Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Antiviral Therapy 9:187-195 Skin reactions in patients with influenza treated with oseltamivir: a retrospective cohort study Beth L Nordstrom1*, Kelly Oh1, Susan T Sacks2 and Gilbert J L’Italien1 1 Ingenix Epidemiology, Auburndale, Mass., USA Roche Global Development, Nutley, NJ, USA 2 *Corresponding author: Tel: +1 617 552 5212; Fax: +1 617 244 9669; E-mail: [email protected] Oseltamivir phosphate is an FDA-approved treatment for influenza that has been available for prescription use in the USA since 1999. The present report describes findings from a post-marketing safety study of skin reactions associated with oseltamivir use. All patients in the claims-derived Ingenix Research Database with a physician diagnosis of influenza and/or a dispensing of oseltamivir between 1 December 1999 and 31 March 2002 were identified. Cohort eligibility criteria included minimum baseline enrolment duration of 3 months, age of at least 1 year and no influenza vaccination on the date of influenza diagnosis or oseltamivir dispensing. Patients were classified into two primary cohorts, influenza diagnosis and oseltamivir dispensing on the same day, and influenza diagnosis but no oseltamivir at any time, and a cohort included for secondary analyses comprising patients who received an oseltamivir dispensing without an influenza diagnosis on the same day. Outcomes included general skin reactions and several specific skin reactions. Events occurring during the 30 days following the date of influenza diagnosis or oseltamivir dispensing were examined using Cox proportional hazards models. Model covariates included age, use of another influenza drug, month and year of index date, and use of antitussives. Adjusted rate ratios for the general class of skin reactions among the primary cohort of oseltamivir users versus non-users were 1.05 (95% CI: 0.88–1.24) for incident cases and 0.98 (95% CI: 0.77–1.24) among patients with a history of a skin reaction. Similar results were seen for the other skin reaction categories, and secondary analyses investigating the oseltamivir users without influenza revealed no elevation in risk. It is concluded that oseltamivir use does not appear to be associated with an increased risk of skin reactions. Introduction Influenza is a highly contagious illness that is associated with an average of 36 000 deaths in the USA each year [1]. Although the primary approach to preventing the complications of influenza is vaccination, antiviral medications can serve as an adjunct to the vaccine for prevention and for treatment when the illness occurs [2]. Oseltamivir phosphate is a neuraminidase inhibitor that has been shown to be effective in decreasing the duration and severity of influenza symptoms [3,4]. The safety profile of oseltamivir to date has revealed few concerns, with the most common adverse events found to be minor upper gastrointestinal effects, including nausea and nausea with vomiting [4,5]. A cohort study comparing the incidence of cardiac, neuropsychiatric and respiratory events in patients using oseltamivir, compared with patients with influenza who did not take oseltamivir, found no increased risk of any of these outcomes [6]. During post-marketing use of the drug in clinical practice, ©2004 International Medical Press 1359-6535/02/$17.00 rashes have been reported in patients using oseltamivir [7]. Spontaneous reports of other skin reactions, ranging from mild to severe and including, for example, Stevens-Johnson syndrome, pruritis, urticaria and bullous conditions, among other dermatological diagnoses, have also been received since the drug’s approval. Such adverse event reports raise the question of whether the use of oseltamivir may place patients at heightened risk of experiencing skin reactions. A serious limitation to adverse event reporting, however, is the inability to determine whether the drug was in any way causally related to the reaction. Postmarketing epidemiological studies including a broader population of patients than those included in the clinical trials can help to answer this question by demonstrating whether the drug is associated with an increased risk of skin reactions. The present report describes a retrospective cohort study using medical claims data to examine the risk of various skin reactions associated with use of oseltamivir. 