Association of Psychiatric Comorbidities With Treatment and Outcomes in Pediatric Migraines Free

This multicenter, retrospective cohort study examined data from the Pediatric Health Information System (PHIS), an administrative database of 51 freestanding pediatric hospitals in the United States affiliated with the Children’s Hospital Association (Lenexa, KS), and was approved by the investigator’s site institutional review board.

Children aged 6 to 18 hospitalized (inpatient or observation status) with a principal discharge diagnosis of migraine headache between January 1, 2010, and June 30, 2018, were included. A migraine headache diagnosis was defined according to the International Classification of Diseases, 9th and 10th Revision, Clinical Modification codes. For each patient, the top 2 secondary psychiatric diagnoses were identified according to a PHIS-established list of International Classification of Diseases, 9th revision, and International Classification of Diseases, 10th revision Clinical Modification codes associated with psychiatric diagnoses.¹⁰  Patients were excluded if diagnosed with a complex chronic condition according to PHIS criteria.¹¹ 

For de-identification purposes, patients were assigned a unique encrypted patient code to allow tracking across visits. Patient data retrieved included presence and type of psychiatric comorbid condition(s), demographics (age, gender, race, and ethnicity), medications, and outcomes (length of stay [LOS], readmission within 7 and 30 days, and cost). Costs were estimated through charges using hospital and year-specific cost-to-charge ratios.¹² 

Patients were categorized into 2 groups according to the presence or absence of any psychiatric comorbidity. The yearly rate of psychiatric comorbidities was compared over the study period using the Cochran-Armitage trend test. The distribution of demographics, interventions, and outcomes was summarized within each group of patients, with and without psychiatric comorbidity and compared among groups using the Fisher’s exact or Wilcoxon rank sum test as appropriate for the data distribution. The magnitude of association between psychiatric comorbidity and outcomes was estimated as the difference in means or odds ratio, along with corresponding 95% confidence intervals to indicate variability. For each outcome, possible confounding by demographics associated with psychiatric comorbidity was assessed with multivariable linear and logistic regression models. As the distribution of LOS and cost was right-tailed, the natural logarithm transformation was applied to better approximate a Gaussian distribution for linear regression models. Statistical analyses were performed using the software package SAS (SAS Institute, Cary, NC), and all statistical tests were 2-sided, with significance evaluated at the 5% level.

During the study period, 21 436 patients meeting criteria were hospitalized. The mean age was 13.6 years, and a majority of patients were female (15 419 [72%]), who self-identified as White (16 141 [75%]) and non-Hispanic (17 483 [87%]). The prevalence of psychiatric comorbidities increased significantly, from 22% in 2010 to a high of 39% in 2017 (Table 1). Psychiatric comorbidities were present in 6796 (32%) patients, with 4612 (68%) patients diagnosed with 1 or more of the 3 top diagnoses: anxiety (2415; prevalence 11.2%), depression (1433; prevalence 6.7%), and ADHD (1411; prevalence 6.5%). The remaining 2184 patients (32%) were diagnosed with 1 or more of 250 other psychiatric conditions, each with a prevalence of <1%.

Patients with any psychiatric comorbidities were on average 1 year older and more likely to identify as White (Table 1).

Patients with versus without psychiatric comorbidities were significantly more likely to receive treatment with most medications investigated (Table 1), notably with DHE (61% vs 54%), topiramate (23% vs 18%), and valproate (38% vs 34%).

The mean LOS was significantly longer for patients with versus without psychiatric comorbidities (2.6 days vs 2.0 days). Consistent with longer LOS, average cost was significantly higher for patients with psychiatric comorbidities versus those without ($8749 vs $7040). Readmission rates within 7 and 30 days were also significantly higher among patients with versus without psychiatric comorbidities (10% vs 8% and 21% vs 17%, respectively) (Table 2). Confounding by demographics was not detected for any of the outcomes.

This large cohort study demonstrates that 32% of pediatric patients hospitalized with migraine headache are diagnosed with 1 or more psychiatric conditions. The most common comorbid diagnoses identified in our population were anxiety, depression, and ADHD, with the estimated prevalence of each well above that reported in the general population. Furthermore, associations were identified between the presence of psychiatric comorbidities and increased medication administration, LOS, cost, and readmissions for children hospitalized for migraine.

The association between psychiatric disorders and migraine headache has been well documented in adult patients,^13,14  and recent outpatient studies in children have also suggested a correlation.^8,9,15,16  Our study estimated a high rate of psychiatric disorders among pediatric patients hospitalized for migraine, aligning with recent outpatient studies reporting psychiatric comorbidity in 40% to 56% of children with migraine headache.^8,9  Additionally, anxiety, depression, and ADHD were the most prevalent psychiatric comorbidities in our hospitalized population, which is consistent with previous reports.^5,6 

Our study demonstrates that patients hospitalized for migraine headache with a psychiatric comorbidity have increased use of most medications, both for immediate headache abortion and headache prophylaxis, and also require increased use of agents usually reserved for treatment of severe, intractable migraine (eg, DHE). It is unclear why patients with psychiatric comorbidity require increased medication use for migraine management; however, this factor may have contributed to the longer LOS and higher cost of hospitalization seen in this population. Studies of adult outpatients have shown that, if left untreated, comorbid psychiatric conditions can increase migraine-related disability, reduce quality of life, and negatively impact treatment outcomes.^13,17  Although it is possible that untreated psychiatric conditions in the pediatric migraine headache patients in our study influenced the increased medication use and poor outcomes identified, further study is needed to clarify.

For all patients with migraine headaches, it has been suggested that inpatient migraine treatment requires a biopsychosocial and multidisciplinary approach.¹⁸  A meta-analysis of 35 randomized controlled trials concerning the use of nonpharmacologic pain management methods in children with migraine and tension-type headache have confirmed their efficacy in the outpatient setting.¹⁹  We postulate that the impact of this multimodal, multidisciplinary treatment approach for children with both migraines and psychiatric conditions may be even greater. It has also been suggested that incorporation of treatment of known psychiatric disorders into the overall management of children with migraine may facilitate modification of the trajectories of both conditions across their lifespan.²⁰  Future studies are needed to evaluate the clinical impact of treated versus untreated psychiatric comorbidity and to explore the feasibility and effectiveness of a multidisciplinary treatment model for children with a psychiatric disorder who are hospitalized for the treatment of migraine headache.

This retrospective analysis of a multisite billing database is subject to issues of data quality, including potentially incomplete and inaccurate information. Inaccurate documentation of comorbid diagnoses may have led to false classification of patients into the wrong study group. Unavailable data on treatment of psychiatric comorbidities and duration of each migraine treatment limit the ability to tease out factors contributing to higher LOS and cost. Possible confounding bias was checked using appropriate multivariable models. In addition, the data source is limited to those tertiary care children’s hospitals that contributed data to PHIS and may not be reflective of all children treated for migraine headache in a hospital setting.

Psychiatric comorbidities are prevalent in 32% of pediatric patients hospitalized for treatment of migraine headache. These comorbid states are statistically associated with increased use of therapies, longer LOS, increased cost, and increased readmission. Prospective studies are recommended to confirm clinical relevance and identify optimal multidisciplinary care models that comprehensively address both neurologic and psychiatric conditions for pediatric migraine patients with psychiatric comorbidities in the inpatient setting.