use longer than 2 weeks. 13,38 In the Guidelines for Immunization of Patients with Inflammatory Bowel Disease, Sands et al 13 defined the immune-suppressed IBD patient as having treatment with glucocorticoids (prednisone 20 mg/day equivalent or greater, or 2 mg/kg/day if less than 10 kg, for 2 weeks or more, and within 3 months of stopping), azathioprine/6-mp, methotrexate, infliximab, or significant proteincalorie malnutrition. Our results generally are consistent with this definition, having found an increased risk for zoster infection among IBD patients who received a prescription for oral corticosteroids or immunomodulators at various time intervals within 90 days of zoster diagnosis. The proximity of initiation of AZA/6-MP therapy to risk of infection is characterized poorly, although it is believed that maximal efficacy may require up to 12 weeks of therapy. Our study found increased risk of zoster among patients who received a prescription for AZA/6-MP within 30 days of the index date, regardless of whether they had received an additional prescription in the period from 31 to 90 days before the index date. Interestingly, we did not observe any increased risk among patients who last received a prescription for AZA/6-MP more than 30 days before the index date, suggesting that the immunosuppressive effects of these medications may correct relatively quickly. In contrast, an observational study of 6-MP related complications in IBD patients found that the severity of herpes zoster did not appear to be related to the continuation or withdrawal of 6-MP during infection. 16 However, in clinical practice, AZA/6-MP usually now is withdrawn with the advent of severe infection. Studying drug combinations will become increasingly important. The anti-tnf therapies often are prescribed concurrently with steroids, AZA/6-MP, or methotrexate. We chose to examine the association of zoster and medication use before the era of anti-tnf therapies emerged because the use of anti- TNF therapies generally is not recorded in the database. As well, it is important to understand the relationship of other immunomodulators to zoster in the absence of the exposure to anti-tnf therapies. We anticipated that patients receiving both corticosteroids and AZA/6-MP would have the highest risk of zoster. However, the strength of the association was slightly lower in this combined treatment group than in those on AZA/6-MP alone. Importantly, these were relatively small subgroups and the estimates have wide CIs. As a result, this study cannot rule out a synergistic effect of these medications. Future studies with larger sample sizes and including the use of anti- TNF therapies are needed to clarify the issues of risk of infection with concomitant use of multiple medications. Cell-mediated immunity to varicella zoster virus in older patients with a prior history of chickenpox can be stimulated by immunization against varicella zoster virus. Current guidelines for immunization of patients with IBD suggest administering a 2-dose vaccination series of Varicella Virus Vaccine Live (Varivax, Merck, Whitehouse Station, NJ) in patients ages 50 and older. 13 A new live zoster vaccine (Zoster Vaccine Live [Zostavax, Merck, Whitehouse Station, NJ]) is under development and if approved for marketing potentially could be used in this population as well. 11 However, as a live virus vaccine, Varivax generally is considered contraindicated in immunesuppressed IBD patients. 13 Of note, 2 studies on pediatric patients with leukemia who currently were receiving or had recently completed chemotherapy, including AZA/6-MP, showed more than 90% seroconversion with vaccine-induced immunity effective in preventing varicella after exposure to natural disease. 39,40 Whether Varivax and/or Zostavax may be equally safe in adult patients with IBD on similar medications is unknown. Strengths of this study include the well-defined source population, the use of an internal control group selected from the same source population, the previous validation of the reliability of the IBD diagnosis, the large sample size, and the ability to examine subgroups of patients treated with specific medications. We had access to data before the introduction of the anti-tnf therapies to the market. This can be viewed as both a strength and a limitation, allowing for estimates of risk associated with standard IBD therapies that are not confounded by concomitant use of anti-tnf therapies, but preventing us from examining the association of anti-tnf therapies with the risk of zoster. There are a few other potential limitations worth considering. One is the difficulty in studying the dose of steroids and immunomodulators. Steroid prescriptions often are written to be taken as directed, which limits the ability to analyze steroid dose. Incomplete weight data limits the ability to analyze immunomodulator dose. It is possible that the smaller relative odds observed for corticosteroids relative to AZA/6-MP may reflect relatively low-dose corticosteroid use. Additional studies are needed to clarify the dose-response relationship between these medications and the risk of zoster. The use of an established database that relied on diagnoses recorded by general practitioners also has the potential to lead to misclassification bias. Prior studies have validated the GPRD for use in epidemiologic studies, especially for studies of IBD ,41,42 However, we are unaware of epidemiologic studies that have confirmed the accuracy of diagnoses of herpes zoster or shingles or zoster-related diagnoses as recorded in the GPRD. Fortunately, zoster is a relatively easy condition to diagnose and, as such, misclassification should be rare. To assess the completeness of reporting of zoster in the database, we examined the incidence rate of zoster among our population in the cohort study and compared this with a prior populationbased study from the United Kingdom. 33 In our study we observed an age-associated increase in the incidence rate of zoster similar to that observed by Brisson and Edmunds. 33 Both studies showed, compared with people ages 5 44, a roughly 2- to 3-fold increased risk among patients ages 45 64, and a 4-fold increased risk among patients ages 65 and older. Furthermore, the incidence rates in the 2 studies generally were comparable.