187 BL Nordstrom et al. Materials and methods Data source Data for the study were obtained from the Ingenix Research Database, which consists of demographic, medical and prescription data from UnitedHealthcare. UnitedHealthcare is one of the largest healthcare companies in the USA, with more than 300000 physicians contracted to provide healthcare services to over 14 million members. The Research Database was designed to facilitate drug safety and health outcomes research. UnitedHealthcare members are brought into the Ingenix Research Database if they belong to plans in which the costs of pharmaceuticals and health services are reimbursed to providers on an item-by-item basis, so that it is possible to reconstruct all healthcare utilization. Beginning with the earliest claims, in 1990, the Research Database has detailed information on approximately 8 000000 current and past members. Study population The source population for the study included patients from 25 UnitedHealthcare plans from the following 21 states: Alabama, Arizona, Arkansas, Florida, Georgia, Illinois, Kentucky, Louisiana, Massachusetts, Michigan, Mississippi, Missouri, Nebraska, North Carolina, Ohio, Rhode Island, South Carolina, Tennessee, Texas, Utah and Virginia. The study population included patients who received either a dispensing of oseltamivir or an ICD-9 diagnosis code of influenza (487.*) associated with a physician visit or hospitalization between 1 December 1999 and 31 March 2002. Oseltamivir became available in November 1999, but was rarely prescribed in UnitedHealthcare before 1 December 1999. Since the few people taking oseltamivir in its first month on the market may not represent the general population of users, we excluded any oseltamivir dispensing and influenza diagnoses before 1 December 1999. The date of oseltamivir dispensing for those who received oseltamivir, and of influenza diagnosis for those who did not receive the drug, was designated as the index date. Oseltamivir has been approved by the FDA for use in children over 1 year of age. We thus excluded from the study any infants less than 1 year old on the index date. Since influenza is a seasonal illness, we examined only diagnoses and dispensing occurring during the influenza season, November through March. Any patient with oseltamivir dispensing and/or influenza diagnoses during different influenza seasons was allowed to enter the cohort up to once per influenza season, for a maximum of three episodes included in the analysis. If more than one influenza diagnosis or oseltamivir dispensing occurred during the same season, only the first was included in the analysis. 188 To allow assessment of baseline variables, patients were required to have a minimum of 3 months of enrolment in the health plan before their index date. We assumed that any diagnosis of influenza given on the same day as an influenza vaccination did not indicate a genuine case of influenza, and thus excluded any patient with an influenza vaccination on the index date. The study focused on two primary cohorts. The influenza with oseltamivir cohort included patients meeting the above inclusion criteria who received a dispensing of oseltamivir on the same date as the diagnosis of influenza. Patients who received no dispensing of oseltamivir at any time, but who had a diagnosis of influenza, constituted the influenza without oseltamivir cohort. Patients who received a dispensing of oseltamivir but did not have a diagnosis of influenza on the date of dispensing were examined separately from the two primary cohorts in what must be considered an exploratory fashion only, as it was not possible to construct an appropriate comparison group with similar baseline health status. The oseltamivir without influenza cohort was compared to the influenza with oseltamivir cohort to investigate potential differences between the two groups of patients who received oseltamivir. Outcome definition The outcomes assessed consisted of various categories of skin reactions and other allergic responses. We defined a general skin reaction outcome, including inflammatory skin conditions and other skin diseases, several specific skin reactions, and allergic or other adverse reaction to a drug. The outcomes were defined using the following ICD-9 diagnosis codes: general skin reaction (690.*– 698.* and 700.*– 709.*), atopic dermatitis (691.8), contact dermatitis and other eczema (692.*), dermatitis due to substances taken internally (693.*), dermatitis herpetiformis (694.0), impetigo herpetiformis (694.3), pemphigus (694.4), pemphigoid (694.5), other specified bullous dermatoses (694.8), erythematous conditions (695.* excluding 695.3 and 695.8), unspecified pruritic disorder (698.9), urticaria (708.*) and adverse reaction to a drug (995.2). Appropriate diagnoses given in conjunction with a physician visit or hospitalization during the 30 days following the index date were considered to be outcome events. Baseline and covariate measures For each patient we identified age, sex, geographic area (northeast, southeast, midwest or west), and month and year of influenza diagnosis. We also determined the influenza subcode given, specifically influenza with the fourth digit unspecified (487), influenza with pneumonia (487.0), influenza with other respiratory ©2004 International Medical Press Oseltamivir and skin reactions manifestations (487.1) or influenza with other manifestations (487.8). Patients were classified as using or not using each of several classes of drugs, including amantadine, rimantadine, zanamivir, any influenza medication (that included any of the preceding three drugs), ribavirin, antibiotics (also