The contents herein are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
Address correspondence to Cynthia S. Crowson, MS, Mayo Clinic, 200 First Street SW, Rochester, MN 55905. E-mail: email@example.com.
To determine the incidence, time trends, risk factors, and severity of herpes zoster in a population-based cohort of patients with newly diagnosed rheumatoid arthritis (RA) compared to a group of individuals without RA from the same population.
All residents of Olmsted County, Minnesota fulfilling for the first time the 1987 American College of Rheumatology criteria for RA between January 1, 1980 and December 31, 2007 and a cohort of similar residents without RA were assembled and followed by retrospective chart review until death, migration, or December 31, 2008.
There was no difference in the presence of herpes zoster prior to the RA incidence/index date between the cohorts (P = 0.85). During followup, 84 patients with RA (rate 12.1 cases per 1,000 person-years) and 44 subjects without RA (rate 5.4 cases per 1,000 person-years) developed herpes zoster. Patients with RA were more likely to develop herpes zoster than those without RA (hazard ratio [HR] 2.4 [95% confidence interval (95% CI) 1.7–3.5]). Herpes zoster occurred more frequently in patients diagnosed with RA more recently (HR 1.06 per year [95% CI 1.02–1.10]). Erosive disease, previous joint surgery, and use of hydroxychloroquine and corticosteroids were significantly associated with the development of herpes zoster in RA. There was no apparent association of herpes zoster with the use of methotrexate or biologic agents. Complications of herpes zoster occurred at a similar rate in both cohorts.
The incidence of herpes zoster is increased in RA and has risen in recent years. There also has been an increasing incidence of herpes zoster in more recent years in the general population. RA disease severity is associated with the development of herpes zoster.
Herpes zoster, or shingles, is a common cutaneous disorder caused by the reactivation of latent varicella-zoster virus dormant in the cranial nerve or dorsal root ganglia. It usually manifests as a dermatomal distribution of a vesicular eruption that can cause significant morbidity . Herpes zoster is known to occur more frequently in patients with conditions that depress cell-mediated immunity, including malignancy, human immunodeficiency virus (HIV), transplant-related conditions, immunosuppressive disorders, and treatment with immunosuppressants .
The overall rate of herpes zoster infections in patients with rheumatoid arthritis (RA) is greater than in the general population, with an increased risk in individuals receiving traditional disease-modifying antirheumatic drugs (DMARDs) or biologic agents . The use of prednisone in RA is an important predisposing factor for herpes zoster and, when taken in conjunction with DMARDs, the risk is increased beyond that of DMARDs alone . Many unanswered questions remain regarding the relationship between RA and reactivation of herpes zoster and how RA-related disease features, treatment, and herpes zoster prevention measures might be addressed to avoid recurrent and complicated herpes zoster infections in RA patients.
We performed a population-based study to assess the incidence, time trends, risk factors, and severity of herpes zoster in a well-defined population of patients with RA. The information provided herein should serve the eventual goal of identifying possible strategies to minimize the risk of development of herpes zoster in patients with RA.
Significance & Innovations
This large population-based cohort study reports an increased incidence of herpes zoster in patients with rheumatoid arthritis (RA) compared to the general population.
The occurrence of herpes zoster has risen during recent years in both RA and non-RA subjects but remains consistently higher in RA subjects.
RA disease severity and the use of some antirheumatic medications (in particular corticosteroids) appear to be associated with increased risk of herpes zoster in patients with RA.
SUBJECTS AND METHODS
This retrospective, population-based cohort study was conducted in Olmsted County, Minnesota, which is well suited for a longitudinal population-based analysis because the medical records of all residents seeking medical care by any medical provider dating back to over a half century are readily available. Gathering the medical records is made easier by the records linkage system of the Rochester Epidemiology Project, which allows ready access to the complete records from all health care providers and hospitals for the local population. The potential of this data system for use in population-based research has been well established .
The study included an inception cohort of all cases of RA diagnosed for the first time between January 1, 1980 and December 31, 2007 (n = 813) among Olmsted County residents ≥18 years of age assembled as previously described . The incidence date was defined as the earliest date in which the patient fulfilled at least 4 of the 7 American College of Rheumatology (ACR) 1987 classification criteria for RA . A comparison cohort of subjects without RA was assembled by randomly selecting for each patient with RA a corresponding Olmsted County resident during the same calendar year of the incident RA diagnosis who had a similar age and was the same sex. All patients in both cohorts were followed up longitudinally using their complete medical records until death, migration from Olmsted County, or December 31, 2008. All subjects (irrespective of residency status) were tracked nationally to ascertain vital status.
