Key Points
Question
What is the rate of serious infections requiring hospitalization and outpatient-treated infections among children with psoriasis treated with ustekinumab, etanercept, or methotrexate in clinical practice?
Findings
In this cohort study of 2338 children, the incidence of serious infection was 18.4 per 1000 person-years for ustekinumab users, 25.6 per 1000 person-years for etanercept users, and 14.9 per 1000 person-years for methotrexate users. Ustekinumab showed no increased rate of outpatient infections compared with etanercept and methotrexate, and may suggest a trend toward a decreased risk for ustekinumab.
Meaning
This cohort study suggests that ustekinumab was not associated with an increased risk of infections compared with methotrexate and etanercept among children with psoriasis.
Abstract
Importance
Psoriasis in children is increasingly treated with systemic medications, yet their risk of serious infection is not well characterized in clinical practice. Pediatric clinical trials for these medications were often small and placebo controlled.
Objective
To estimate the 6-month rate of infections among children with psoriasis who started treatment with ustekinumab, etanercept, or methotrexate.
Design, Setting, and Participants
This cohort study used insurance claims data from clinical practices across the US on children aged 17 years or younger with psoriasis who were receiving treatment with a topical medication for psoriasis and started new treatment with ustekinumab, etanercept, or methotrexate. The analysis was stratified by the time before pediatric labeling (2009-2015) and after pediatric approval (2016-2021). Patient follow-up started 1 day after initiating treatment and ended at 6 months.
Exposures
New treatment with ustekinumab, etanercept, and methotrexate.
Main Outcomes and Measures
During follow-up, the frequency of inpatient serious infections and outpatient infections requiring treatment was compared. Event rates and rate ratios were estimated after propensity score decile stratification.
Results
After exclusions, we identified 2338 patients (1368 girls [57.8%]) who initiated new treatment with a targeted immunomodulating agent. In all, 379 patients started treatment with ustekinumab, 779 patients started treatment with etanercept, and 1180 patients started treatment with methotrexate from 2009 through 2021. The propensity score–adjusted incidence rate of serious infection was 18.4 per 1000 person-years (3 events) for ustekinumab users, 25.6 per 1000 person-years (9 events) for etanercept users, and 14.9 per 1000 person-years (8 events) for methotrexate users. The adjusted rate of outpatient infections was 254.9 per 1000 person-years (39 events) for ustekinumab users, 435.7 per 1000 person-years (139 events) for etanercept users, and 433.6 per 1000 person-years (209 events) for methotrexate users. The adjusted rate ratio of outpatient infections was 0.58 (95% CI, 0.41-0.83) for ustekinumab vs etanercept, 0.66 (95% CI, 0.48-0.91) for ustekinumab vs methotrexate, and 0.95 (95% CI, 0.75-1.21) for etanercept vs methotrexate. Rate ratios were similar during the off-label use era and after pediatric labeling.
Conclusions and Relevance
Among children with psoriasis who started treatment with immunomodulating agents, serious infections were infrequent. This cohort study suggests that there was no increase in the risk of outpatient infections for children who started treatment with ustekinumab compared with etanercept or methotrexate.
This cohort study uses insurance claims data to estimate the 6-month rate of infections among children with psoriasis who started treatment with ustekinumab, etanercept, or methotrexate.
Introduction
Clinical trials have demonstrated high efficacy of new immunomodulatory agents in treating children with psoriasis; however, the risk of infections in clinical practice has not been fully characterized by comparing these medications against each other in pairwise comparisons.1,2 Psoriasis is a chronic inflammatory skin disease affecting about 1% of children in the US, with an incidence of about 33.2 per 100 000 children.3,4,5 Physicians often initiate treatment with systemic immunomodulatory agents for children with psoriasis that is refractory to topical treatments, has a higher body surface area of involvement, causes significant burden on quality of life, or involves body sites that are challenging to treat. Although no nonbiologic immunomodulatory agents are approved for use in children with psoriasis, methotrexate is recommended as an off-label treatment.6 Several biologic immunomodulatory agents have been approved in the US for the treatment of moderate to severe pediatric psoriasis, including etanercept in 2016 (for children aged 4-17 years), ustekinumab in 2017 (for children aged 12-17 years) and 2020 (for children aged 6-17 years), and ixekizumab (for children aged 6-17 years) and secukinumab (for children aged 6-17 years) in 2020.2
These biologic immunomodulatory agents are increasingly considered as first-line systemic agents given their efficacy in treating children with psoriasis and improving their quality of life.6,7 The approval of biologic immunomodulatory agents for children with psoriasis was based on small, mostly placebo-controlled clinical trials, often with fewer than 100 children and limited pairwise comparisons.8,9,10 Potentially serious adverse events, including infection requiring hospitalization or treatment, are difficult to observe in these small trials. The controlled research setting, required for conducting robust clinical trials, may not be generalizable to the broad population seen in clinical practice. Quantifying the risk of infection associated with these immunomodulatory agents for children treated in routine care can provide a better benefit-risk assessment, especially as these immunomodulatory agents are increasingly used as first-line systemic treatments.
This study used US claims data to evaluate the risk of infections, including serious infections and infections requiring treatment, among children with psoriasis who initiated additional treatment with ustekinumab, etanercept, or methotrexate.
Methods
Data Source
We used longitudinal, nationwide insurance claims data from 2 US databases (IBM MarketScan and Optum’s Clinformatics Data Mart) covering enrollees between December 31, 2002, and June 1, 2021. The databases contain administrative health claims for members of large commercial and Medicare Advantage health plans across the US, including dated information on plan enrollment, health care use, demographic characteristics, hospital discharge diagnoses, outpatient events, and pharmacy dispensing. The Brigham and Women’s Hospital’s institutional review board approved this study. Consent was waived because all patient data were deidentified. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Patients
Imitating a head-to-head clinical trial, we implemented a new-user, active comparator cohort study (eFigure in Supplement 1). To compare the 3 drugs, we created 3 direct pairwise comparison cohorts for analysis. We identified patients younger than 18 years who initiated treatment with a systemic immunomodulatory agent of interest between January 1, 2016, and June 30, 2021. Within the 180 days before starting a treatment of interest, all children were required to have a recorded diagnosis of psoriasis (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] code L40.x [psoriasis], except code L40.5 [arthropathic psoriasis]) and at least 1 filled prescription for a topical corticosteroid or topical vitamin D derivative.11
We compared groups of pediatric patients with psoriasis initially treated with topical agents by types of treatment escalation: initiation of ustekinumab, etanercept, or methotrexate. Because secukinumab and ixekizumab were approved for pediatric use in 2020, they were not included in this study. We created 3 pairwise comparisons for analysis: cohort A compared new users of ustekinumab vs etanercept (reference drug), cohort B compared ustekinumab vs methotrexate (reference drug), and cohort C compared etanercept vs methotrexate (reference drug). The cohort entry date was the first use of any of these treatments.12 First use was defined as no prior use of the exposure drug and no prior use of the reference drug, within each pairwise cohort, during the 180 days before treatment initiation.