separated into cephalosporins, macrolides, penicillins, tetracyclines, quinolones and sulfonamides), antitussives and analgesics, during the 3-month baseline period. Pre-existing cancer, diabetes, rheumatoid arthritis and HIV, and history of each of the study outcomes were noted if the patient had an appropriate diagnosis code in conjunction with a physician visit or hospitalization during the baseline period. All primary analyses consisted of comparisons between the influenza with oseltamivir and influenza without oseltamivir cohorts. Exploratory analyses compared the oseltamivir without influenza to the influenza with oseltamivir cohort. Differences between groups on baseline characteristics were assessed using χ2 tests. For the general skin reaction outcome and each of the 11 specific skin reaction outcomes, we performed Cox proportional hazards regressions, in which adjusted incidence rate ratios were estimated for the oseltamivir users compared to the non-users. For statistical reasons, proportional hazards models were computed only for outcomes with at least 20 events. We investigated potential confounding from all baseline health and demographic characteristics described above. Drugs given during the baseline period were investigated both within lower-level (such as, amantadine, cephalosporins) and higher-level groupings (such as, influenza medications, antibiotics). The final Cox models included any of these variables that modified the hazard ratio for oseltamivir exposure by at least 5%. To separate incident cases of each event from cases that may be ongoing or recurring, the adjusted models were calculated separately for those with and those without a history of the outcome diagnoses during the baseline period. oseltamivir users and 10 914 influenza patients without oseltamivir. Only nine of the oseltamivir users were excluded because of the age restriction, in contrast with 1752 other influenza patients who were less than 1 year old on the index date. Fifty-two oseltamivir users and 365 influenza patients without oseltamivir were given an influenza vaccine on the index date, leading to their exclusion from the study. Please note that the same patient may have met more than one of these exclusion criteria. The final study cohorts included 11 625 patients with a dispensing of oseltamivir on the same day as an influenza diagnosis, 70 999 patients with influenza but no oseltamivir, and 17 365 patients who received oseltamivir without a diagnosis of influenza on the index date. Approximately 8% (1351) of the oseltamivir patients without an influenza diagnosis on the date of dispensing did receive a diagnosis of influenza on a different day than the index date. There were 1665 patients who received a diagnosis of influenza without oseltamivir in two different influenza seasons and 77 patients in all three seasons under study. A total of 66 patients received oseltamivir with an influenza diagnosis during two different influenza seasons, but none during all three seasons studied, while 225 and 10 patients, respectively, received oseltamivir without an influenza diagnosis during two and three seasons. Some patients also appeared in different cohorts during different influenza seasons; 259 were diagnosed with influenza both with and without oseltamivir, 100 received oseltamivir both with and without an influenza diagnosis, 568 were diagnosed with influenza without oseltamivir in one season and received oseltamivir without influenza during a different season, and three patients entered all three study cohorts in the three seasons. Allowing each patient to contribute one episode to each influenza season in which he or she met the inclusion criteria resulted in a total count for each cohort of 11 691 influenza with oseltamivir patients, 72 818 influenza patients without oseltamivir, and 17 610 oseltamivir users without influenza. Results Baseline characteristics A total of 32 459 patients received one or more dispensings of oseltamivir during the 1999 to 2002 influenza seasons. An additional 82 993 patients had one or more physician visits or hospitalizations with a diagnosis of influenza but did not receive a dispensing of oseltamivir during the same influenza season as the diagnosis. Approximately 13% of patients were excluded because of failure to meet the 3-month baseline enrolment criterion; this led to the exclusion of 3956 Demographic and other baseline characteristics of the final study cohorts are presented in Table 1. Children under the age of 13 with an influenza diagnosis were less likely to receive oseltamivir than older individuals, though otherwise the age distributions of the two influenza cohorts were fairly similar, overall χ2(6)=3090.4, P<0.001. The patients who received oseltamivir without an influenza diagnosis were more likely to be older than the oseltamivir users with influenza, χ2(6)=426.0, P<0.001. There was a slightly lower proportion of females among the influenza with Data analysis Antiviral Therapy 9:2 189 BL Nordstrom et al. Table 