The original and complete medical records of all subjects were reviewed longitudinally by trained nurse abstractors who were supervised by the principal investigator (SEG). Information about demographics (e.g., age, sex, smoking status, and body mass index [BMI]), RA disease characteristics (e.g., erythrocyte sedimentation rate [ESR], rheumatoid nodules, joint erosions/destructive changes on radiographs, joint surgeries [arthroplasty and synovectomy], use of traditional DMARDs, biologic agents, and systemic corticosteroids), and severe extraarticular manifestations (defined according to the Malmö criteria and including pericarditis, pleuritis, Felty's syndrome, glomerulonephritis, vasculitis, peripheral neuropathy, scleritis, and episcleritis) was abstracted . The diagnostic index was also searched using International Classification of Diseases, Ninth Revision (ICD-9) codes for herpes zoster and its complications (ICD-9 codes 053.0–053.9 and 052.1). The medical records of all patients with codes for herpes zoster were reviewed to confirm the diagnosis, to ascertain the extent of the skin disease (single or multidermatomal/disseminated lesions), and to determine the outcome of the herpes zoster infection (extent of organ involvement/complications and hospitalizations). Postherpetic neuralgia was defined as documented zoster-associated pain persisting at least 90 days following the episode of herpes zoster. The diagnosis of herpes zoster was defined as the date of onset of the rash diagnosed as herpes zoster or, if the timing of the rash onset was not known, the date of the first diagnosis of herpes zoster either at a clinic visit or by a positive culture or biopsy. The information regarding vaccination against herpes zoster was gathered from the medical records.
For statistical analysis, descriptive statistics (means, percentages, etc.) were used to summarize the data. The cumulative incidence of herpes zoster adjusted for the competing risk of death was estimated . These methods are similar to the Kaplan-Meier method, with censoring of patients who were still alive at the final followup. However, the patients who died before developing herpes zoster were appropriately accounted for to avoid the overestimation of the rate of occurrence of herpes zoster, which can happen if such subjects are simply censored. Patients who were diagnosed with herpes zoster prior to the diagnosis of RA, or prior to the index date for subjects in the non-RA comparison cohort, were excluded from the analysis of cumulative incidence. Cumulative incidence comparisons between the cohorts were performed using the methods described by Gray . Poisson regression models were used to model the rates of herpes zoster over calendar time. Smoothing splines were used to allow for nonlinear time trends.
Cox proportional hazards models were used to compare the rate of the development of herpes zoster between patients with RA and the non-RA comparison cohort. In addition, Cox proportional hazards models were used to assess the association of risk factors with the development of herpes zoster among patients with RA. These models were assessed on an age time scale, stratified by sex and adjusted for calendar year of the first diagnosis of herpes zoster. The age time scale was used because age is strongly and perhaps nonlinearly associated with the incidence of herpes zoster, and using age as the time scale achieves the best adjustment for age without the need to specify whether age has a linear or nonlinear effect on the development of herpes zoster. The risk factors included RA disease characteristics and antirheumatic medications, as well as smoking status and BMI ≥30 kg/m2. For antirheumatic medications, current use and exposure at any time during the followup were examined using time-dependent covariates. Time-dependent covariates were used to model the risk factors that developed over time (i.e., RA duration, BMI ≥30 kg/m2, joint erosions, large joint swelling, joint surgery, rheumatoid nodules, and severe extra-articular manifestations), allowing patients to be modeled as unexposed to the risk factor during the followup time prior to the development of the risk factor and then changed to be modeled as exposed following the development of the risk factor. In the case of RA duration, which is a continuous time-dependent covariate, the RA duration of each subject was updated for each day of followup.
Kaplan-Meier methods were used to estimate mortality following herpes zoster occurrence. Cox proportional hazards models were also used to assess the differences between the RA and non-RA cohorts in mortality after herpes zoster, adjusting for age, sex, and calendar year of the first herpes zoster diagnosis.