We excluded patients who had less than 180 days of continuous enrollment before cohort entry,13 or had any preexisting conditions that could increase the risk of serious infections, including congenital or acquired immunodeficiency, chronic infections, any malignant neoplasm, chemotherapy, organ transplantation, and prior use of other systemic immunomodulating agents as well as patients who started both the reference and exposure medications on the same day. To help ensure the medication of interest was being prescribed for treatment of psoriasis and not for another comorbid condition, we further excluded patients with psoriasis who had comorbid rheumatoid arthritis, juvenile idiopathic arthritis, ankylosing spondylitis, dermatomyositis, and inflammatory bowel disease (eFigure in Supplement 1).14
Outcomes
All new infections were recorded during follow-up, which started the day after cohort entry and ended at 180 days after cohort entry. Our primary outcome was serious infection, defined as any events of serious bacterial, viral, or opportunistic infections (eTable 1 in Supplement 1) that led to an emergency department visit or hospital admissions.15,16,17,18 The codes have been previously validated in International Classification of Diseases, Ninth Revision (ICD-9) and have shown positive predictive values greater than 80%.15,16 We used the Centers for Medicare & Medicaid Services–recommended ICD-9 to ICD-10 conversion rules to identify ICD-10 codes for the outcome, as no US-based ICD-10 code validation study had been conducted at the time of the analysis. We also cross-checked our ICD-10 codes against 2 European-based validation studies, which showed positive predictive values greater than 90%, and earlier US database studies that used these codes.17,18,19,20
Our secondary end point was outpatient infection, defined as any outpatient physician visit with a diagnosis of infection plus a filled prescription for its treatment within 3 days after the diagnosis. According to this criterion, we counted all events of bacterial infection that required use of an antibiotic, mycobacterial infections that required use of a mycobacterial-specific medication, and herpes simplex or zoster infection that required use of an antiviral (eTable 2 in Supplement 1). The infection date was the date on which all criteria were fulfilled. We further stratified by each outpatient infection separately. Patients were censored at the date of qualifying infection, end of follow-up (180 days), death, disenrollment, or end of the data stream, whichever came first (eFigure in Supplement 1).14
Patient Characteristics
All patient characteristics were assessed during the 180 days before cohort entry, including the day of cohort entry. The following patient characteristics were considered: age at cohort entry; sex; race and ethnicity (race data were available only in the Optum database, which uses proprietary algorithms to define race; race is a derived ethnicity; the member’s ethnicity is derived by using the member’s name and geography, and once the ethnicity is determined, the member is mapped to 1 of 4 race categories [Asian, Black, Hispanic, or White]); calendar year of cohort entry; history of past infections that required an office visit or hospitalization; prior use of antibiotic, antifungal, or antiviral medication; history of recent surgery; number of prior nonbiologic or biologic immunomodulatory agents; systemic glucocorticoid use; cumulative dose of systemic glucocorticoids in prednisone milligram equivalences; and health care use (ie, numbers of outpatient physician visits, unique medications, and dermatologist visits). The full list of patient characteristics is detailed in Table 1. We further controlled for comorbidities using a validated pediatric comorbidity score (eTable 3 in Supplement 1).21
Table 1. Patient Characteristics Within the 180 Days Before Starting Treatment.
| Characteristic | Patients, No. (%) | |||||
|---|---|---|---|---|---|---|
| Ustekinumab vs etanercept | Ustekinumab vs methotrexate | Etanercept vs methotrexate | ||||
| Ustekinumab | Etanercept | Ustekinumaba | Methotrexate | Etanercept | Methotrexate | |
| No. of patients | 394 | 786 | 406 | 1189 | 586 | 1139 |
| Demographic characteristic | ||||||
| Sex | ||||||
| Male | 162 (41.1) | 312 (39.7) | 164 (40.4) | 527 (44.3) | 225 (38.4) | 510 (44.8) |
| Female | 232 (58.9) | 474 (60.3) | 242 (59.6) | 662 (55.7) | 361 (61.6) | 629 (55.2) |
| Age categories, y | ||||||
| ≥5 | 1 (0.3) | 14 (1.8) | 1 (0.2) | 37 (3.1) | 11 (1.9) | 36 (3.2) |
| 6-11 | 16 (4.1) | 187 (23.8) | 19 (4.7) | 345 (29.0) | 144 (24.6) | 335 (29.4) |
| 12-17 | 377 (95.7) | 585 (74.4) | 386 (95.1) | 807 (67.9) | 431 (73.5) | 768 (67.4) |
| Raceb | ||||||
| Asian | 10 (6.7) | 11 (5.2) | 11 (7.3) | 15 (5.2) | 8 (4.8) | 13 (4.9) |