1. Demographic characteristics of study cohorts Influenza with oseltamivir (n=11 691) Influenza without oseltamivir (n=72 818) Age group (years) 1–12 13–19 20–29 30–39 40–49 50–64 65+ 694 (5.9%) 1059 (9.1%) 2308 (19.7%) 3108 (26.6%) 2591 (22.2%) 1844 (15.8%) 87 (0.7%) Gender Male Female Region Northeast Southeast Month of index date Season 1 December 1999 January 2000 February 2000 March 2000 Season 2 November 2000 December 2000 January 2001 February 2001 March 2001 Season 3 November 2001 December 2001 January 2002 February 2002 March 2002 P–value* Oseltamivir without influenza (n=17 610) P–value † 20 554 (28.2%) 8207 (11.3%) 9297 (12.8%) 13 066 (17.9%) 11 740 (16.1%) 9175 (12.6%) 779 (1.1%) <0.001 657 (3.7%) 1196 (6.8%) 2662 (15.1%) 4408 (25.0%) 4618 (26.2%) 3780 (21.5%) 289 (1.6%) <0.001 5670 (48.5%) 6021 (51.5%) 34 182 (46.9%) 38 636 (53.1%) 0.002 8231 (46.7%) 9379 (53.3%) 0.003 150 (1.3%) 7029 (60.1%) 3441 (4.7%) 30 302 (41.6%) 279 (1.6%) 10 140 (57.6%) 810 (6.9%) 2027 (17.3%) 640 (5.5%) 81 (0.7%) 10 354 (14.2%) 15 023 (20.6%) 4867 (6.7%) 1666 (2.3%) 1232 (7.0%) 3121 (17.7%) 1086 (6.2%) 255 (1.4%) 153 (1.3%) 438 (3.7%) 1677 (14.3%) 1679 (14.4%) 310 (2.7%) 1966 (2.7%) 3402 (4.7%) 8632 (11.9%) 6771 (9.3%) 2398 (3.3%) 480 (2.7%) 938 (5.3%) 2767 (15.7%) 2125 (12.1%) 610 (3.5%) 133 (1.1%) 227 (1.9%) 863 (7.4%) 1794 (15.3%) 859 (7.3%) 1495 (2.1%) 1659 (2.3%) 3470 (4.8%) 6969 (9.6%) 4146 (5.7%) 291 (1.7%) 468 (2.7%) 1052 (6.0%) 2074 (11.8%) 1111 (6.3%) <0.001 <0.001 Values are expressed as number (%). *P-value obtained from χ2 test comparing influenza without oseltamivir to influenza with oseltamivir. † P-value obtained from χ2 test comparing oseltamivir without influenza to influenza with oseltamivir. oseltamivir cohort than the influenza without oseltamivir, χ2(1)=9.80, P=0.002, and oseltamivir without influenza cohorts, χ2(1)=8.71, P=0.003. Regional differences were seen, with a higher proportion of oseltamivir users with influenza in the southeast, and a lower proportion in the midwest and northeast compared with non-users, χ2(3)=1537.6, P<0.001, and oseltamivir users without influenza, χ2(3)=24.98, P<0.001. Use of oseltamivir peaked in January and February of each year, with the peaks somewhat sharper than the peaks in influenza diagnoses in the same months, χ2(13)=1579.1, P<0.001. The decline in numbers of patients with influenza across the three seasons reflects the patterns seen in the 190 USA as a whole, with the highest incidence of influenza occurring during the 1999–2000 season [8–10]. Table 2 presents the frequency of diagnoses and drugs given during the 91-day period from 3 months before the index date through the index date. Most influenza diagnoses given on the index date were coded as 487.1, influenza with other respiratory manifestations. The oseltamivir users with influenza were somewhat less likely to receive a diagnosis of influenza with pneumonia than were the non-users, overall χ2(3)=88.70, P<0.001. Most influenza patients did not receive any influenza medication. Most patients who did use an influenza drug used only one antiviral. Approximately ©2004 International Medical Press Oseltamivir and skin reactions 17% of the non-users received an influenza drug other than oseltamivir, whereas only 0.5% of the influenza with oseltamivir group received another antiviral in addition to oseltamivir, χ2(1)=2250.2, P<0.001. Amantadine was the most frequently dispensed of the other three influenza drugs. The oseltamivir users without influenza were about twice as likely to receive another influenza medication in addition to oseltamivir than the primary oseltamivir cohort, χ2(1)=25.26, P<0.001, though prevalence of use within this group was still only 1%. Ribavirin was very rarely used in any cohort (results not shown). The oseltamivir users with influenza were somewhat less likely to use antibiotics than were the non-users, χ2(1)=23.31, P<0.001 and the oseltamivir users without influenza, χ2(1)=4.81, P=0.03. The influenza patients receiving oseltamivir were more likely than the comparison influenza patients, χ2(1)=1046.0, P<0.001, and than the oseltamivir users without influenza, χ2(1)=767.3, P<0.001, to receive antitussives during the baseline period. The primary oseltamivir cohort was also more likely than non-users to receive analgesics, χ2(1)=207.7, P<0.001. The oseltamivir without influenza cohort showed a similar prevalence of analgesic use to the oseltamivir with influenza cohort, χ2(1)=1.58, P=0.21. Table 2. Diagnoses and drugs given during the 3-month baseline period Influenza with oseltamivir (n=11 691) Influenza without oseltamivir (n=72 818) Influenza ICD-9 diagnosis code Influenza, no fourth digit (487) Influenza with pneumonia (487.0) Influenza with other respiratory manifestations (487.1) Influenza with other manifestations (487.8) 300 (2.6%) 243 (2.1%) 10822 (92.6%) 326 (2.8%) Drug use history Amantadine Rimantadine Zanamivir Any influenza medication Cephalosporins Macrolides Penicillins Tetracyclines Quinolones Sulfonamides Any