The study population included 813 patients with RA and 813 subjects without RA. The mean ± SD age at RA diagnosis (at the index date for the non-RA cohort) was 55.9 ± 15.7 years and the median age was 55.3 years (range 18–94 years); 556 individuals (68%) in each cohort were women. There was a statistically significant difference in smoking status between the RA and non-RA cohorts, as shown in Table 1. The prevalence of herpes zoster (P = 0.85) and malignancy (P = 0.22) prior to the RA diagnosis/index date were similar in both cohorts.
Table 1. Characteristics of 813 patients with RA and 813 subjects without RAa
RA (n = 813)
Non-RA (n = 813)
aValues are the number (percentage) unless otherwise indicated. RA = rheumatoid arthritis.
Age at incidence/index, mean ± SD years
55.9 ± 15.7
55.9 ± 15.7
Length of followup, mean ± SD years
9.6 ± 6.9
10.9 ± 7.2
Smoking status at incidence/index
Body mass index at incidence/index, mean ± SD kg/m2
27.8 ± 6.2
27.8 ± 7.8
Herpes zoster prior to incidence/index
Any malignancy prior to incidence/index
Among the patients who did not have herpes zoster prior to the RA diagnosis/index date, 84 RA patients and 44 non-RA subjects developed herpes zoster during followup. The cumulative incidence rate of herpes zoster was higher in the RA patients compared to the non-RA subjects (8.9% [95% confidence interval (95% CI) 6.6–11.2] at 10 years in the RA patients versus 4.3% [95% CI 2.7–5.9] at 10 years in the non-RA subjects [P < 0.001]) (Figure 1). In other words, the patients with RA were more likely to develop herpes zoster during followup (hazard ratio [HR] 2.4 [95% CI 1.7–3.5] adjusted for age, sex, and calendar year). The rate of the development of herpes zoster was 12.1 cases per 1,000 person-years (95% CI 9.6–14.9) in patients with RA and 5.4 cases per 1,000 person-years (95% CI 3.9–7.2) in non-RA subjects.
The development of herpes zoster occurred more frequently in patients diagnosed with RA more recently (HR 1.06 per year [95% CI 1.02–1.10], P = 0.004). Likewise, the non-RA subjects showed a similar trend (HR 1.07 per year [95% CI 1.02–1.12], P = 0.009), reflecting an apparent increase in the incidence of herpes zoster in the general population in more recent years. Figure 2 shows the change in rates of development of herpes zoster over time for both cohorts. There was no evidence that the increase in herpes zoster was different in patients with RA than in the subjects without RA (P = 0.49 for the interaction between RA/non-RA and calendar year). Furthermore, there was no statistically significant evidence that age (P = 0.28) or sex (P = 0.23) influenced the development of herpes zoster differently in RA patients than in non-RA subjects. In addition, there was no apparent seasonal variation in the development of herpes zoster among either the RA or the non-RA cohort (data not shown), which agrees with the findings from the general population .
Risk factors for herpes zoster in RA
The examination of RA disease characteristics revealed that erosive disease, large joint swelling, previous joint surgery, and use of hydroxychloroquine and corticosteroids at any point during the followup period were significantly associated with the development of herpes zoster in patients with RA (Table 2). The association between the exposure to other antirheumatic medications, including biologic agents (HR 1.24 [95% CI 0.56–2.74]) with the development of herpes zoster did not reach statistical significance. However, the confidence intervals were wide, indicating limited statistical power, likely related to the low prevalence of use of biologic agents (17%). When the current use of antirheumatic medications was analyzed, only corticosteroids were found to be significantly associated with the occurrence of herpes zoster (HR 1.78 [95% CI 1.14–2.76]).
Table 2. Association between RA disease characteristics and herpes zoster in patients with RAa
Hazard ratio (95% CI)
aRA = rheumatoid arthritis; 95% CI = 95% confidence interval; DMARDs = disease-modifying antirheumatic drugs.
bPer 10 mm/hour increase.
dDefined as ≥6 tablets of aspirin per day (>1,950 mg/day) for ≥3 months.