7 December 2006 IBD AND HERPES ZOSTER 1489 Thus, it appears that the recording of zoster within GPRD is relatively complete. Nonetheless, it could be argued that general practitioners, in atypical clinical cases, are more prone to make the diagnosis of zoster in IBD patients receiving immunomodulators than in those not receiving immunomodulators. This would potentially bias our results away from the null. However, the degree of misdiagnosis of zoster would need to be substantial to eliminate the association with AZA/6-MP. In sensitivity analyses, we estimate that if 20% of exposed case subjects did not actually have zoster, the association with AZA/6-MP would remain statistically significant with an OR of between 2.2 and 2.5 (data not shown). Thus, differential misdiagnosis of zoster is unlikely to explain the association between immunomodulator use and zoster. Because the observed association with steroids was weaker than that with AZA/6-MP, such misclassification could have a larger impact on results of analyses focusing on steroid exposure. It is difficult to assess disease activity in GPRD. Our analyses of the association between medication use and risk of zoster could be subject to confounding by indication if patients with active IBD have a higher risk of zoster. There was some evidence of confounding by active disease because the association between immunomodulator use and zoster was the strongest in the new users; however, this was based on small numbers of exposed case patients and the CIs were extremely wide (adjusted OR, 4.2; 95% CI, ). Several pieces of evidence argue against confounding by indication representing a major bias. Importantly, we did not see any association between use of mesalamine and the risk of zoster, even among patients who newly started these medications. Of course, it is possible that mesalamine is preferentially used for patients with mild exacerbations of IBD whereas corticosteroids and AZA/6-MP are used for more severe disease. However, it is unknown whether, among patients with active disease, if severity of the exacerbation is associated with the risk of zoster. Also arguing against confounding by indication in the association between AZA/6-MP and zoster risk is that combined therapy with steroids and AZA/6-MP was not associated more strongly with zoster than use of AZA/6-MP alone. Furthermore, among the 22 patients with zoster who had received a prescription for AZA/6-MP in the preceding 30 days, 17 (77%) had started this medication more than 30 days before the diagnosis with zoster (data not shown), thereby likely reflecting chronic maintenance therapy. Thus, although it is not possible to completely rule-out confounding by indication, each of these suggests that our results are unlikely to be significantly biased by lack of data on disease activity. This study provides important insights into the relationship between zoster and IBD. Patients with IBD have a higher incidence rate of zoster compared with the general population, with drug-induced immunosuppression as an important risk factor for development of zoster among IBD patients. The risk of zoster with AZA/6-MP was as large as or larger than that with steroids. Given that patients with IBD represent a potentially high-risk population for zoster, future studies should focus on the efficacy and safety of varicella zoster virus vaccines in patients with IBD. References 1. Aberra FN, Lichtenstein GR. Methods to avoid infections in patients with inflammatory bowel disease. Inflamm Bowel Dis 2005; 11: Hambleton S, Gershon AA. Preventing varicella-zoster disease. Clin Microbiol Rev 2005;18: Aberra FN, Lewis JD, Hass D, et al. Corticosteroids and immunomodulators: postoperative infectious complication risk in inflammatory bowel disease patients. Gastroenterology 2003;125: Sohail MR, Khan AH, Holmes DR Jr, et al. Infectious complications of percutaneous vascular closure devices. Mayo Clin Proc 2005;80: Marchal L, D Haens G, Van Assche G, et al. The risk of postoperative complications associated with infliximab therapy for Crohn s disease: a controlled cohort study. Aliment Pharmacol Ther 2004;19: Bresnihan B, Cunnane G. Infection complications associated with the use of biologic agents. Rheum Dis Clin North Am 2003;29: Kleinschmidt-DeMasters BK, Tyler KL. Progressive multifocal leukoencephalopathy complicating treatment with natalizumab and interferon beta-1a for multiple sclerosis. N Engl J Med 2005; 353: Langer-Gould A, Atlas SW, Green AJ, et al. Progressive multifocal leukoencephalopathy in a patient treated with natalizumab. N Engl J Med 2005;353: 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: Gnann JW Jr, Whitley RJ. Clinical practice. Herpes zoster. N Engl J Med 2002;347: Oxman MN, Levin MJ, Johnson GR, et al. A vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. N Engl J Med 2005;352: Gilden DH. Varicella-zoster virus vaccine grown-ups need it, too. N Engl J Med 2005;352: Sands BE, Cuffari C, Katz J, et al. Guidelines for immunizations in patients with inflammatory bowel disease. Inflamm Bowel Dis 2004;10: Korelitz BI, Fuller SR, Warman JI, et al. Shingles