antibiotic Antitussives Analgesics Diagnostic history Diabetes (250) Cancer (140–209) Rheumatoid arthritis (714) HIV (042, V08) General skin reaction (690–698, 700–709) Atopic dermatitis (691.8) Contact dermatitis and other eczema (692) Dermatitis due to substance taken internally (693) Erythematous conditions (695 excluding 695.3 and 695.8) Unspecified pruritic disorder (698.9) Urticaria (708) Adverse reaction to drug (995.2) P–value* Oseltamivir without influenza (n=17 610) P–value † 1670 (2.3%) 2593 (3.6%) 65999 (90.6%) 2556 (3.5%) <0.001 N/A N/A N/A N/A N/A 21 (0.2%) 25 (0.2%) 7 (0.1%) 53 (0.5%) 941 (8.0%) 2458 (21.0%) 1613 (13.8%) 370 (3.2%) 684 (5.9%) 9 (0.1%) 5012 (42.9%) 4701 (40.2%) 1557 (13.3%) 5435 (7.5%) 3877 (5.3%) 3343 (4.6%) 12616 (17.3%) 6250 (8.6%) 14644 (20.1%) 14162 (19.4%) 1778 (2.4%) 3068 (4.2%) 81 (0.1%) 32960 (45.3%) 18771 (25.8%) 6609 (9.1%) <0.001 <0.001 <0.001 <0.001 0.06 0.02 <0.001 <0.001 <0.001 0.29 <0.001 <0.001 <0.001 55 (0.3%) 66 (0.4%) 54 (0.3%) 172 (1.0%) 1391 (7.9%) 3963 (22.5%) 2295 (13.0%) 568 (3.2%) 1441 (8.2%) 18 (0.1%) 7778 (44.2%) 4389 (24.9%) 2436 (13.8%) 0.03 0.02 <0.001 <0.001 0.64 0.003 0.06 0.77 <0.001 0.49 0.03 <0.001 0.21 261 (2.2%) 77 (0.7%) 29 (0.2%) 6 (0.1%) 681 (5.8%) 30 (0.3%) 143 (1.2%) 9 (0.1%) 5 (0.04%) 1711 (2.3%) 465 (0.6%) 154 (0.2%) 41 (0.1%) 4484 (6.2%) 308 (0.4%) 1221 (1.7%) 78 (0.1%) 49 (0.07%) 0.44 0.80 0.43 0.83 0.16 0.008 <0.001 0.35 0.33 529 (3.0%) 227 (1.3%) 52 (0.3%) 34 (0.2%) 1075 (6.1%) 37 (0.2%) 211 (1.2%) 13 (0.1%) 10 (0.06%) <0.001 <0.001 0.36 0.001 0.32 0.41 0.85 0.92 0.60 6 (0.05%) 31 (0.3%) 25 (0.2%) 83 (0.11%) 333 (0.5%) 145 (0.2%) 0.05 0.003 0.74 17 (0.10%) 53 (0.3%) 36 (0.2%) 0.18 0.58 0.86 Values are expressed as number (%). *P-value obtained from χ2 test comparing influenza without oseltamivir to influenza with oseltamivir. † P-value obtained from χ2 test comparing oseltamivir without influenza to influenza with oseltamivir. Antiviral Therapy 9:2 191 BL Nordstrom et al. History of diabetes, cancer, rheumatoid arthritis, HIV and each of the skin reaction outcomes did not differ markedly among the cohorts, though some comparisons did achieve statistical significance at the 0.05 level. The prevalence of diabetes, cancer and HIV were similar between the two influenza cohorts, but the oseltamivir without influenza group showed a higher prevalence of each, χ2(1)=15.94, P<0.001 for diabetes, χ2(1)=27.20, P<0.001 for cancer and χ2(1)=10.36, P=0.001 for HIV. Although the general skin reaction was seen in approximately 6% of patients, the frequency of each of the specific reactions investigated was low, with all diagnoses except contact dermatitis and other eczema present in less than 1% of patients. Compared with the oseltamivir users with influenza, the non-users showed a somewhat higher prevalence of atopic dermatitis, χ2(1)=7.00, P=0.008, contact dermatitis and other eczema, χ2(1)=13.05, P<0.001, unspecified pruritic disorder, χ2(1)=3.76, P=0.05 and urticaria, χ2(1)=8.67, P=0.003. In contrast, baseline rates of each skin reaction were similar between the two oseltamivir cohorts. Diagnoses of dermatitis herpetiformis, impetigo herpetiformis, pemphigus, pemphigoid and other specified bullous dermatoses were seen in almost no patients in the baseline or follow-up period, so the results for these skin reactions are not shown. Skin reaction outcomes Table 3 shows the frequency of each skin reaction outcome during the 30 days following the index date for the two primary study cohorts, along with crude and adjusted incidence rate ratios obtained from Cox proportional hazards models. The adjusted incidence rate ratios were computed separately for those with and without a history of the outcome diagnoses during the 3-month baseline period. Age, grouped into the categories shown in Table 1, and use of another influenza medication during the baseline period were found to be important predictors of almost all of the skin reaction outcomes, so we included these two variables in all models. The analysis of the general skin reaction outcome included only age and use of another influenza drug as covariates. For this broad class of events, we observed little difference in risk between the two groups, either with or without a history of the outcome. Figure 1 presents Kaplan-Meier curves showing the cumulative incidence of this event for the two groups. It is clear from the figure that the risk of a general skin reaction was much higher among those with a history of a skin reaction than among those who had not received a skin reaction diagnosis during the 3 months prior to the index date. The figure also reveals that the skin reactions occurred at a relatively constant rate across the 192 month of follow-up; if influenza and/or oseltamivir caused skin reactions concurrent with the illness or drug use, the curves should have shown a sharper increase during the first several days followed by a levelling off as patients recovered from influenza and stopped using oseltamivir. Atopic dermatitis occurred