Duration of RA (per 10-year increase)
Erythrocyte sedimentation rate at index, mean ± SD
Multidermatomal skin involvement occurred in 17 patients with RA (20%) and in 7 subjects without RA (16%; P = 0.69). Secondary infection of the skin and cellulitis occurred in 4 RA patients (5%) and 1 non-RA subject (2%; P = 0.66). Thirteen patients with RA (15%) and 7 non-RA subjects (16%) were diagnosed with postherpetic neuralgia (P = 0.97). Ramsay Hunt syndrome type II (geniculate ganglion involvement) occurred in 1 patient with RA; 1 patient with RA had disseminated herpes zoster. Hospitalization for herpes zoster was required for 11 patients with RA (13%) and 1 non-RA subject (2%; P = 0.057 by Fisher's exact test). Eight of 11 RA patients who were hospitalized had ≥1 complications of herpes zoster: 4 patients had post-herpetic neuralgias alone, 2 patients had both cellulitis and post-herpetic neuralgias, and 2 patients had eye complications. The one non-RA patient who was hospitalized for herpes zoster did not have any complications.
A second episode of herpes zoster occurred during followup in 3 patients with RA and 3 non-RA subjects (HR 0.17 [95% CI 0.02–1.60, P = 0.12] adjusted for age, sex, and calendar year of the first diagnosis of herpes zoster). During followup, 32 RA patients and 11 non-RA subjects who developed herpes zoster died, corresponding to a 10-year mortality rate of 53% (95% CI 39–68) among the RA patients and 26% (95% CI 11–41) among the non-RA patients with herpes zoster. Although not reaching statistical significance, patients with RA who developed herpes zoster had a somewhat higher risk of death than patients without RA (HR 2.1 [95% CI 0.98–4.4, P = 0.058] adjusted for age, sex, and calendar year of herpes zoster).
Some of the patients in this cohort had received a vaccine for herpes zoster since 2006, when the vaccine was introduced. We evaluated the vaccine in a subset of 477 patients with RA and 518 subjects without RA who were followed up between January 1, 2006 and December 31, 2010. A total of 170 RA patients (rate 6.6 [95% CI 5.6–7.8] per 100 person-years) and 230 non-RA subjects (rate 8.4 [95% CI 7.3–9.6] per 100 person-years) received herpes zoster vaccine during followup. The patients with RA were less likely to receive the herpes zoster vaccine compared to the subjects without RA (HR 0.77 [95% CI 0.62–0.95]). When these analyses were repeated in the subset of patients age ≥60 years, the rates of vaccination in RA (number vaccinated = 145, rate 11.6 [95% CI 9.8–13.6] per 100 person-years) remained lower than in the non-RA cohort (number vaccinated = 206, rate 13.6 [95% CI 11.8–15.6] per 100 person-years). The likelihood of herpes zoster vaccination in patients with RA was significantly lower than in the non-RA cohort (HR 0.71 [95% CI 0.58–0.89] adjusted for age and sex). There were 4 RA patients and 2 non-RA subjects with herpes zoster after administration of the vaccine (P = 1.0 by Fisher's exact test).
We examined associations between RA characteristics and the likelihood of receiving herpes zoster vaccine in RA patients age ≥60 years. The presence of erosions/destructive changes (HR 0.67 [95% CI 0.47–0.95]) and current treatment with methotrexate (HR 0.63 [95% CI 0.43–0.93]) but not other RA characteristics and medications were significantly associated with lower likelihood of being vaccinated. The associations of ESR at baseline (HR 0.93 [95% CI 0.85–1.02]) with vaccination approached statistical significance (P = 0.13).
The incidence of herpes zoster is increased in patients with RA compared to patients without RA and has risen in recent years. The increases in the incidence of herpes zoster among patients with RA mirror the increases seen in the general population, with the incidence of herpes zoster remaining consistently higher among individuals with RA than among those without RA. Several indicators of disease severity were found to be significantly associated with the occurrence of herpes zoster among patients with RA.
Data from prior studies have shown that herpes zoster infections occur in ∼1 million individuals annually in the US [1, 12]. Patients with RA are a particularly vulnerable population, as evidenced by the approximately 2-fold increase in risk of herpes zoster compared to the general population . Our findings of elevated rates of herpes zoster occurrences in RA of 12.1 cases per 1,000 person-years are consistent with the estimate from a retrospective cohort study of US veterans (9.96 cases per 1,000 patient-years) .
Potential explanations for the increase in the incidence of herpes zoster in patients with RA, particularly in recent years, include increasing and early use of immunosuppressive medications for the management of RA, aging of the population, and increasing burden of comorbidity and immunocompromised conditions [13-15]. Although these factors may account for the recent incidence trend, they do not fully explain why the incidence of herpes zoster increased in both the patients with RA and the general population over the duration of the study. This recent increase in the incidence of herpes zoster in the general population has been noted by other investigators as well [16, 17].
