Family report compared to clinician-documented diagnoses for psychiatric conditions among hospitalized children

At a tertiary-care children’s hospital, we found high point prevalence of comorbid psychiatric conditions and low agreement between FR- and CD-identified psychiatric conditions. Estimates of the prevalence of psychiatric comorbidity among pediatric medical and surgical inpatients were similar for FR- and CD-identified psychiatric conditions, though each method missed about one third of the cases identified by the other method. FR only and CD only each identified about 1 in 4 or 5 hospitalized children and adolescents with a psychiatric comorbidity. When FR and CD were combined, a comorbid psychiatric diagnosis was identified in about 1 in 3 medical and surgical inpatients aged 4 to 21 years. FR–CD agreement was substantial only for ADHD and was fair to slight for most other psychiatric conditions, including autism, depression, anxiety, and disruptive behavior disorders (eg, conduct disorder, oppositional defiant disorder).

Our finding that psychiatric conditions were more commonly reported by families and documented by clinicians for white patients is consistent with a large body of evidence showing that racial or ethnic minority patients experience more stigma related to mental health diagnoses and use mental health services less.^29-33 Families were more likely to report use of mental health services than a known mental health diagnosis. This finding may reflect families’ willingness to use services even if they do not understand or experience stigma related to psychiatric diagnoses. Alternatively, use of mental health services without a diagnosis may reflect clinicians’ willingness to refer a child for services when the child is perceived to have an impairment even in the absence of a clear psychiatric diagnosis.

The low FR–CD agreement regarding psychiatric conditions in hospitalized children and adolescents raises 3 issues for pediatric hospital care. First, earlier studies likely underestimated the prevalence of these conditions. A 2014 study of a national sample found that 13% of children hospitalized for a physical health condition had psychiatric comorbidity.²⁵ That study and other large-scale studies showing a high and increasing prevalence of primary psychiatric conditions in hospitalized children and adolescents have relied on administrative data derived from clinician-documented diagnoses.^25-27 Our study findings suggest that reliance on administrative data could result in underestimation of the prevalence of psychiatric comorbidity in hospitalized children by as much as 40%. Pediatric hospitals are reporting a shortage of pediatric mental health specialists.³⁴ Augmenting estimates of the prevalence of psychiatric comorbidity in hospitalized children with reports from other sources, including families or outpatient administrative records, may aid health systems in allocating mental health resources for pediatric inpatients.

The second issue is that the present data suggest that families and clinicians do not share the same information about a child’s psychiatric diagnoses when the child is hospitalized for a medical condition or surgical procedure. Low FR–CD agreement regarding psychiatric diagnoses suggests families and clinical teams are not always “on the same page” about psychiatric needs during hospitalization. Implications of this finding are relevant to inpatient and ambulatory care settings. In cases in which a clinician recognizes a psychiatric condition but the family does not, the family may not seek outpatient treatment. In the present study, one third of patients with a psychiatric diagnosis identified by CD but not FR were not engaged in ambulatory treatment for the condition. Conversely, a psychiatric diagnosis identified by FR but not CD suggests clinical teams lack the skills and knowledge needed to elicit information about psychiatric conditions and their potential relevance to inpatient care. As a result, clinicians may miss opportunities to provide interventions that may improve physical or mental health outcomes. For example, clinical teams with information about a patient’s anxiety disorder may be better able to provide brief interventions to prevent medical treatments from triggering anxiety symptoms and to mitigate the risk for traumatic stress symptoms related to the hospitalization.

The third issue is that anxiety disorders were most likely to be the subject of FR–CD disagreement. This finding identifies children with anxiety disorders as a priority population for research into differences between families and clinicians in understanding patients’ psychiatric diagnoses. Our findings suggest families and clinicians have different views of patients’ anxiety symptoms. Anxiety disorders are a risk factor for worse outcomes in children with chronic physical conditions,^3,35-37 and acute hospitalization is associated with posthospital anxiety symptoms.^38,39 Thus, anxiety disorders are particularly relevant to hospital care and are a priority for research on the differences between families’ and clinicians’ perspectives on children’s psychiatric diagnoses.

Our findings should be interpreted in the context of study limitations. First, because of resource limitations, we did not obtain psychiatric diagnostic evaluations or records to confirm FR- and CD-identified psychiatric diagnoses. Although this lack of clinical confirmation could have resulted in misclassification bias, the risk of bias was no higher than in many other studies that have successfully used hospital records^21,25 and family reports to identify psychiatric comorbidity.⁴⁰ Second, because the study included only English-speaking patients and families, results cannot be generalized to non-English-speaking populations. Third, this was a single-center study, conducted in a free-standing tertiary-care children’s hospital. Sample size was small, particularly for estimating the prevalence of individual psychiatric conditions. Patient characteristics and clinical practice patterns may differ at other types of hospitals. Larger multicenter studies are warranted. Despite these limitations, our results provide important new information that can further our understanding of the epidemiology of psychiatric conditions in hospitalized children. This information should interest clinical teams caring for children with comorbid physical and mental health conditions.