The Occurrence of Delirium Diagnosis among Youth Hospitalizations in the United States: a Kids’ Inpatient Database Analysis

Abstract

Objectives:

Delirium is an acute neuropsychiatric condition associated with increased morbidity and mortality. There is increasing recognition of delirium as a substantial health burden in younger patients, although few studies have characterized its occurrence. This study analyzes the occurrence of delirium diagnosis, its comorbidities, and cost among youth hospitalized in the United States.

Methods:

The Kids’ Inpatient Database, a national all-payors sample of pediatric hospitalizations in general hospitals, was examined for the year 2019. Hospitalizations with a discharge diagnosis of delirium among patients aged 1 to 20 years were included in the analysis.

Results:

Delirium was diagnosed in 43,138 hospitalizations (95% CI: 41,170 to 45,106), or 2.3% of studied hospitalizations. Delirium was diagnosed in a broad range of illnesses, with suicide and self-inflicted injury as the most common primary discharge diagnosis among patients with delirium. In hospital mortality was seven times greater in hospitalizations caring a delirium diagnosis. The diagnosis of delirium was associated with an adjusted increased hospital cost of $8,648 per hospitalization, or $373 million in aggregate cost.

Conclusions:

Based on a large national claims database, delirium was diagnosed in youth at a lower rate than expected based on prospective studies. The relative absence of delirium diagnosis in claims data may reflect underdiagnosis, a failure to code, and/or a lower rate of delirium in general hospitals compared to other settings. Further research is needed to better characterize the incidence and prevalence of delirium in young people in the hospital setting.

Introduction

Delirium is a neuropsychiatric syndrome characterized by acute changes in attention, awareness, and cognition secondary to an underlying illness.^1,2 Because of its presence in diverse hospital settings and its status as a core clinical competency of consult-liaison psychiatry,^3,4 delirium is a topic of burgeoning study in general hospital settings,⁵ specialized cohorts including the critically ill,⁶ and those hospitalized with COVID-19.^7,8 Although the majority of delirium-associated literature is based in adults,⁹ pediatric delirium is a long-recognized distinct entity with particular significance in the pediatric ICU (PICU).¹⁰ Delirium is associated with increases in mortality,¹¹ prolongation of hospital stay,¹² and worse cognitive outcomes¹³ in adults. In critically ill children, delirium is also associated with extended duration of mechanical ventilation and prolonged ICU stay,¹⁴ with a large-single center cohort study of PICU patients showing that delirium is associated with 20% increased length of critical care.¹⁵ Functional outcomes are also impacted poorly by delirium in children: studies demonstrate decline in cognitive status¹⁵ and quality-of-life scores¹⁶ associated with delirium during admission. Healthcare costs are also significantly impacted by delirium in adults and children. One estimate attributes the cost of delirium in adults to greater than $150 billion per year,¹⁷ and in critically ill children delirium was associated with an 85% increase in cost of hospitalization.¹⁸

Estimations of delirium incidence in hospitalized adult populations vary widely. In prospective assessment, delirium occurs in approximately 1 in 5 hospitalized adults.¹⁹ In contrast, the pool rate of diagnosis appears to be 5% of hospitalized patients, with variation by treating department of several orders of magnitude.⁵ Estimates in pediatric populations have focused primarily on critically ill patients, and range from 20% of patients in the general pediatric ICU¹⁴ to 50–60% in surgical²⁰ or cardiothoracic²¹ pediatric ICUs, with a recent meta-analysis calculating a pooled delirium prevalence of 34% in the PICU setting.²² These estimates may additionally be impacted by a noted tendency towards underreporting of delirium in claims data.²³

Efforts to mitigate the impact of delirium on health and healthcare costs through prevention been encouraging.²⁴ Multiple validated scoring approaches exist for identification of pediatric delirium;^25–28 however, given the existences of multi component prevention,^29,30 and the absence of robust treatments,^31,32 population level risk stratification is needed to allocate scarce prevention resources. As in adults, the pediatric literature shows a significant burden outside of the ICU and those who are critically ill raising the possibility of substantial missed prevention opportunities.⁵ This study characterizes the occurrence of delirium diagnosis in hospitalized pediatric patients using nationally representative claims data to elucidate clinical impacts, case distribution, and potential diagnostic shortfalls.²³