We explored what factors, disease or treatment related, if any, could account for the increase in the incidence of herpes zoster in RA. Among the RA disease characteristics, erosive disease and previous joint surgery stand out for their significant association with the development of herpes zoster, presumably because these characteristics are indicators of increased disease activity and disease burden. This is consistent with the findings of other authors reporting on the association of measures of disease activity and severity (i.e., Disease Activity Score and Health Assessment Questionnaire score) with the risk of herpes zoster in RA [4, 18].The reasons for the lack of an association of another marker of RA severity, namely rheumatoid factor positivity, are unclear, but this finding appears to be concordant with a previous study .
The existing evidence regarding the impact of cigarette smoking on herpes zoster is inconclusive. Some population-based studies reported a protective effect of current smoking on the occurrence of herpes zoster, while in other more recent studies smoking has been linked to an increased severity of herpes zoster (i.e., increased pain burden and risk of complications) [19, 20]. Current smoking was not associated with the occurrence of herpes zoster in our study.
The effect of antirheumatic medications, in particular DMARDs and biologic agents, on the risk of infections in patients with RA has been debated in the rheumatology community. A nested case–control study by Smitten et al  found the risk of herpes zoster to be elevated in patients with RA who were receiving biologic DMARDs (odds ratio [OR] 1.54) or traditional DMARDs (OR 1.37) compared with individuals not receiving DMARD treatment. Antonelli et al reported an increased risk of herpes zoster in RA patients treated with low-dose methotrexate compared with the general population . A prospective German biologic agent registry  showed an increased risk of herpes zoster in patients receiving tumor necrosis factor (TNF) inhibitors (i.e., adalimumab and infliximab) compared with those receiving traditional DMARDs. The data from the Consortium of Rheumatology Researchers of North America registry showed that varicella-zoster virus infection was the most frequent infection in patients who received methotrexate, TNF blockers, or other DMARDs and made up 44% of all cases of opportunistic infections .
Somewhat conversely, some studies did not find any apparent association of TNF antagonists with the risk of herpes zoster. Wolfe et al  studied 10,614 patients with RA and found that cyclophosphamide, azathioprine, prednisone, leflunomide, and some nonsteroidal antiinflammatory drugs were risk factors for herpes zoster, but that TNF antagonists and methotrexate were not. However, the use of self-administered patient questionnaires was a limitation in this study. Concordant results were obtained in a large population-based study . The authors noted no difference in herpes zoster risk for those receiving TNF antagonists compared to those taking traditional DMARDs.
In our study, the association of biologic agents with the risk of herpes zoster did not reach statistical significance, which may be in part due to the limited statistical power. Of traditional DMARDs, only hydroxychloroquine was associated with herpes zoster occurrence. This finding resonates with some recent reports of the effects of hydroxychloroquine in patients with HIV . A recent randomized double-blind placebo-controlled trial of hydroxychloroquine (400 mg/day for 48 weeks) versus placebo in patients with chronic asymptomatic HIV who were not taking antiretroviral therapy has demonstrated a detrimental effect of hydroxychloroquine on CD4 cell count and increased HIV viral replication. The reasons for this unfavorable effect of hydroxychloroquine are unclear and require further study. Taken together with our findings, these results suggest the need for reconsideration of the immunosuppressive effects of hydroxychloroquine. The only clear predisposing factor was the use of corticosteroids.
The use of corticosteroids was strongly associated with herpes zoster in our study, which is concordant with some previous reports [3, 13, 18]. Smitten et al  also demonstrated a significant association between corticosteroids and herpes zoster among patients with RA, with corticosteroid use being a surrogate for increased RA disease activity. Decreased cell-mediated immunity associated with exposure to glucocorticoids may also contribute to this association.
In addition to RA characteristics and medication use, the presence of comorbid conditions can also contribute to the increased risk of herpes zoster in RA. The association between comorbidities and herpes zoster occurrence in RA was not examined in this study; however, some previous reports suggested an association of some comorbidities, including malignancy, chronic lung disease, renal failure, and liver disease with the risk of herpes zoster in RA .