during the course of treatment with 6-mercaptopurine for inflammatory bowel disease. Am J Gastroenterol 1999;94: Fraser AG, Orchard TR, Jewell DP. The efficacy of azathioprine for the treatment of inflammatory bowel disease: a 30 year review. Gut 2002;50: Present DH, Meltzer SJ, Krumholz MP, et al. 6-Mercaptopurine in the management of inflammatory bowel disease: short- and longterm toxicity. Ann Intern Med 1989;111: Gupta G, Gelfand JM, Lewis JD. Increased risk for demyelinating diseases in patients with inflammatory bowel disease. Gastroenterology 2005;129: Lewis JD, Aberra FN, Lichtenstein GR, et al. Seasonal variation in flares of inflammatory bowel disease. Gastroenterology 2004; 126: 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: Lewis JD, Brensinger C, Bilker WB, et al. Validity and completeness of the General Practice Research Database for studies of inflammatory bowel disease. Pharmacoepidemiol Drug Saf 2002; 11: Anonymous. The general practice research database: information for researchers. London: Office for National Statistics, Lis Y, Mann RD. The VAMP Research multi-purpose database in the U.K. J Clin Epidemiol 1995;48:

8 1490 GUPTA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No Garcia Rodriguez LA, Perez Gutthann S. Use of the UK General Practice Research Database for pharmacoepidemiology. Br J Clin Pharmacol 1998;45: Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991;302: Jick SS, Kaye JA, Vasilakis-Scaramozza C, et al. Validity of the general practice research database. Pharmacotherapy 2003;23: Lewis JD, Bilker WB, Brensinger C, et al. Inflammatory bowel disease is not associated with an increased risk of lymphoma. Gastroenterology 2001;121: Lewis JD, Schwartz JS, Lichtenstein GR. Azathioprine for maintenance of remission in Crohn s disease: benefits outweigh the risk of lymphoma. Gastroenterology 2000;118: Rothman KJ, Greenland S. Case-control studies. In: Rothman KJ, Greenland S, eds. Modern epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven, 1998: Irwin M, Costlow C, Williams H, et al. Cellular immunity to varicella-zoster virus in patients with major depression. J Infect Dis 1998;178(Suppl 1):S104 S Thomas SL, Hall AJ. What does epidemiology tell us about risk factors for herpes zoster? Lancet Infect Dis 2004;4: McCormick A, Charlton J. Morbidity statistics from general practice: 4th national study A study carried out by the Royal College of General Practitioners, the Office of Population Censuses and Surveys, and the Department of Health. London: HM Stationery Office, Schmader K, George LK, Burchett BM, et al. Race and stress in the incidence of herpes zoster in older adults. J Am Geriatr Soc 1998;46: Brisson M, Edmunds WJ. Epidemiology of Varicella-Zoster virus in England and Wales. J Med Virol 2003;70(Suppl 1):S9 S Franchimont D. Overview of the actions of glucocorticoids on the immune response: a good model to characterize new pathways of immunosuppression for new treatment strategies. Ann N Y Acad Sci 2004;1024: Mueller XM. Drug immunosuppression therapy for adult heart transplantation. Part 1: immune response to allograft and mechanism of action of immunosuppressants. Ann Thorac Surg 2004; 77: Pelaia G, Vatrella A, Cuda G, et al. Molecular mechanisms of corticosteroid actions in chronic inflammatory airway diseases. Life Sci 2003;72: Arnott ID, Watts D, Satsangi J. Azathioprine and anti-tnf alpha therapies in Crohn s disease: a review of pharmacology, clinical efficacy and safety. Pharmacol Res 2003;47: Stuck AE, Minder CE, Frey FJ. Risk of infectious complications in patients taking glucocorticosteroids. Rev Infect Dis 1989;11: Arbeter AM, Granowetter L, Starr SE, et al. Immunization of children with acute lymphoblastic leukemia with live attenuated varicella vaccine without complete suspension of chemotherapy. Pediatrics 1990;85: Takahashi M, Kamiya H, Baba K, et al. Clinical experience with Oka live varicella vaccine in Japan. Postgrad Med J 1985; 61(Suppl 4): Kennerfalk A, Ruigomez A, Wallander MA, et al. Geriatric drug therapy and healthcare utilization in the United Kingdom. Ann Pharmacother 2002;36: Strom BL. Pharmacoepidemiology. Chichester, NY: John Wiley, Address requests for reprints to: James D. Lewis, MD, MSCE, Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, 720 Blockley Hall, 423 Guardian Drive, Philadelphia, Pennsylvania jlewis@cceb.med.upenn.edu; fax: (215) Supported in part by a grant from the Agency for Healthcare Research and Quality, Centers for Education and Research on Therapeutics (U18-HS10399), and in part by an Alpha Omega Alpha Carolyn L. Kuckein Student Research Fellowship. J.D.L. has received consulting honorarium and research support from Wyeth and Johnson and Johnson for work unrelated to anti tumor necrosis factor therapies and from Centocor for work related to anti tumor necrosis factor therapies, and has received consulting honorarium from Merck, Berlex, and Elan. Note after submission: The US Food and Drug Administration has now approved Zostavax (Zoster Vaccine Live [Oka/Merck]) for prevention of herpes zoster (shingles) in individuals 60 years of age and older.