in a total of 82 patients in the two cohorts. For this outcome, use of antitussives during the baseline period modified the hazard ratio for exposure by more than 5%, so this variable was included in the model along with age and use of another influenza medication. The adjusted incidence rate ratio for the patients without a history of atopic dermatitis was 0.81, though the confidence interval for this rate ratio was fairly wide. Since there were fewer than 20 events among those with a history of the outcome, we did not construct a Cox model for this subgroup. Contact dermatitis and other eczema occurred more frequently, so we were able to perform adjusted analyses for both patients with and without a history of this outcome. Only age group and use of another influenza drug were included in these models. For the patients without a history of the outcome, the adjusted incidence rate ratio was 0.79. The confidence interval was narrower than in the analysis of atopic dermatitis, but it still included one, suggesting that the apparently lower risk in the oseltamivir group may not reflect a true difference. The rate ratio for the group with a history of contact dermatitis and other eczema was close to one. Dermatitis due to a substance taken internally is of particular interest in this study, because a skin reaction ascribed by the physician to oseltamivir may be assigned this code. There were, however, only 43 of these events, which limits the power of our analysis. Fewer than 20 events occurred in the group with a history of the event, so an adjusted incidence rate ratio is shown only for the group without a history of dermatitis due to a substance. Month (November through March) and use of antitussives during the baseline period were important predictors in the model, so these two variables were added as covariates. The adjusted incidence rate ratio for oseltamivir exposure was somewhat greater than one, but the confidence interval was very wide, so we cannot conclude that oseltamivir was associated with an increased risk of this event. Erythematous conditions were seen only among the non-users. The model for this event, computed only for the group without a history of the outcome, included only the age and other influenza drug covariates. The incidence rate ratio was zero and the exact confidence limit crossed one but closely approached statistical significance. With so few events, however, these results must be interpreted with caution. ©2004 International Medical Press Oseltamivir and skin reactions Table 3. Incidence rate ratios of skin reactions for influenza cohorts in 30 days following index date Number of outcomes General skin reaction Atopic dermatitis Contact dermatitis and other eczema Dermatitis due to substance taken internally Erythematous conditions Unspecified pruritic disorder Urticaria Adverse reaction to drug Influenza with oseltamivir (n=11 691) Influenza without oseltamivir (n=72 818) Crude incidence rate ratio* Adjusted incidence rate ratio, no history Adjusted incidence rate ratio, history 247 7 47 7 0 5 18 16 1440 75 356 36 29 24 132 74 1.06 (0.93, 1.22) 0.58 (0.27, 1.26) 0.82 (0.60, 1.11) 1.21 (0.54, 2.71) 0 (0.00, 0.84) 1.29 (0.49, 3.39) 0.85 (0.52, 1.38) 1.35 (0.78, 2.31) 1.05 (0.88, 1.24) † 0.81 (0.36, 1.84) § 0.79 (0.56, 1.12) † 1.26 (0.53, 3.01) ‡ 0 (0.00, 1.03) † 1.27 (0.45, 3.56) ¢ 1.27 (0.72, 2.25) § 1.10 (0.60, 1.99) # 0.98 (0.77, 1.24) † N/A 0.94 (0.46, 1.91) † N/A N/A N/A 1.01 (0.30, 3.42) § N/A *Incidence rate ratios are presented as rate ratio (95% CI). † Rate ratio adjusted for age and use of another influenza drug. § Rate ratio adjusted for age, use of another influenza drug and use of antitussives. ‡ Rate ratio adjusted for age, use of another influenza drug, month of index date and use of antitussives. ¢ Rate ratio adjusted for age, use of another influenza drug and year of index date. # Rate ratio adjusted for age, use of another influenza drug and month of index date. A total of 29 patients had a diagnosis of unspecified pruritic disorder during the outcome period. Again the analyses were computed only for the group without a history of the outcome. Year of index date was found to modify the hazard ratio for exposure, so it was added to age and use of another influenza drug as a covariate. The adjusted incidence rate ratio was slightly over one but again was made questionable because of low power; the confidence interval ranged from less than 0.5 to 3.5. Urticaria was more common, with 150 patients receiving this diagnosis. Use of antitussives in the baseline period was an important predictor and was added to the models both for those with and those without a history of the outcome. For patients without a history of urticaria during the baseline period, the adjusted incidence rate ratio was 1.27, but again the confidence interval is sufficiently wide that we cannot conclude that oseltamivir was associated with an increased risk of this outcome. For those with Figure 1. General skin reaction between the two influenza