Introduction
Delirium is characterized by fluctuating consciousness, cognitive dysfunction, with reduced ability to focus, sustain, or shift attention (American Psychiatric Association, 2013). A combination of a systemic illness, effects of treatment, and the abnormal hospital environment contribute to developing delirium (Traube et al., Reference Traube, Ariagno and Thau2017a). In adults, delirium is associated with increased morbidity and mortality, increased length of hospitalization, higher health care costs, and higher levels of distress in patients and families (Breitbart and Alici, Reference Breitbart and Alici2012). More recently, recognition of delirium in the pediatric population and its impact on clinical outcomes has increased research, mostly in the area of pediatric critical care. The development of validated screening tools has allowed for universal screening in pediatric intensive care units, yielding prevalence rates of 13–28% (Daoud et al., Reference Daoud, Duff and Joffe2014). In a large study of over almost 1,600 critically ill pediatric patients, delirium was diagnosed in 17% of patients. Identified risk factors for developing delirium in this study included age ≤2 years, developmental delay, severity of illness, prior coma, mechanical ventilation, and benzodiazepine/anticholinergic use. Delirium increased hospital length of stay in these children and was an independent predictor of mortality (Traube et al., Reference Traube, Silver and Gerber2017b).
Despite the medical complexity of pediatric cancer patients, less is known about delirium in this population than adults (Combs et al., Reference Combs, Rice and Kopp2014; Traube et al., Reference Traube, Augenstein and Greenwald2014b, Reference Traube, Ariagno and Thau2017a; Winsnes et al., Reference Winsnes, Sochacki and Eriksson2019). A retrospective study of admissions over a 3-month period to the pediatric cancer service found a delirium incidence of 18.8%. Age ≤5 years old, primary diagnosis of brain tumor, postoperative status and benzodiazepines were independently associated with the development of delirium (Traube et al., Reference Traube, Ariagno and Thau2017a). A 1-year prospective study in pediatric hematology, oncology, and bone marrow transplant patients found a delirium prevalence of only 5% and significant associations with increased length of stay, admission to the bone marrow transplant service, patient location [Pediatric Intensive Care Unit (PICU) vs. Pediatric Hematology/Oncology (PHO) unit], benzodiazepine, opioid, and anticholinergic administration (Winsnes et al., Reference Winsnes, Sochacki and Eriksson2019).
Given preliminary evidence that delirium occurs frequently within pediatric oncology population with serious consequences of increased length of hospital stay and mortality, we wanted to add to the emerging body of the literature by evaluating delirium identification and documentation in hospitalized pediatric cancer and bone marrow transplant patients at our institution. We hypothesized that delirium is poorly recognized and documented within this population.
Methods
A retrospective, observational study conducted at a National Cancer Institute (NCI)-designated cancer center. The Cancer Informatics Core at Massey Cancer Center conducted a search of the Massey Data Analysis System (MDAS) and Clinical Trials Eligibility Database (CTED) for hospitalized pediatric oncology patients aged less than 21 years of age with active cancer from January 2013 to March 2016. Billing administrative data was searched for the following ICD codes: 780.09, 780.97, 293.0, 293.1, 293.89, 292.81, 300.11, R40.41, F05, F44.89, F06.1, F44.4, F44.6, F53, R40.0, R40.1, R41.82 (see Appendix). Administrative data yielded specific medications used during hospitalization (Lorazepam, Midazolam, Alprazolam, Morphine, Hydromorphone, Fentanyl, Oxycodone, Haloperidol, Quetiapine, Risperidone, Diazepam, and Olanzapine) and for the presence of a psychiatry consult. A term search of the medical record used the following key words: delirium, confusion, encephalopathy, agitation, inattentive/inattentiveness/inattention, altered mental status, ICU psychosis, and delirious. Patients with developmental delay, autism, or co-morbid psychiatric conditions (i.e., schizophrenia, bipolar disorder) were all excluded given the difficulty of retrospectively diagnosing delirium in these populations.
For patient encounters identified by any one of the above mechanisms, the electronic medical record was reviewed to confirm the diagnosis of delirium using Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria. For patient encounters determined to have documentation of delirium, the reviewers collected the following data: age, race/ethnicity, underlying diagnoses (including primary cancer diagnosis) reason for admission, medication use, length of stay, and discharge plan. As a quality control measure, every tenth patient encounter was reviewed by a clinical psychologist with expertise in delirium to ensure the diagnosis was assigned appropriately by the chart reviewers.
Data were collected and managed from using Research Electronic Data Capture (REDCap), a secure web-based application. The Institutional Review Board (IRB) at Virginia Commonwealth University approved this study.
Results
We identified 1,688 admissions of hospitalized pediatric oncology patients from January 2013 to March 2016. After screening patients, and excluding those with developmental delay, autism or psychiatric disorder, 201 admissions remained for the evaluation of delirium. Excluded patients are listed in Figure 1.
Fig. 1. Diagram of medical records identified and reviewed.