| Black | 15 (10.1) | 19 (8.9) | 13 (8.7) | 21 (7.3) | 15 (8.9) | 20 (7.5) |
| Hispanic | 13 (8.7) | 25 (11.7) | 14 (9.3) | 25 (8.7) | 21 (12.5) | 21 (7.8) |
| White | 91 (61.1) | 144 (67.6) | 93 (62.0) | 206 (71.5) | 112 (66.7) | 193 (72.0) |
| Health care use | ||||||
| Office visits, mean (SD), No. | 5.9 (6.7) | 6.8 (6.6) | 5.4 (5.8) | 6.2 (6.6) | 6.8 (6.9) | 6.2 (6.7) |
| Unique medications, mean (SD), No. | ||||||
| Topical | 2.5 (2.5) | 2.9 (2.8) | 2.3 (2.4) | 3.0 (2.9) | 2.9 (2.6) | 3.0 (3.0) |
| Nontopical | 3.9 (4.5) | 4.3 (4.8) | 3.7 (3.9) | 3.7 (4.6) | 3.3 (4.1) | 3.6 (4.6) |
| Hospitalization | 11 (2.8) | 20 (2.5) | 8 (2.0) | 34 (4.3) | 13 (2.2) | 32 (2.8) |
| Dermatologist office visits, mean (SD), No. | 2.2 (3.4) | 2.9 (4.7) | 2.1 (3.3) | 2.4 (4.3) | 2.9 (4.8) | 2.4 (4.4) |
| Infection risk factors | ||||||
| History of surgery | 1 (0.3) | 1 (0.1) | 1 (0.3) | 2 (0.3) | 0 | 2 (0.2) |
| Physician-administered intravenous antifungal, antibiotic, or antiviral medication | 6 (1.5) | 22 (2.8) | 5 (1.3) | 31 (3.9) | 15 (2.6) | 28 (2.5) |
| Use of antibiotics, antivirals, antifungals, or tuberculosis medication | 145 (36.8) | 319 (40.6) | 158 (40.1) | 546 (69.5) | 233 (39.8) | 522 (45.8) |
| Prior infection | ||||||
| Bacterial or viral (inpatient or outpatient) | 62 (15.7) | 126 (16.0) | 61 (15.5) | 232 (29.5) | 93 (15.9) | 220 (19.3) |
| Opportunistic (inpatient or outpatient) | 0 | 5 (0.6) | 0 | 3 (0.4) | 4 (0.7) | 3 (0.3) |
| Type 1 diabetes | 1 (0.3) | 8 (1.0) | 3 (0.8) | 4 (0.5) | 7 (1.2) | 3 (0.3) |
| Sickle-cell disease | 0 | 0 | 0 | 0 | 0 | 0 |
| Pediatric comorbidity score, mean (SD)c | 1.1 (2.2) | 0.9 (1.7) | 1.1 (2.1) | 1.0 (1.8) | 0.9 (1.7) | 0.9 (1.8) |
| Comorbid conditions | ||||||
| Acne | 59 (15.0) | 70 (8.9) | 59 (15.0) | 93 (11.8) | 52 (8.9) | 85 (7.5) |
| Hidradenitis suppurativa | 2 (0.5) | 1 (0.1) | 2 (0.5) | 1 (0.1) | 0 | 1 (0.1) |
| Psoriatic arthritis | 18 (4.6) | 124 (15.8) | 18 (4.6) | 132 (16.8) | 69 (11.8) | 119 (10.4) |
| Uveitis | 0 | 0 | 0 | 9 (1.1) | 0 | 9 (0.8) |
| Systemic corticosteroid use | ||||||
| Prior use | 39 (9.9) | 99 (12.6) | 38 (9.6) | 216 (27.5) | 71 (12.1) | 213 (18.7) |
| Cumulative sum of daily dose of oral corticosteroids (oral prednisone milligram equivalents), mean (SD) | 15.6 (85.5) | 13.5 (88.4) | 17.6 (99.9) | 24.5 (159.5) | 13.8 (86.1) | 25.1 (162.3) |
| Prior use of immunomodulatory treatmentsd | ||||||
| Recent nonbiologic use (30 d prior to 1 d prior to starting the exposure agent) | 23 (5.8) | 94 (12.0) | 5 (1.2) | 23 (1.9) | 9 (1.5) | 21 (1.8) |
| Prior nonbiologic use (60 d prior to 31 d prior to starting the exposure agent) | 30 (7.6) | 107 (13.6) | 7 (1.7) | 19 (1.6) | 11 (1.9) | 17 (1.5) |
| Recent biologic use (30 d prior to 1 d prior to starting the exposure agent) | 16 (4.1) | 13 (1.7) | 23 (5.7) | 43 (3.6) | 7 (1.2) | 21 (1.8) |
| Prior biologic use (60 d prior to 31 d prior to starting the exposure agent) | 35 (8.9) | 12 (1.5) | 57 (14.0) | 41 (3.4) | 9 (1.5) | 18 (1.6) |
| No. of prior systemic immunomodulators, nonbiologice | ||||||
| 0 | 337 (85.5) | 582 (74.0) | 391 (96.3) | 1129 (95.0) | 564 (96.2) | 1083 (95.1) |
| 1 | 56 (14.2) | 197 (25.1) | 15 (3.7) | 60 (5.0) | 21 (3.6) | 56 (4.9) |
| ≥2 | 1 (0.3) | 7 (0.9) | 0 | 0 | 1 (0.2) | 0 |
| No. of prior systemic immunomodulators, biologic or targeted syntheticf | ||||||
| 0 | 335 (85.0) | 749 (95.3) | 304 (74.9) | 1115 (93.8) | 557 (95.1) | 1105 (97.0) |
| 1 | 56 (14.2) | 37 (4.7) | 97 (23.9) | 69 (5.8) | 29 (4.9) | 32 (2.8) |
| ≥2 | 3 (0.8) | 0 | 5 (1.2) | 5 (0.4) | 0 | 2 (0.2) |
Included ustekinumab users who contributed to 1 or more pairwise comparisons.
Race data were available only in the Optum database, which uses proprietary algorithms to define race. Race is a derived ethnicity. The member’s ethnicity is derived by using the member’s name and geography. Once the ethnicity is determined, the member is mapped to 1 of 4 race categories (Asian, Black, Hispanic, or White).
Scores range from 0 to 5, with 0 being the lowest risk of hospitalization and 5 being the highest risk of hospitalization.
Patients were new users (180-day washout) of both the reference agent and the exposure agent within each pairwise comparison. However, patients were allowed to use other systemic agents.
Nonbiologic agents: methotrexate, cyclosporine, mycophenolate, and leflunomide.
Biologic agents: apremilast, risankizumab, tildrakizumab, guselkumab, ustekinumab, brodalumab, golimumab, infliximab, certolizumab, adalimumab, etanercept, ixekizumab, and secukinumab.
Statistical Analysis
We computed the 6-month event rates carrying the exposure status at cohort entry forward for the 180 days of follow-up. We analyzed 3 pairwise comparisons: ustekinumab vs etanercept (cohort A), ustekinumab vs methotrexate (cohort B), and etanercept vs methotrexate (cohort C).