Methods

Data Source

This study utilized the 2019 version of the Kids’ Inpatient Database (KID), produced by the Healthcare Cost and Utilization Project (HCUP) of the Agency for Healthcare Research and Quality. The KID provides administrative data on pediatric hospitalizations (defined as individuals aged 20 years or younger) in the United States, sampling 80% of non-newborn discharges from 3,998 non-federal acute care hospitals in 49 states. This database is produced every three years, with the 2019 data being the most recent. Data within this claims database includes patient demographics, information on up to 40 discharge diagnoses per admission, the total hospital cost, and length of stay.

Sample

Due to challenges in diagnosing delirium in the youngest patients, and clinical consensus that delirium diagnoses beginning at the age of 12 months are valid,²⁵ this study included all hospital encounters from the 2019 KID involving patients aged 1 to 20 years.

Case Selection

Delirium discharges were identified based on the Procedure Coding System from the International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM/PCS). In order to make this pediatric analysis as comparable as possible to prior studies utilizing billing records to assess the incidence of delirium in adults,^5,33–35 delirium was defined based off the ICD-10-CM codes F01.51 (vascular dementia with behavioral disturbance), F03.90 (unspecified dementia without behavioral disturbance), F05 (delirium due to known physiological condition), F06.0 (psychotic disorder with hallucinations due to known physiological condition), F06.1 (catatonic disorder due to known physiological condition), F06.2 (psychotic disorder with delusions due to known physiological condition), F06.30 (mood disorder due to known physiological condition, unspecified), F06.4 (anxiety disorder due to known physiological condition), F06.8 (other specified mental disorders due to known physiological condition), F10.231 (alcohol dependence with withdrawal delirium), F15.920 (other stimulant use, unspecified with intoxication, uncomplicated), F19.921 (other psychoactive substance use, unspecified with intoxication with delirium), F19.939 (other psychoactive substance use, unspecified with withdrawal, unspecified), F19.950 (other psychoactive substance use, unspecified with psychoactive substance-induced psychotic disorder with delusions), F19.951 (other psychoactive substance use, unspecified with psychoactive substance-induced psychotic disorder with hallucinations), F19.97 (other psychoactive substance use, unspecified with psychoactive substance-induced persisting dementia), G92 (toxic encephalopathy), G93.40 (encephalopathy, unspecified), G93.41 (metabolic encephalopathy), G93.49 (other encephalopathy), R40.0 (somnolence), R40.4 (transient alteration of awareness), and R41.82 (altered mental status, unspecified). Hospitalizations were included in the sample if any of these codes was among the discharge diagnosis list. Primary hospital diagnoses were identified from the discharge diagnoses, and classified according to the Pediatric Clinical Classification System (PECCS),³⁶ which groups the 72,446 ICD-10-CM diagnosis codes into 834 clinically distinctive categories for pediatric medical conditions. Hospitalizations were classified as primary delirium if one of the delirium codes was the first discharge diagnosis, while those with delirium code(s) not listed as the first discharge diagnosis were defined as secondary delirium hospitalization.

Statistical Analysis

The survey design of the KID, in which hospitals are stratified and then sampled within strata, produces variance around all reported values for both continuous and categorical variables. This variance is presented here for the total number of hospitalizations, while all other variables are given as weighted point estimates. Due to the non-normal distribution of age, length of stay and total hospital charges, these values are reported as medians with interquartile range (IQR). Differences between groups for categorical variables were assessed using the χ² test, and differences between groups for continuous variables were assessed using the two-sided Mann-Whitney U Test. All analyses were conducted on data weighted according to the appropriate KID discharge weight to obtain national estimates.

For the primary statistical analysis, the log-transformed total hospital charges were modeled using a general linear model incorporating the sampling methodology of the KID, with delirium diagnosis (yes or no), age (z score), sex, hospital region, admission type (elective vs. non-elective), primary service line, injury status, whether the admission involved a major operative procedure, primary payor, and the log-transformed sum of total length of stay plus 1 as descriptor variables. As a supplemental analysis, a logistic regression was conducted on the outcome of a primary or secondary diagnosis of delirium, with age (z score), sex, hospital region, admission type (elective vs. non-elective), primary service line, injury status, whether the admission involved a major operative procedure, and primary payor as descriptor variables.