Since 2006, a live, attenuated zoster vaccine has been available for use in immunocompetent persons ages ≥60 years. In the face of manufacturer's recommendations contraindicating the use of this vaccine in immunosuppressed individuals, the Centers for Disease Control and Prevention Advisory Committee on Immunization Practices recommend that patients receiving prednisone <20 mg/day, short-term corticosteroids (<2 weeks), topical or intraarticular corticosteroids, low-dosage methotrexate (defined as <0.4 mg/kg/week), azathioprine (<3.0 mg/kg/day), or 6-mercaptopurine (<1.5 mg/kg/day) could receive the vaccine . However, these guidelines do not address vaccine use when these agents are combined. The current ACR guidelines for the use of DMARDs and biologic agents in RA do not recommend giving a varicella-zoster vaccine to patients receiving biologic agents [25, 26]. However, some recent findings have suggested that there was a lack of an increase in short-term risk of herpes zoster infection following administration of the vaccine in patients receiving biologic agents compared to those taking only nonbiologic DMARDs . Considering the number of contraindications for the use of a varicella-zoster vaccine, it is not surprising that patients with RA had reduced zoster vaccination rates in our cohort. Lower likelihood of vaccination in patients with more severe RA (i.e., those with erosive disease, and perhaps those with increased ESR) as well as in patients receiving methotrexate may be due to physicians' concerns about providing the vaccination to these patients. However, we do not have specific information about this. In our population, there were too few patients who had herpes zoster after receiving the vaccine to draw any meaningful conclusions regarding the effect of vaccination on the occurrence and/or severity of herpes zoster in the RA versus the non-RA cohort.
Because herpes zoster is caused by the reactivation of the latent virus, the immunocompetence of the patient is of particular importance regarding the risk of development of herpes zoster and its associated complications. It is thought that immunocompromised individuals are more likely to experience complications from herpes zoster , such as ocular and neurologic syndromes . We found no statistically significant differences in the occurrence of herpes zoster complications in patients with RA and in subjects without RA. However, the number of events was limited in these analyses suggesting that some differences could have been missed. Patients with RA who developed herpes zoster were at a somewhat higher risk of hospitalization and all-cause premature mortality than the patients without RA. This is consistent with findings from the general population where conditions associated with immunosuppression, including RA, have been found to account for up to 43.4% of hospitalizations and up to 21.6% of all deaths in patients with herpes zoster . Because of the small number of events it was not possible to explore the reasons for the trend toward increased likelihood of hospitalization and mortality in this study.
To our knowledge, this is the first population-based study to examine the differences in the incidence and severity of herpes zoster in patients with RA compared to the non-RA subjects. Strengths of our study include its longitudinal followup of a population-based cohort of newly diagnosed RA patients with medical record confirmation of all diagnoses. With the exception of a higher proportion of the working population employed in the health care industry and a corresponding higher education level, and the predominately white racial composition, the results of this study utilizing the population of Olmsted County, Minnesota are generalizable to the results from studies of other populations of interest . As is the case with any retrospective study, only persons who had a medical encounter during which herpes zoster could be identified were included. Occurrences of herpes zoster in persons not seeking medical care for this problem would not be included. Disease activity scores were not available for this cohort. The statistical power could be limited in some analyses, due to the limited number of events, particularly when assessing the rates of outcomes/complications of herpes zoster and associations between the use of biologic agents and occurrence of herpes zoster. These results should be interpreted with caution.
In conclusion, there is an apparent increase in the incidence of herpes zoster in patients with RA relative to the general population. The incidence of herpes zoster in RA has further increased in recent years; a similar increasing trend has been noted in subjects without RA. In the present study, there was no significant difference in the severity of herpes zoster in RA versus non-RA subjects. RA disease severity and the use of some antirheumatic medications (in particular, corticosteroids) were found to be associated with the occurrence of herpes zoster. Given the increasing trend of herpes zoster in patients with rheumatic diseases, a careful and thorough review of existing risk factors, risk modification through the use of a vaccine, and judicious use of immunosuppressive medications would hopefully prevent the recurrence and complications of herpes zoster.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Ms Crowson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Veetil, Myasoedova, Matteson, Gabriel, Crowson.
Acquisition of data. Veetil, Matteson, Gabriel, Crowson.
Analysis and interpretation of data. Veetil, Myasoedova, Matteson, Gabriel, Green, Crowson.
ROLE OF THE STUDY SPONSOR
Pfizer had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Pfizer.