We used propensity scores to achieve balance across covariates for each of the pairwise comparisons. All confounders and patient characteristics listed in the preceding subsection were entered as independent variables without further variable selection.22 Patients’ propensity score values were calculated using the estimated values from the resulting logistic regression model.23 For our primary analysis, we used a propensity score decile–stratified analysis with trimming as recommended for studies with very few events.24 We conducted propensity score decile stratification with symmetric trimming of the extremes of the propensity score distribution by 2.5%.25 We estimated a propensity score, used propensity score deciles to construct 10 propensity score strata, and computed a weighted mean of event rates, including events and person-time, as well as 6-month risks across the 10 propensity score strata.26 Event rate ratios (RRs) and rate differences were computed from those adjusted metrics. To demonstrate balance, we present 1:1 propensity score–matched results (eTable 5 in Supplement 1).27,28 We stratified our pediatric analysis by off-label users (2009-2015) and postapproval users (2016-2021), then pooled the data for all years (2009-2021). All analyses were implemented separately in each database, then the adjusted number of events and person-time were pooled across years and across databases to compute pooled effect estimates. All analyses were conducted using the Aetion Evidence Platform, version 4.40 (Aetion Inc), including R, version 3.4.2 (R Group for Statistical Computing), which has been validated in multiple studies.29,30,31,32
Results
After exclusions, we identified 2338 pediatric patients with psoriasis who were receiving treatment with a topical medication for psoriasis and initiated new treatment with a targeted immunomodulating agent between 2009 and 2021. Of these, 1001 (42.3%) were boys and 1368 (57.8%) were girls. For the 3 pairwise comparisons, we identified cohort A with 394 ustekinumab initiators vs 786 etanercept initiators, cohort B with 406 ustekinumab initiators vs 1189 methotrexate initiators, and cohort C with 586 etanercept initiators vs 1139 methotrexate initiators (eTable 4 in Supplement 1). After trimming, cohort A had 379 ustekinumab initiators vs 779 etanercept initiators, cohort B had 394 ustekinumab initiators vs 1180 methotrexate initiators, and cohort C had 571 etanercept initiators vs 1126 methotrexate initiators (Table 2). Our cohort included 52 children (2.2%) aged 5 years or younger, 548 (23.1%) aged 6 to 11 years, and 1769 (74.7%) aged 12 to 17 years (Table 1). After propensity score matching, the distribution of all patient characteristics between individuals in each pairwise comparison was balanced across all 3 cohorts (eTable 5 in Supplement 1).
Table 2. Pooled Incidence Rate and RR of Infections (2009-2021), After Propensity Score Decile Stratification.
| Outcome | Ustekinumab vs etanercept | Ustekinumab vs methotrexate | Etanercept vs methotrexate | |||
|---|---|---|---|---|---|---|
| Ustekinumaba | Etanercept | Ustekinumaba | Methotrexate | Etanercept | Methotrexate | |
| Serious infectionb | ||||||
| No. of patients | 379 | 779 | 394 | 1180 | 571 | 1126 |
| No. of person-years | 163 | 352 | 171 | 536 | 257 | 510 |
| No. of events | 3 | 9 | 3 | 8 | 8 | 8 |
| Rate per 1000 person-years | 18.4 | 25.6 | 17.5 | 14.9 | 31.1 | 15.7 |
| RR (95% CI) | 0.72 (0.19 to 2.66) | 1 [Reference] | 1.18 (0.31 to 4.43) | 1 [Reference] | 1.98 (0.74 to 5.29) | 1 [Reference] |
| Rate difference per 1000 person-years (95% CI) | −7.2 (−33.9 to 19.5) | 1 [Reference] | 2.6 (−19.8 to 25.0) | 1 [Reference] | 15.4 (−8.7 to 39.6) | 1 [Reference] |
| Outpatient infectionc | ||||||
| No. of patients | 379 | 779 | 394 | 1180 | 571 | 1126 |
| No. of person-years | 153 | 319 | 161 | 482 | 236 | 459 |
| No. of events | 39 | 139 | 46 | 209 | 97 | 198 |
| Rate per 1000 person-years | 254.9 | 435.7 | 285.7 | 433.6 | 411.0 | 431.4 |
| RR (95% CI) | 0.58 (0.41 to 0.83) | 1 [Reference] | 0.66 (0.48 to 0.91) | 1 [Reference] | 0.95 (0.75 to 1.21) | 1 [Reference] |
| Rate difference per 1000 person-years (95% CI) | −180.8 (−288.8 to −72.9) | 1 [Reference] | −147.9 (−249.3 to −46.5) | 1 [Reference] | −20.4 (−121.9 to 81.1) | 1 [Reference] |
Abbreviation: RR, rate ratio.
Included ustekinumab users who contributed to 1 or more pairwise comparisons.
Defined as infection requiring an emergency department visit or hospitalization.
Defined as an outpatient physician visit for a bacterial or viral infection with a filled prescription for an antibiotic or antiviral agent within 3 days after the diagnosis.
Serious Infection
During the first 6 months of treatment, the propensity score decile–adjusted risk of serious infection was 0.8% for ustekinumab (3 events among 379 users), 1.2% for etanercept (9 events among 779 users), and 0.7% for methotrexate (8 events among 1180 users). The adjusted rate of serious infection was 18.4 per 1000 person-years for ustekinumab, 25.6 per 1000 person-years for etanercept, and 14.9 per 1000 person-years for methotrexate (Table 2). Rate ratio estimates after propensity score decile adjustment varied by contrast: 0.72 (95% CI, 0.19-2.66) for ustekinumab vs etanercept, 1.18 (95% CI, 0.31-4.43) for ustekinumab vs methotrexate, and 1.98 (95% CI, 0.74-5.29) for etanercept vs methotrexate. Estimates remained similar with or without trimming; 1 serious infection event was trimmed in the etanercept group of the etanercept vs methotrexate comparison (eTable 6 in Supplement 1).
Outpatient Infection
The 6-month risk of outpatient infection requiring treatment was 10.3% for ustekinumab (39 events among 379 users), 17.8% for etanercept (139 events among 779 users), and 17.7% for methotrexate (209 events among 1180 users). The adjusted rate of outpatient infection was 254.9 per 1000 person-years for ustekinumab, 435.7 per 1000 person-years for etanercept, and 433.6 per 1000 person-years for methotrexate (Table 2). The adjusted RR of outpatient infection was consistently reduced for ustekinumab: 0.58 (95% CI, 0.41-0.83) for ustekinumab vs etanercept, 0.66 (95% CI, 0.48-0.91) for ustekinumab vs methotrexate, and 0.95 (95% CI, 0.75-1.21) for etanercept vs methotrexate.
Across all comparisons, there were no events of outpatient mycobacterial infection or bone and joint infection (Table 3). During the first 6 months, there was 1 event of herpes simplex or zoster virus requiring antiviral treatment in the etanercept group. Overall, the most frequently occurring outpatient infections were skin and soft tissue infections, sinusitis, streptococcal pharyngitis, and acute otitis media. Skin infections and acute otitis media were most frequent with ustekinumab, while sinusitis was most frequent with etanercept and methotrexate.