Due to the de-identified nature of the KID, this study was determined to be Not Human Subjects Research by the Mass General Brigham Institutional Review Board. This study is reported in accordance with the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement.³⁷ Analyses were conducted using SPSS (version 25; IBM Software, Inc, Armonk, NY).

Results

In the 2019 KID, 43,138 hospitalizations (95% CI: 41,170 to 45,106) among patients aged 1 to 20 years involved a diagnosis of delirium compared to 1,833,685 (95% CI:1,781,218 to 1,886,151) hospitalizations without a delirium diagnosis, for an overall case rate of 2.3%. Partial demographic information for hospitalizations with and without delirium are given in Table 1, with further details in Table S1. The specific billing codes identifying hospitalizations as involving delirium are listed in Table S2. Hospitalizations with and without delirium were significantly different in all categories at the level of p<0.001. There were more hospitalizations with delirium among males (22,930; 53.2%) than females (20,198; 46.8%), whereas 58.0% of hospitalizations without delirium involved female patients. The median age for hospitalizations with delirium was 14, with an IQR of 7 to 18 years (Figure 1). Most delirium-associated hospitalizations (36,482; 84.6%) were non-elective, and 74.6% (32,164) did not involve a transfer from another healthcare facility. A total of 888 hospitalizations with delirium (2.1%) concluded in death compared to 0.3% of hospitalizations without delirium. Hospitalizations with delirium involved a median hospital length of stay of 3 days (IQR 2 to 8 days), and median hospital charges of $38,494 (IQR $18,744 to $101,475).

Table 1:

demographics of hospitalizations with and without a discharge diagnosis of delirium in the 2019 KID. The two groups differ significantly in all categorical and continuous variables at the level of p<0.001.

	Delirium Diagnosis		No Delirium Diagnosis
	n	%	n	%
N	43,138 (41,170 to 45,106)		1,833,685 (1,781,218 to 1,886,151)
Age (yrs), median, IQR	14 (7 to 18)		14 (7 to 18)
<10	14,407	33.4	602,601	32.9
10–16	13,644	31.6	531,196	29.0
17–20	15,087	35.0	699,888	38.2
Sex
Male	22,930	53.2	769,552	42
Female	20,198	46.8	1,063,859	58
Hospital Region
Northeast	6,669	15.5	302,230	16.5
Midwest	10,737	24.9	411,410	22.4
South	15,945	37	731,305	39.9
West	9,788	22.7	388,739	21.2
Admission Type
Elective	6,582	15.3	406,382	22.2
Non-Elective	36,482	84.6	1,423,577	77.6
Primary Service Line
Maternal and Neonatal	217	0.5	319,282	17.4
Mental Health/Substance Use	5,446	12.6	276,199	15.1
Injury	5,898	13.7	114,546	6.2
Surgical	4,351	10.1	244,167	13.3
Medical	27,226	63.1	879,490	48
Injury Status
No Injury Diagnosis	30,180	70	1,639,497	89.4
Injury Diagnosis is Primary	10,336	24	137,062	7.5
Injury Diagnosis, non-Primary	2,623	6.1	57,126	3.1
Operating Room
No Major Procedure on Discharge Record	37,848	87.7	1,409,095	76.8
Major OR Procedure on Discharge Record	5,290	12.3	424,590	23.2
Primary Payor
Medicare	217	0.5	8,518	0.5
Medicaid	23,224	53.8	983,783	53.7
Private Insurance	16,001	37.1	694,761	37.9
Self Pay	1,908	4.4	69,965	3.8
No charge	72	0.2	3,316	0.2
Other	1,628	3.8	70,611	3.9
Died	888	2.1	5,477	0.3
Hospital Length of Stay, days (median, IQR)	3 (2 to 8)		3 (2 to 4)
Total Charges, $ (median, IQR)	$38,494 ($18,744 to $101,475)		$22,257 ($12,562 to $43,656)

Figure 1:

Percentage of overall hospitalizations that involve a diagnosis of delirium vs. the age of the patient. While the absolute number of hospitalizations and delirium cases varies by age, the proportion of hospitalizations with a delirium diagnosis is in the range of 1.8% to 2.8% for all ages.