cohorts for 30 days after the index date 0.14 Oseltamivir, with history of outcome Proportion of patients with event 0.12 0.10 No oseltamivir, with history of outcome 0.08 0.06 0.04 Oseltamivir, no history of outcome 0.02 No oseltamivir, no history of outcome 0.00 0 5 10 15 20 25 30 Days since index date Plots are separate for patients with and without claims indicating a history of the outcome diagnoses given during the 3-month period prior to the index date. Antiviral Therapy 9:2 193 BL Nordstrom et al. a history of the outcome, the rate ratio was approximately one. Adverse reaction to a drug occurred in a total of 90 patients during follow-up. Month of the index date modified the hazard ratio for exposure, so this variable was added to the model. The adjusted incidence rate for patients without a history of the diagnosis was slightly greater than one but with a wide confidence interval, so that again there is no clear sign of an increase in risk of this outcome with oseltamivir. Five of the specific outcomes, dermatitis herpetiformis, impetigo herpetiformis, pemphigus, pemphigoid and other specified bullous dermatoses, were extremely rare in both cohorts, so we were unable to perform survival analyses of these outcomes. There were no incident cases of any of these disorders in the oseltamivir cohort, so there was clearly no indication of an elevation of risk associated with oseltamivir. Comparisons of the skin reaction outcomes between the two oseltamivir groups are presented in Table 4. Many outcomes failed to reach the 20 event minimum we had set for performing Cox regressions. For all events that we were able to analyse, including the general skin reaction, contact dermatitis and other eczema, urticaria and adverse reaction to a drug, the oseltamivir without influenza cohort showed no sign of any increase in risk compared with the primary oseltamivir cohort. All confidence intervals were one or less, though none were significantly (P<0.05) less than one. Discussion The skin reactions investigated as outcomes after oseltamivir and/or influenza diagnosis were found to be uncommon in the present cohorts. The broadest class of diagnoses, consisting of a wide range of inflammatory skin conditions and other skin diseases, was seen in approximately 2% of patients during the month following the index date. The more specific reactions examined were overall very rare. We found no indication of an increase in risk of any of these events associated with use of oseltamivir. We must, however, question the frequency with which these skin reactions, especially mild to moderate reactions of fairly brief duration, are reported on medical claims. Under-reporting of such events seems likely to occur similarly across patients who did and those who did not receive a dispensing of oseltamivir, so that the effect of this under-reporting is most likely to reduce the statistical power of our comparisons rather than to bias the size of the effect. Although severe reactions including StevensJohnson syndrome have been reported with post-marketing use of oseltamivir, we did not find a heightened risk of these reactions among patients receiving oseltamivir in the present database. StevensJohnson syndrome cannot be identified in claims data with any precision, because there is no unique ICD-9 code associated with it. In our skin reaction categories, this reaction would fall under the heading of erythematous conditions, of which we identified none in the primary oseltamivir cohort and only four among the oseltamivir users without influenza. The power of this study may have been inadequate to detect a very rare event. If the true incidence of erythematous conditions among users of oseltamivir was one in 10 000, the likelihood of our detecting such events in the primary oseltamivir cohort would be 0.69, but if the true incidence was one in 100 000, the probability of detecting the events would be only 0.11. We cannot, therefore, Table 4. Incidence rate ratios of skin reactions for oseltamivir cohorts in 30 days following index date Number of outcomes General skin reaction Atopic dermatitis Contact dermatitis and other eczema Dermatitis due to substance taken internally Erythematous conditions Unspecified pruritic disorder Urticaria Adverse reaction to drug Influenza with oseltamivir (n=11 691) Oseltamivir without influenza (n=17 610) Crude incidence rate ratio* Adjusted incidence rate ratio, no history Adjusted incidence rate ratio, history 247 7 47 7 0 5 18 16 341 10 63 4 4 10 15 18 0.92 (0.78, 1.08) N/A 0.89 (0.61, 1.30) N/A N/A N/A 0.55 (0.28, 1.10) 0.75 (0.38, 1.46) 0.99 (0.81, 1.21) † N/A 1.00 (0.66, 1.51) † N/A N/A N/A 0.57 (0.26, 1.23) § 0.88 (0.43, 1.79) ‡ 0.77 (0.57, 1.03) † N/A N/A N/A N/A N/A N/A N/A *Incidence rate ratios are presented as rate ratio (95% CI). † Rate ratio adjusted for age and use of another influenza drug. § Rate ratio adjusted for age, use of another influenza drug and use of antitussives. ‡ Rate ratio adjusted for age, use of another influenza drug