Table 3. Pooled IRs of Individual Outpatient Infections Across 2009-2021, After Propensity Score Decile Stratification.
| Outcome | Ustekinumab vs etanercept | Ustekinumab vs methotrexate | Etanercept vs methotrexate | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. of events (IR per 1000 person-years) | RR (95% CI) | No. of events (IR per 1000 person-years) | RR (95% CI) | No. of events (IR per 1000 person-years) | RR (95% CI) | ||||
| Ustekinumab (n = 379) | Etanercept (n = 779) | Ustekinumab (n = 394) | Methotrexate (n = 1180) | Etanercept (n = 671) | Methotrexate (n = 1321) | ||||
| All outpatient infectionsa | 39 (254.9) | 139 (435.7) | 0.58 (0.48-0.71) | 46 (285.7) | 209 (433.6) | 0.66 (0.55-0.79) | 104 (372.8) | 222 (410.4) | 0.91 (0.79-1.05) |
| Cellulitis | 10 (62.5) | 24 (69.2) | 0.90 (0.75-1.09) | 12 (71.0) | 42 (79.9) | 0.89 (0.75-1.06) | 12 (39.9) | 47 (80.2) | 0.50 (0.43-0.57) |
| Sinusitis, acute | 7 (43.2) | 53 (155.4) | 0.28 (0.23-0.34) | 12 (71.0) | 68 (130.5) | 0.54 (0.46-0.65) | 41 (139.5) | 74 (128.0) | 1.09 (0.95-1.25) |
| Streptococcal pharyngitis and tonsillitis | 5 (30.9) | 24 (68.8) | 0.45 (0.37-0.54) | 5 (29.2) | 34 (64.4) | 0.45 (0.38-0.54) | 22 (73.3) | 42 (71.4) | 1.03 (0.89-1.18) |
| Acute otitis media | 12 (75.0) | 23 (65.7) | 1.14 (0.95-1.38) | 14 (83.3) | 41 (77.8) | 1.07 (0.90-1.27) | 18 (60.0) | 51 (87.2) | 0.69 (0.60-0.79) |
| Acute otitis externa | 1 (6.1) | 6 (17.0) | 0.36 (0.30-0.43) | 2 (11.7) | 20 (37.5) | 0.31 (0.26-0.37) | 4 (13.2) | 21 (35.4) | 0.37 (0.33-0.43) |
| Pneumonia | 1 (6.1) | 2 (5.6) | 1.09 (0.90-1.31) | 1 (5.9) | 7 (482.3) | 0.45 (0.38-0.53) | 1 (3.3) | 9 (15.1) | 0.22 (0.19-0.25) |
| UTI | 2 (12.3) | 2 (34.1) | 0.36 (0.30-0.43) | 2 (11.7) | 11 (20.5) | 0.57 (0.48-0.68) | 13 (43.2) | 11 (18.5) | 2.34 (2.04-2.69) |
| Bone and joint infection | 0 | 0 | NA | 0 | 0 | NA | 0 | 0 | NA |
| Herpes simplex or zoster virus | 1 (6.1) | 1 (2.8) | 2.18 (1.81-2.62) | 1 (5.9) | 3 (5.6) | 1.05 (0.88-1.24) | 1 (3.3) | 4 (6.7) | 0.49 (0.43-0.57) |
| TB or atypical mycobacterial infection | 0 | 0 | NA | 0 | 0 | NA | 0 | 0 | NA |
Abbreviations: IRs, incidence rates; NA, not applicable; RR, rate ratio; TB, tuberculosis; UTI, urinary tract infection.
Outpatient bacterial infection plus a filled prescription for antibiotics within 3 days after diagnosis. For herpetic infections, we required a filled prescription for an antiviral medication within 3 days after diagnosis.
Off-label and Postapproval Use
During the years before pediatric labeling, when the medications were prescribed off-label (2009-2015), there were fewer ustekinumab users compared with etanercept users (50 vs 443); in the postapproval years (2016-2021), the use of ustekinumab and etanercept was similar (329 vs 336). Although methotrexate is not approved for pediatric psoriasis, it remained the most used agent across all years. Given the small numbers of serious infections, it was difficult to compare serious infection event rates in the off-label years vs the postapproval years (eTable 7 in Supplement 1).
The RRs for outpatient infections consistently showed no increase in risk among ustekinumab users during the off-label and postapproval years: for ustekinumab vs etanercept, the RR was 0.83 (95% CI, 0.38-1.80) in the off-label period and 0.53 (95% CI, 0.34-0.81) after approval; for ustekinumab vs methotrexate, the RR was 0.84 (95% CI, 0.41-1.72) in the off-label period and 0.62 (95% CI, 0.42-0.92) after approval; and for etanercept vs methotrexate, the RR was 0.97 (95% CI, 0.70-1.33) in the off-label period and 0.93 (95% CI, 0.64-1.36) after approval (Table 4).
Table 4. Pooled Incidence and RR of Outpatient Infections During Off-label Years (2009-2015) Compared With Postapproval Years (2016-2021), After Propensity Score Decile Stratification.
| Outcome | Ustekinumab vs etanercept | Ustekinumab vs methotrexate | Etanercept vs methotrexate | |||
|---|---|---|---|---|---|---|
| Ustekinumaba | Etanercept | Ustekinumaba | Methotrexate | Etanercept | Methotrexate | |
| Off-label years (2009-2015): outpatient infectionb | ||||||
| No. of patients | 50 | 443 | 53 | 724 | 313 | 685 |
| No. of person-years | 20 | 185 | 22 | 297 | 131 | 281 |
| No. of events | 7 | 78 | 8 | 128 | 54 | 120 |
| Rate per 1000 person-years | 350.0 | 421.6 | 363.6 | 431.0 | 412.2 | 427.1 |
| RR (95% CI) | 0.83 (0.38-1.80) | 1 [Reference] | 0.84 (0.41-1.72) | 1 [Reference] | 0.97 (0.70-1.33) | 1 [Reference] |
| Postapproval years (2016-2021): outpatient infection | ||||||
| No. of patients | 329 | 336 | 341 | 456 | 358 | 636 |
| No. of person-years | 133 | 134 | 139 | 185 | 148 | 260 |
| No. of events | 32 | 61 | 38 | 81 | 50 | 102 |
| Rate per 1000 person-years | 240.6 | 455.2 | 273.4 | 437.8 | 337.8 | 392.3 |
| RR (95% CI) | 0.53 (0.34-0.81) | 1 [Reference] | 0.62 (0.42-0.92) | 1 [Reference] | 0.93 (0.64-1.36) | 1 [Reference] |
Abbreviation: RR, rate ratio.
