A retrospective cohort study to evaluate the development of comorbidities, including psychiatric comorbidities, among a pediatric psoriasis population

Abstract Background/Objective Compared with the adult psoriasis population, knowledge about the incidence of comorbidities in the pediatric psoriasis population is limited. The objective of this study was to assess the prevalence and incidence of comorbidities, including psychiatric comorbidities, in patients with pediatric psoriasis. Methods In this claims‐based, retrospective cohort study, patients with pediatric psoriasis were matched 1:3 with a nonpsoriasis cohort based on age, sex, and index date (the earliest of inpatient claims or the latter of two outpatient claims). Results Obesity, serious infection, and juvenile idiopathic arthropathy had higher prevalence and incidence rates in the psoriasis cohort than the nonpsoriasis cohort. Psychiatric comorbidities were also more common in the psoriasis cohort than the nonpsoriasis cohort, as were ulcerative colitis and Crohn disease. Stratifying the psoriasis cohort by disease severity—mild and moderate‐to‐severe—found no differences in incidence rates of comorbidities between the two subsets. Conclusion The incidence rates of many comorbid conditions were higher for patients with pediatric psoriasis compared with patients without pediatric psoriasis, and similar between patients with moderate‐to‐severe and mild pediatric psoriasis.


| Study design and patient population
This claims-based, retrospective cohort study utilized data from the MarketScan ® Commercial Claims and Encounters Database between Institutional review board approval and formal consent were not required because patient data were de-identified by an independent third party before author review, in compliance with Health Insurance Portability and Accountability Act regulations.

| Outcome measures
Patients were followed from the index date until the earliest of the following: disenrollment from the database, end of data collection (June 30, 2015), or occurrence of each comorbidity being assessed. Prevalent comorbidities were captured during the baseline period; incident comorbidities were captured during the follow-up period and excluded prevalent conditions. Comorbidities in this analysis were assessed individually and defined by one inpatient or two outpatient claims; comorbidities were not mutually exclusive, as a prevalent comorbidity did not preclude a patient from being evaluated for another incident comorbidity. A summary outcome-"any comorbidity"-was defined by the number of patients with at least one comorbid condition. Each comorbidity was defined by its specific ICD-9 code (Supporting Information). In particular, serious infections were defined by a primary diagnosis of the ICD-9 code "serious infection" for an inpatient hospitalization with at least one overnight stay, as previously described. 8 Psychiatric comorbidities were defined by one inpatient or outpatient claim accompanied by a pharmacy or medication procedure claim for psychiatric medications (Supporting Information); a summary outcome-"any psychiatric comorbidity"-was defined by the number of patients with at least one psychiatric comorbid condition. Comorbidities were evaluated separately in the moderate-to-severe and mild subsets.

| Statistical analyses
Descriptive statistics were generated for all baseline data, for example, sample size, percentage, and mean (standard deviation).
Prevalence per 1000 patients and IRs per 1000 person-years with 95% confidence intervals (CIs) were calculated for each comorbidity during the baseline period. IRRs with 95% CIs compared psoriasis and nonpsoriasis cohorts, and moderate-to-severe and mild subsets; P-values for IRRs were based on z-score and calculated using Mid-P exact test. The mean number of comorbidities and psychiatric comorbidities was estimated among all patients and patients with at least one comorbidity within each cohort.

| Patient disposition
This study included 38 430 patients (N = 7686 psoriasis, N = 30 744 nonpsoriasis). Patients from the two cohorts differed significantly (chi-squared P < 0.0001) in follow-up time, number of visits to any doctor for any reason during follow-up, and region (Table 1).
Within the psoriasis cohort, 65% of patients were diagnosed by a dermatologist. Within the nonpsoriasis cohort, the most frequent physician specialties were pediatrician, family practice, medical doctor, unknown, and dermatologist. During the baseline period, 4.4% of patients in the psoriasis cohort and 0.2% in the nonpsoriasis cohort had csDMARD exposure; 19.6% in the psoriasis and 7.9% in the nonpsoriasis cohort had oral steroids exposure.

| Comorbidities
The prevalence (95% CI) per 1000 patients of any comorbidity was  Table 3). Types of serious infection were not different between cohorts (Table S1).
Because steroid use may increase the risk of serious infections, prevalence and IR of serious infections were determined in patients with baseline steroid exposure. Among those patients, prevalence (21.28 vs 11.10) and IR (10.82 vs 6.03) of serious infections were higher in the psoriasis vs the nonpsoriasis cohort, respectively.

| Psychiatric comorbidities
The prevalence of any psychiatric comorbidity was 22.64 (19.40-26.26) in the psoriasis cohort and 13.40 (12.14-14.76) in the nonpsoriasis ( Table 2). Prevalence of each psychiatric comorbidity is presented in Table 2.

| Comorbidities by disease severity
In the psoriasis cohort, 1149 (15.0%) patients had moderate-to-severe psoriasis. They were treated with phototherapy (46.3%; of whom 96.8% received UVB and 5.6% PUVA), csDMARDs (35.7%), TNFi's (31.0%), and other biologics (4.9%) on or after the index date. They differed from the mild subset in baseline medication exposure (Table S2). csDMARD, conventional synthetic disease-modifying antirheumatic drug; SD, standard deviation. a In this case-control study, patients were matched by gender and age at index; therefore, there is no difference between the patients with pediatric psoriasis and patients without pediatric psoriasis based on these variables. b Physician specialty refers to the physician who provided the confirmed diagnosis of psoriasis for the inpatient claim or latter of the two outpatient claims. The five most common physician specialties in "Other" included acute care hospital, physician assistant, internal medicine, nurse practitioner, and rheumatologist.