Of 201 admissions reviewed, 54 (26.9%) were assessed to have documentation of delirium, from 109 unique patients. Their demographic characteristics are shown in Table 1. Of note, the overall incidence of documented delirium was 3.2% (54 of the total 1,688 admissions) or 8.2% of unique patients (43 of the total 525 hospitalized pediatric oncology patients).
Table 1. Demographic characteristics of unique patient records reviewed and evaluated for delirium
Patients prescribed opioids and benzodiazepines were more likely to have documentation of delirium during a hospitalization than those who did not receive these medications (Table 2). Over half of the patients with documentation of delirium were prescribed both opioids and benzodiazepines within 7 days before or after delirium was documented. Only five patients with documentation of delirium were not prescribed either opioid or benzodiazepines.
Table 2. Presence of delirium in relation to drugs prescribed
The accuracy of diagnoses codes in billing was compared with charts manually assessed for delirium (Table 3). Of the hospitalizations that had a hospital code and/or physician billing code for delirium, 44.8% were assessed to have no delirium through chart review (false positives). Of the hospitalizations assessed to have delirium, 70.3% had no codes for delirium on either the hospital or physician billing.
Table 3. Accuracy of billing codes vs. chart review for delirium
Discussion
In this retrospective study of pediatric cancer and bone marrow transplant admissions, delirium was documented in at least 8.2% of unique patients. This is lower than previously reported in the pediatric critical care literature (17%) or in hospitalized children with cancer (13–18.8%), likely due to the retrospective nature of our study and the lack of routine delirium screening (Traube et al., Reference Traube, Ariagno and Thau2017a, Reference Traube, Silver and Gerber2017b; Winsnes et al., Reference Winsnes, Sochacki and Eriksson2019).
Similar to other studies done in both pediatric critical care and oncology patients, we found a significant association between opioid and benzodiazepine use and delirium (Traube et al., Reference Traube, Ariagno and Thau2017a, Reference Traube, Silver and Gerber2017b; Winsnes et al., Reference Winsnes, Sochacki and Eriksson2019). Previous studies involving pediatric oncology patients reported frequent use of opioids (41.0–60.7% of admissions) and benzodiazepines (39.5–54.2%) and independent associations with increased risk of delirium (Traube et al., Reference Traube, Ariagno and Thau2017a; Winsnes et al., Reference Winsnes, Sochacki and Eriksson2019). Limiting benzodiazepine and opioid administration and optimizing opioid/benzodiazepine-sparing pharmacologic and non-pharmacologic interventions for pain, anxiety, and nausea may decrease delirium (Dupuis et al., Reference Dupuis, Boodhan and Holdsworth2013, Reference Dupuis, Robinson and Boodhan2014; Traube et al., Reference Traube, Ariagno and Thau2017a).
Findings from our study also indicate a sizable discrepancy in ICD codes and charts manually assessed for delirium. One study evaluating diagnostic coding of delirium found in-hospital delirium incidence unreliable and under-reported (Casey et al., Reference Casey, Cross and Mart2019); ICD codes as compared to a point of prevalence survey using validated tools were 30% lower. Of note, the study was conducted in adult patients in whom recognition may be more robust than the pediatric population. Our results affirm under-coding for delirium is prevalent in pediatric oncology but also reveal some concerns for over-diagnosing delirium (false positives). Education of healthcare professionals and a standardized assessment improves the identification of delirium in adults (Fadul et al., Reference Fadul, Kaur and Zhang2007). Similarly, familiarity and daily clinical use of CAP-D, a validated screening tool designed to also detect hypoactive delirium, may improve diagnosis in pediatric patients (Traube et al., Reference Traube, Silver and Kearney2014a).
Our study has several limitations including the retrospective nature and the relatively small sample size from a single NCI center. We also excluded patients with developmental delay, autism, or major psychiatric disorders, conditions previously identified as risk factors for delirium (Traube et al., Reference Traube, Silver and Gerber2017b). We excluded these patients because evaluating delirium in children with developmental delays or major psychiatric disorders often requires knowledge of child's baseline through comprehensive patient interaction. Documentation of this information may not be readily available through a hospital chart review (Traube et al., Reference Traube, Silver and Gerber2017b).
Conclusion
Delirium is a known complication of a pediatric cancer diagnosis and treatment, documented in at least 3.2% of the total admissions or 8.2% of unique patients in our retrospective study. Despite increasing awareness, delirium remains a frequently undocumented or miscoded entity. Implementing a validated, universal screening tool for delirium may improve identification and clinical outcomes. Patients who received opioids and benzodiazepines were more likely to have documentation of delirium than those patients that did not receive these medications. Further prospective research is also necessary to determine the true incidence of delirium, associated risk factors and establish best practices for treatment and prevention of delirium within the pediatric cancer and bone marrow transplant population.
Funding
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors. Egidio Del Fabbro, M.D., receives research support from R01AG061558 and R01CA225701.
Conflict of interest
There are no conflicts of interest.
Appendix
ICD codes used to identify patients