Included ustekinumab users who contributed to 1 or more pairwise comparisons.
Outpatient infection was defined as an outpatient physician visit for a bacterial or viral infection with a filled prescription for an antibiotic or antiviral agent within 3 days after the diagnosis. The years 2009 to 2015 correspond to off-label use of these medications in pediatric patients. The years 2016 to 2021 correspond to the years after approval of biologic agents etanercept and ustekinumab for pediatric patients.
Discussion
Within the first 6 months of treatment, we observed that serious infections, requiring emergency department care or hospitalization, were infrequent across all treatment groups (incidence rates, 14.9 to 25.6 per 1000 person-years). Outpatient infections requiring treatment were more frequent, with an incidence of 254.9 to 435.7 per 1000 person-years. Although there was a numeric trend toward a decreased rate of infection among patients initiating treatment with ustekinumab, there was no meaningful difference in risk of infection between systemic treatment with ustekinumab, etanercept, or methotrexate. The evidence of no difference in risk of infection for ustekinumab was observed during the time when it was used off-label, before its approval for use in children, and equally so after it received pediatric approval. This finding confirms the usefulness of pediatric pharmacoepidemiology studies in the absence of adequately powered randomized clinical trials for formal pediatric labeling.
For pediatric patients with psoriasis, there are no head-to-head randomized clinical trials for ustekinumab vs etanercept or vs methotrexate, to our knowledge; however, the rates of serious infection are similar to those observed in placebo-controlled trials for these agents. A pediatric clinical trial investigating etanercept vs placebo found 4 events of severe infection among 210 etanercept-treated children (16.5 person-years), resulting in an incidence rate of 2.4 per 100 person-years.8 This finding is consistent with our study findings, in which the incidence rate of serious infection among patients receiving etanercept was 2.6 per 100 person-years (25.6 per 1000 person-years) based on 9 events in 779 patients (352 person-years). In randomized clinical trials for ustekinumab vs placebo, the risk of serious infection at week 16 was 1.8% (2 events among 110 ustekinumab initiators) for children aged 12 to 17 years and 2.3% (1 event among 44 ustekinumab initiators) for those aged 6 to 11 years.9,10 These rates are somewhat higher than in our study, where the 6-month risk of serious infection in patients initiating treatment with ustekinumab was 0.8% (3 events among 379 ustekinumab initiators). A randomized clinical trial of pediatric patients with psoriasis compared methotrexate vs adalimumab and found no serious infection events among the 37 children who initiated treatment with methotrexate.33 A registry study analyzed the data of 25 patients initiating treatment with methotrexate for psoriasis and reported 1 serious infection event requiring hospitalization.34 This finding is consistent with the low rate of serious infection observed in our study, where the 6-month risk of serious infection among patients taking methotrexate patients was 0.7% (8 events among 1180 methotrexate initiators).
For outpatient infections, it is difficult to compare our results with those in randomized clinical trials, as most randomized clinical trial definitions for outpatient infection included viral infections (such as nasopharyngitis) that did not require treatment and counted multiple events per patient. Compared with other database studies conducted among adults with psoriasis, the rate of serious infection observed in our study was similar to the rate of infection among adults. A French database study19 recently compared the risk of serious infection in adults with psoriasis initiating treatment with biologic agents. The rate of infection among adults with psoriasis initiating treatment with ustekinumab was 16.3 per 1000 person-years and among those initiating treatment with etanercept was 24.8 per 1000 person-years, which was similar to our findings in children (ustekinumab, 18.4 per 1000 person-years, and etanercept, 25.6 per 1000 person-years).19
Strengths and Limitations
Our study had several strengths. First, to our knowledge, this is the largest US pairwise comparison study of children with psoriasis receiving treatment with ustekinumab, etanercept, or methotrexate. Earlier large-scale studies compared classes of immunomodulators used in pediatric psoriasis but did not stratify by individual agents.35 Second, in this nationwide cohort study, we used claims data to complement randomized clinical trials by comparing infection risk in the clinical practice setting, which can enhance generalizability.36 Longitudinal claims data allow us to follow a patient’s timeline across multiple health systems and care settings in the US, including information from specialist visits, outpatient or emergency department visits and diagnoses, inpatient hospitalizations and discharge diagnoses, filled prescriptions, and procedures. Third, while there are known challenges to using claims data, we used best practices in pharmacoepidemiologic study design, including implementing the new-user active comparator cohort design in the absence of baseline randomization, to mitigate these challenges.15 Randomization in clinical trials is important to ensure that groups are balanced, with similar patients in both treatment groups. A nonrandomized setting is prone to confounding bias when comparing ongoing users against nonusers; for example, nonusers likely have mild psoriasis that does not require treatment with a systemic immunomodulator. As such, in a nonrandomized setting, we focus on only a new-user design to limit confounding and mimic a clinical trial design. Fourth, we used pairwise comparisons to achieve robust confounder adjustment and propensity score decile stratification as the primary mode of adjustment, given the small number of events. The additional trimming makes the analysis less susceptible to residual confounding among patients with extreme propensity scores when compared with an inverse probability of treatment weighting approach.37,38,39 This method excludes patients who have a strong propensity to only receive one treatment and not the other.
Some limitations remain. By stratifying into patients receiving specific immunomodulating agents, the cohort sizes were much smaller. Although we still had 400 to 600 pediatric patients per individual treatment group, we had very few serious infection events. The small event count in this population suggests that serious infection is infrequent among children treated with immunomodulators. Given the small number of events, it is important to be aware of overfitting the model, which can lead to overestimation bias away from the null. This overestimation bias is relevant to the serious infection RR estimates, which were based on sometimes fewer than 10 events.40 We used claims data that were collected from commercial insurance, which does not cover Medicaid patients. A small percentage of patients were younger than 5 years. In addition, small overlap between the data sources would not change the point estimate but may result in minimal widening of the 95% CIs that we did not adjust for. Last, the extent of psoriatic disease is difficult to fully capture in claims data, which may lead to residual confounding if psoriasis severity was a strong risk factor for the infections under study and if ustekinumab was selectively used in children with less severe disease. This study population captures patients who were using topical agents in the past 6 months and then started treatment with a systemic immunomodulator (ustekinumab, etanercept, or methotrexate). These characteristics may suggest that the population is composed of more patients with moderate to severe disease. Claims data also could not adjust for environmental and behavioral risk factors of serious infection.