and month of index date. 194 ©2004 International Medical Press Oseltamivir and skin reactions be confident that such an extremely rare event as Stevens-Johnson syndrome is likely to be would have appeared in the present study. For other severe reactions that can be ascertained from medical claims, such as pemphigus and pemphigoid, we found no incident cases among oseltamivir users. Unlike many mild reactions, severe reactions should be given some form of medical care that will ensure that the diagnosis appears on health insurance claims, making it unlikely that such events were missed in the present study. Given the limitations of the diagnostic coding system, however, it is difficult to determine which events may be considered severe. Over half of the patients who received a dispensing of oseltamivir did so without having a diagnosis of influenza noted in their medical claims at any time during the same influenza season. The resulting oseltamivir without influenza cohort probably consisted of patients using oseltamivir prophylactically, patients who did have influenza but simply called their physicians rather than making an office visit, and patients who received an oseltamivir dispensing on a different day from the date of diagnosis of influenza. The proportion of patients in the former two categories is impossible to determine using claims data alone, as telephone calls do not result in medical claims. The finding of similar risks of each outcome between the primary oseltamivir cohort and the oseltamivir cohort without an influenza diagnosis provides reassurance that the influenza with oseltamivir group was fairly representative of all oseltamivir users, and hence that the principal comparisons between influenza patients with and without oseltamivir were valid. Please note that oseltamivir exposure may have been misclassified for some patients, as those who received a pharmacy dispensing of the drug may not have taken it, while those with no claims for oseltamivir dispensing in the database may have obtained the drug through a different source than a pharmacy dispensing processed by their healthcare insurer (such as, samples provided in the physician’s office). In conclusion, the skin reactions investigated occurred infrequently in this claims-based study of influenza patients. Oseltamivir use appears not to be associated with an increase in risk of any of the skin reactions examined. This finding holds both for incident cases and for recurring reactions, suggesting that oseltamivir is likely to be safe with respect to skin reactions even among those with a history of skin disorders. Source of support Supported by a contract between Roche Global Development and Ingenix Epidemiology. References 1. Centers for Disease Control and Prevention. Update: Influenza activity – United States, 2002–03 season. Morbidity & Mortality Weekly Report 2003; 52:26–28. 2 Centers for Disease Control and Prevention. Prevention and control of influenza: recommendations of the Advisory Committee on Immunization Practices (ACIP). Morbidity & Mortality Weekly Report 2001; 50:1–46. 3. Hayden FG, Treanor JJ, Fritz RS, Lobo M, Betts RF, Miller M, Kinnersley N, Mills RG, Ward P & Straus SE. Use of the oral neuraminidase inhibitor oseltamivir in experimental human influenza: randomized controlled trials for prevention and treatment. Journal of the American Medical Association 1999; 282:1240–1246. 4. Nicholson KG, Aoki FY, Osterhaus AD, Trottier S, Carewicz O, Mercier CH, Rode A, Kinnersley N & Ward P. Efficacy and safety of oseltamivir in treatment of acute influenza: a randomised controlled trial. Lancet 2000; 355:1845–1850. 5. Treanor JJ, Hayden FG, Vrooman PS, Barbarash R, Bettis R, Riff D, Singh S, Kinnersley N, Ward P & Mills RG. Efficacy and safety of the oral neuraminidase inhibitor oseltamivir in treating acute influenza: a randomized controlled trial. Journal of the American Medical Association 2000; 283:1016–1024. 6. Enger C, Nordstrom BL, Thakrar B, Sacks S & Rothman KJ. Health outcomes among patients receiving oseltamivir. Pharmacoepidemiology & Drug Safety (In Press). 7. Roche Laboratories, Inc. Tamiflu (oseltamivir phosphate) capsules and for oral suspension (Package insert). Roche Laboratories, Inc.: Nutley, NJ, USA. 2001. 8. Centers for Disease Control and Prevention. Update: Influenza activity – United States and worldwide, 19992000 season and composition of the 2000–2001 influenza vaccine. Morbidity & Mortality Weekly Report 2000; 49:375–381. 9. Centers for Disease Control and Prevention. Update: Influenza activity – United States and worldwide, 2000-01 season, and composition of the 2001–2002 influenza vaccine. Morbidity & Mortality Weekly Report 2001; 50:466–470. 10. Centers for Disease Control and Prevention. Update: Influenza activity – United States and worldwide, 20012002 season and composition of the 2002–2003 influenza vaccine. Morbidity & Mortality Weekly Report 2002; 51:503–506. Received 20 August 2003; accepted 31 October 2003 Antiviral Therapy 9:2 195