| Mean number of comorbidities
The mean number of comorbidities (nonpsychiatric and psychiatric) did not differ between patients with and without psoriasis (Table S3).  4,[9][10][11] Estimates for the prevalence of obesity ranged from 1.8% to 20.2%. 1,12 We found that obesity and hyperlipidemia were associated with psoriasis based on incidence rates in the psoriasis and nonpsoriasis cohorts; in contrast, we did not find an association between psoriasis and hypertension or diabetes. 5,6 The number of obese patients in this study is smaller than found in other studies, likely due to the nature of adjudicated claims databases and data collection, which relies on ICD-9 coding by physicians. In a U.S. study of patients with moderate-to-severe psoriasis, 37% were obese; in a German study of patients with mild-to-severe psoriasis, 48% were overweight. 4,13 Obesity may be associated with psoriasis severity, as one study found a greater number of patients with mild psoriasis were overweight and a greater number with severe TA B L E 3 Incidence rates and incidence rate ratios of any comorbidity, including psychiatric comorbidities psoriasis were obese; 4 we found similar rates of obesity between moderate-to-severe and mild patients. Because psoriasis is a risk factor for cardiovascular disease, recent guidelines recommend screening yearly for overweightness, obesity, and hypertension and every 3 years for type 2 diabetes mellitus. 14 The serious infection rate observed in this study was higher than expected. Incidence rates were similar in moderate-to-severe and mild patients but were higher in patients with steroid exposure, which was more prevalent in the psoriasis cohort than nonpsoriasis.

| D ISCUSS I ON
In the absence of additional information on the timing between steroid exposure and serious infection, an association between the two is speculative.
The relative risk of incident CD and UC was greater in the psoriasis cohort compared with nonpsoriasis. 15 Similarly, Augustin et al.
found a fourfold increase in CD; however, that study also found a nonsignificant increase in UC in patients with psoriasis. 5,16,17 Because patients with JIA, PsA, RA, CD, and UC were excluded from the analysis between patients with moderate-to-severe and mild psoriasis, we could not assess incidence of these inflammatory conditions in patients by disease severity.
Psychiatric comorbidities in this study were associated with higher IRs in patients with psoriasis than those without, consistent with previous studies. A previous study based on the U.S.
MarketScan ® database found that depression, anxiety, and bipolar disorder were more common in patients with psoriasis. 6 In the Danish Civil Registration System, IRs of all psychiatric comorbidities evaluated-alcohol abuse, anxiety, depression, and eating disorders-were higher among patients with psoriasis, although the difference was not significant for anxiety. 7 Some studies have reported psychiatric disorders to be common in patients with moderate-to-severe psoriasis; however, similar results were not found in this analysis. 5 within a similar range as reported previously. 6 Another limitation of claims-based data is that we cannot determine why patients received steroids, though likely related to other comorbid conditions such as CD. Further exploration of steroid use in this population is needed.
As mentioned above, obesity is likely to have been underreported, considering obesity was captured only by diagnosis codes and not by body mass index levels. 4,6 Incidence of obesity may also be low because obesity has been found to predate the psoriasis diagnosis in the pediatric population, 29 resulting in a higher prevalence of obesity in the psoriasis cohort in this study; incident diagnoses of obesity were only captured if developed after psoriasis onset. The reliance of claims-based data on ICD-9 coding may also contribute to the underreporting of obesity. Similar to obesity, diagnoses of hypertension, hyperlipidemia, and diabetes mellitus type 2 were based on ICD-9 codes rather than laboratory values, which were not available in this claims database. Because patients in the psoriasis cohort visited the doctor more often (Table 1), there may be ascertainment bias, as those patients may have been screened more often for comorbidities. Because this study required a 12-month wash-out period, if a patient was diagnosed with a comorbidity before that period and again during the follow-up period, that comorbidity would have been considered incident rather than prevalent; however, this applies to both cohorts. Additionally, comorbidities with a longer latency period (eg, more than 2 years, as for malignancies) may not have been captured during study follow-up; as both cohorts had similar follow-up periods, this is unlikely to result in ascertainment bias. Because this study was intended to be descriptive, we were unable to determine whether psoriasis is independently associated with the comorbidities assessed or driven by confounders.
In conclusion, this descriptive study offers more insight, based on a very large sample size, of the association between psoriasis and comorbidities in pediatric patients. The incidence rate ratios of many comorbidities were higher for pediatric patients with psoriasis than those without. Patients with moderate-to-severe psoriasis did not appear to have different IRRs compared with those with mild psoriasis. Future queries of databases where disease activity measures are captured would be useful in addressing differences in comorbidities between mild vs moderate-to-severe psoriasis, particularly as it relates to obesity and other components of the metabolic syndrome (ie, cardiovascular disease risk factors). Exploration of comorbidities in patients stratified by therapy as well as the dose-response relationship between psoriasis severity and obesity is also needed.

ACK N OWLED G M ENTS
Jessica Ma, Ph.D. (Amgen Inc.) provided medical writing assistance.
Qualified researchers may request data from Amgen clinical studies.

CONFLICT OF INTEREST
A.S. Paller has been an investigator without personal compensation for AbbVie, Amgen Inc., Celgene, Janssen, Leo, and Novartis; and a consultant with honorarium for Amgen Inc., Eli Lilly, Novartis, and UCB. J. Schenfeld is a contract worker for Amgen Inc. and receives salary from Amgen Inc. through DOCS Global, Inc. N.A. Accortt and G. Kricorian are employees and shareholders of Amgen Inc.

INSTITUTIONAL REVIEW BOARD APPROVAL
This study did not require approval by an institutional review board because the patient data in this analysis were de-identified by an independent third party prior to initial review by the authors, in compliance with Health Insurance Portability and Accountability Act regulations. For this type of study, formal consent is not required.