Conclusions
This cohort study found that, among pediatric patients with psoriasis who started treatment with ustekinumab, etanercept, or methotrexate, serious infections were infrequent, with a 6-month incidence rate of 14.9 to 25.6 per 1000 person-years. In these data, we found no indication of an increased rate of serious infection among ustekinumab users. We observed a 42% reduced risk of outpatient infections in users of ustekinumab compared with etanercept and methotrexate.
eFigure. Study Design Diagram
eTable 1. Inpatient Infections Included in Analysis
eTable 2. Outpatient Infections Included in Analysis
eTable 3. Pediatric Comorbidity Score
eTable 4. Pooled CONSORT Table
eTable 5. Pooled Patient Characteristics With Standardized Differences for Ustekinumab vs Etanercept, Before and After Propensity-Score Matching
eTable 6. Pooled Incidence Rate and Rate Ratio of Infections (2009-2021), Unadjusted
eTable 7. Pooled Incidence Rate of Serious Infections Stratified by Years, After PS-Decile Stratification
eReferences
Data Sharing Statement
References
- 1.Tollefson MM, Van Houten HK, Asante D, Yao X, Maradit Kremers H. Association of psoriasis with comorbidity development in children with psoriasis. JAMA Dermatol. 2018;154(3):286-292. doi: 10.1001/jamadermatol.2017.5417 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Nogueira M, Paller AS, Torres T. Targeted therapy for pediatric psoriasis. Paediatr Drugs. 2021;23(3):203-212. doi: 10.1007/s40272-021-00443-5 [DOI] [PubMed] [Google Scholar]
- 3.Gelfand JM, Weinstein R, Porter SB, Neimann AL, Berlin JA, Margolis DJ. Prevalence and treatment of psoriasis in the United Kingdom: a population-based study. Arch Dermatol. 2005;141(12):1537-1541. doi: 10.1001/archderm.141.12.1537 [DOI] [PubMed] [Google Scholar]
- 4.De Jager MEA, De Jong EMGJ, Evers AWM, Van De Kerkhof PCM, Seyger MMB. The burden of childhood psoriasis. Pediatr Dermatol. 2011;28(6):736-737. doi: 10.1111/j.1525-1470.2011.01489.x [DOI] [PubMed] [Google Scholar]
- 5.Tollefson MM, Crowson CS, McEvoy MT, Maradit Kremers H. Incidence of psoriasis in children: a population-based study. J Am Acad Dermatol. 2010;62(6):979-987. doi: 10.1016/j.jaad.2009.07.029 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Menter A, Cordoro KM, Davis DMR, et al. Joint American Academy of Dermatology–National Psoriasis Foundation guidelines of care for the management and treatment of psoriasis in pediatric patients. J Am Acad Dermatol. 2020;82(1):161-201. doi: 10.1016/j.jaad.2019.08.049 [DOI] [PubMed] [Google Scholar]
- 7.Aslam N, Saleem H, Murtazaliev S, Quazi SJ, Khan S. FDA approved biologics: can etanercept and ustekinumab be considered a first-line systemic therapy for pediatric/adolescents in moderate to severe psoriasis? a systematic review. Cureus. 2020;12(8):e9812. doi: 10.7759/cureus.9812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Paller AS, Siegfried EC, Langley RG, et al. ; Etanercept Pediatric Psoriasis Study Group . Etanercept treatment for children and adolescents with plaque psoriasis. N Engl J Med. 2008;358(3):241-251. doi: 10.1056/NEJMoa066886 [DOI] [PubMed] [Google Scholar]
- 9.Landells I, Marano C, Hsu MC, et al. Ustekinumab in adolescent patients age 12 to 17 years with moderate-to-severe plaque psoriasis: results of the randomized phase 3 CADMUS study. J Am Acad Dermatol. 2015;73(4):594-603. doi: 10.1016/j.jaad.2015.07.002 [DOI] [PubMed] [Google Scholar]
- 10.Philipp S, Menter A, Nikkels AF, et al. Ustekinumab for the treatment of moderate-to-severe plaque psoriasis in paediatric patients (≥6 to <12 years of age): efficacy, safety, pharmacokinetic and biomarker results from the open-label CADMUS Jr study. Br J Dermatol. 2020;183(4):664-672. doi: 10.1111/bjd.19018 [DOI] [PubMed] [Google Scholar]
- 11.Asgari MM, Wu JJ, Gelfand JM, et al. Validity of diagnostic codes and prevalence of psoriasis and psoriatic arthritis in a managed care population, 1996-2009. Pharmacoepidemiol Drug Saf. 2013;22(8):842-849. doi: 10.1002/pds.3447 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Eichenfield LF, Ahluwalia J, Waldman A, Borok J, Udkoff J, Boguniewicz M. Current guidelines for the evaluation and management of atopic dermatitis: a comparison of the Joint Task Force Practice Parameter and American Academy of Dermatology guidelines. J Allergy Clin Immunol. 2017;139(4S):S49-S57. doi: 10.1016/j.jaci.2017.01.009 [DOI] [PubMed] [Google Scholar]
- 13.Johnson ES, Bartman BA, Briesacher BA, et al. The incident user design in comparative effectiveness research. Pharmacoepidemiol Drug Saf. 2013;22(1):1-6. doi: 10.1002/pds.3334 [DOI] [PubMed] [Google Scholar]
- 14.Schneeweiss S, Rassen JA, Brown JS, et al. Graphical depiction of longitudinal study designs in health care databases. Ann Intern Med. 2019;170(6):398-406. doi: 10.7326/M18-3079 [DOI] [PubMed] [Google Scholar]
- 15.Schneeweiss S, Robicsek A, Scranton R, Zuckerman D, Solomon DH. Veteran’s Affairs hospital discharge databases coded serious bacterial infections accurately. J Clin Epidemiol. 2007;60(4):397-409. doi: 10.1016/j.jclinepi.2006.07.011 [DOI] [PubMed] [Google Scholar]
- 16.Wiese AD, Griffin MR, Stein CM, et al. Validation of discharge diagnosis codes to identify serious infections among middle age and older adults. BMJ Open. 2018;8(6):e020857. doi: 10.1136/bmjopen-2017-020857 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sahli L, Lapeyre-Mestre M, Derumeaux H, Moulis G. Positive predictive values of selected hospital discharge diagnoses to identify infections responsible for hospitalization in the French national hospital database. Pharmacoepidemiol Drug Saf. 2016;25(7):785-789. doi: 10.1002/pds.4006 [DOI] [PubMed] [Google Scholar]
- 18.Holland-Bill L, Xu H, Sørensen HT, et al. Positive predictive value of primary inpatient discharge diagnoses of infection among cancer patients in the Danish National Registry of Patients. Ann Epidemiol. 2014;24(8):593-597, 597.e1-597.e18. doi: 10.1016/j.annepidem.2014.05.011 [DOI] [PubMed] [Google Scholar]
- 19.Penso L, Dray-Spira R, Weill A, Pina Vegas L, Zureik M, Sbidian E. Association between biologics use and risk of serious infection in patients with psoriasis. JAMA Dermatol. 2021;157(9):1056-1065. doi: 10.1001/jamadermatol.2021.2599 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Kirchgesner J, Desai RJ, Beaugerie L, Schneeweiss S, Kim SC. Risk of serious infections with vedolizumab versus tumor necrosis factor antagonists in patients with inflammatory bowel disease. Clin Gastroenterol Hepatol. 2022;20(2):314-324.e16. doi: 10.1016/j.cgh.2020.12.030 [DOI] [PubMed] [Google Scholar]
- 21.Sun JW, Bourgeois FT, Haneuse S, et al. Development and validation of a pediatric comorbidity index. Am J Epidemiol. 2021;190(5):918-927. doi: 10.1093/aje/kwaa244 [DOI] [PubMed] [Google Scholar]
- 22.Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Stürmer T. Variable selection for propensity score models. Am J Epidemiol. 2006;163(12):1149-1156. doi: 10.1093/aje/kwj149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41-55. doi: 10.1093/biomet/70.1.41 [DOI] [Google Scholar]
- 24.Desai RJ, Rothman KJ, Bateman BT, Hernandez-Diaz S, Huybrechts KF. A propensity-score–based fine stratification approach for confounding adjustment when exposure is infrequent. Epidemiology. 2017;28(2):249-257. doi: 10.1097/EDE.0000000000000595 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Yoshida K, Solomon DH, Haneuse S, et al. Multinomial extension of propensity score trimming methods: a simulation study. Am J Epidemiol. 2019;188(3):609-616. doi: 10.1093/aje/kwy263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rothman KJ, Lash TL, VanderWeele TJ, Haneuse S. Modern Epidemiology. 4th ed. Wolters Kluwer; 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Franklin JM, Rassen JA, Ackermann D, Bartels DB, Schneeweiss S. Metrics for covariate balance in cohort studies of causal effects. Stat Med. 2014;33(10):1685-1699. doi: 10.1002/sim.6058 [DOI] [PubMed] [Google Scholar]
- 28.Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083-3107. doi: 10.1002/sim.3697 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wang SV, Verpillat P, Rassen JA, Patrick A, Garry EM, Bartels DB. Transparency and reproducibility of observational cohort studies using large healthcare databases. Clin Pharmacol Ther. 2016;99(3):325-332. doi: 10.1002/cpt.329 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Kim SC, Solomon DH, Rogers JR, et al. Cardiovascular safety of tocilizumab versus tumor necrosis factor inhibitors in patients with rheumatoid arthritis: a multi-database cohort study. Arthritis Rheumatol. 2017;69(6):1154-1164. doi: 10.1002/art.40084 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Patorno E, Schneeweiss S, Gopalakrishnan C, Martin D, Franklin JM. Using real-world data to predict findings of an ongoing phase IV cardiovascular outcome trial: cardiovascular safety of linagliptin versus glimepiride. Diabetes Care. 2019;42(12):2204-2210. doi: 10.2337/dc19-0069 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Aetion. Aetion Evidence Platform: software for real-world data analysis. Accessed December 28, 2022. https://aetion.com/technology/platform/
- 33.Papp K, Thaçi D, Marcoux D, et al. Efficacy and safety of adalimumab every other week versus methotrexate once weekly in children and adolescents with severe chronic plaque psoriasis: a randomised, double-blind, phase 3 trial. Lancet. 2017;390(10089):40-49. doi: 10.1016/S0140-6736(17)31189-3 [DOI] [PubMed] [Google Scholar]
- 34.van Geel MJ, Oostveen AM, Hoppenreijs EP, et al. Methotrexate in pediatric plaque-type psoriasis: long-term daily clinical practice results from the Child-CAPTURE registry. J Dermatolog Treat. 2015;26(5):406-412. doi: 10.3109/09546634.2014.996515 [DOI] [PubMed] [Google Scholar]
- 35.Schneeweiss MC, Huang JT, Wyss R, Schneeweiss S, Merola JF. Serious infection risk in children with psoriasis on systemic treatment: a propensity score–matched population-based study. J Am Acad Dermatol. 2020;82(6):1337-1345. doi: 10.1016/j.jaad.2020.02.065 [DOI] [PubMed] [Google Scholar]
- 36.Sherman RE, Anderson SA, Dal Pan GJ, et al. Real-world evidence—what is it and what can it tell us? N Engl J Med. 2016;375(23):2293-2297. doi: 10.1056/NEJMsb1609216 [DOI] [PubMed] [Google Scholar]
- 37.Seeger JD, Bykov K, Bartels DB, Huybrechts K, Schneeweiss S. Propensity score weighting compared to matching in a study of dabigatran and warfarin. Drug Saf. 2017;40(2):169-181. doi: 10.1007/s40264-016-0480-3 [DOI] [PubMed] [Google Scholar]
- 38.Zhou Y, Matsouaka RA, Thomas L. Propensity score weighting under limited overlap and model misspecification. Stat Methods Med Res. 2020;29(12):3721-3756. doi: 10.1177/0962280220940334 [DOI] [PubMed] [Google Scholar]
- 39.Stürmer T, Rothman KJ, Avorn J, Glynn RJ. Treatment effects in the presence of unmeasured confounding: dealing with observations in the tails of the propensity score distribution—a simulation study. Am J Epidemiol. 2010;172(7):843-854. doi: 10.1093/aje/kwq198 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165(6):710-718. doi: 10.1093/aje/kwk052 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure. Study Design Diagram
eTable 1. Inpatient Infections Included in Analysis
eTable 2. Outpatient Infections Included in Analysis
eTable 3. Pediatric Comorbidity Score
eTable 4. Pooled CONSORT Table
eTable 5. Pooled Patient Characteristics With Standardized Differences for Ustekinumab vs Etanercept, Before and After Propensity-Score Matching
eTable 6. Pooled Incidence Rate and Rate Ratio of Infections (2009-2021), Unadjusted
eTable 7. Pooled Incidence Rate of Serious Infections Stratified by Years, After PS-Decile Stratification
eReferences
Data Sharing Statement