In the majority of cases, delirium was identified as a secondary discharge diagnosis (38,322; 88.9% vs 4,806; 11.1%). When delirium was the primary diagnosis, the hospitalizations involved younger patients (median age 11; IQR 5 to 16) than secondary cases (median age 14; IQR 7 to 18; Mann-Whitney U, U=4.23×10⁷, P < 0.001), were more frequently elective admissions (36.6% vs. 12.6%), rarely included a major surgical procedure (0.9% vs. 13.7%), and had a larger proportion of mental health and substance abuse diagnoses (47.8% vs. 8.2%). Median charges were lower in hospitalizations with delirium as a primary discharge diagnosis compared to those with delirium as a secondary diagnosis (median of $22,257 for primary delirium diagnosis vs. $42,164 for secondary diagnosis; Mann-Whitney U, U=3.43×10⁷, P < 0.001). Demographics for hospitalizations involving delirium as the primary discharge diagnosis compared to those with delirium as a secondary discharge diagnosis are given in Table 2, with additional detail in Table S3. In a logistic regression on the outcome of primary vs. secondary diagnosis of delirium, male sex and the lack of an injury diagnosis were associated with a higher odds of a primary delirium diagnosis, while age, non-elective admission, and injury diagnosis were associated with a higher odds of a secondary delirium diagnosis (Table S4).

Table 2:

demographics of hospitalizations for which delirium is listed as the principal discharge diagnosis vs. those for which delirium is a secondary discharge diagnosis. The two groups differ significantly in all categorical and continuous variables at the level of p<0.001.

	primary delirium diagnosis		secondary delirium diagnosis
	n	%	n	%
N	4,806 (4,546 to 5,066)		38,332 (36,706 to 39,959)
Age (yrs), median, IQR	11 (5 to 16)		14 (7 to 18)
<10	2,128	44.3	12,278	32.0
10–16	1,636	34.0	12,009	31.3
17–20	1,041	21.7	14,046	36.6
Sex
Male	2,672	55.6	20,257	52.8
Female	2,133	44.4	18,064	47.1
Hospital Region
Northeast	950	19.8	5,719	14.9
Midwest	1,307	27.2	9,429	24.6
South	1,487	30.9	14,457	37.7
West	1,062	22.1	8,727	22.8
Admission Type
Elective	1,759	36.6	4,822	12.6
Non-Elective	3,038	63.2	33,444	87.2
Primary Service Line
Maternal and Neonatal	0	0	217	0.6
Mental Health/Substance Use	2,295	47.8	3,151	8.2
Injury	0	0	5,898	15.4
Surgical	39	0.8	4,312	11.2
Medical	2,472	51.4	24,754	64.6
Injury Status
No Injury Diagnosis	4,571	95.1	25,609	66.8
Injury Diagnosis is Primary	0	0	10,336	27
Injury Diagnosis, non-Primary	235	4.9	2,387	6.2
Operating Room
No Major Procedure on Discharge Record	4,762	99.1	33,087	86.3
Major OR Procedure on Discharge Record	44	0.9	5,246	13.7
Primary Payor
Medicare	18	0.4	199	0.5
Medicaid	2,454	51.1	20,770	54.2
Private Insurance	1,973	41.1	14,028	36.6
Self Pay	163	3.4	1,745	4.6
No charge	<11	<0.2	64	0.2
Other	177	3.7	1,450	3.8
Died	<11	<0.2	879	2.3
Hospital Length of Stay, days (median, IQR)	2 (1 to 3)		4 (2 to 8)
Total Charges, $ (median, IQR)	$23,412 ($12,113 to $40,289)		$42,164 ($19,913 to $114,370)

Diagnostically, the most common principal discharge diagnoses (grouped by PECCS categories) among hospitalizations involving delirium were suicide and intentional self-inflicted injury (4,413; 10.3%), transient alteration of awareness (3,190; 7.5%), seizures (2,467; 5.8%), sepsis (2,290; 5.3%), and respiratory failure (2,125; 5.0%). Together these top 5 categories represented 33.9% of hospitalizations with delirium (Table 3). Stratifying primary discharge diagnoses by age shows different diagnoses at different developmental stages. In children younger than 10 years, suicide and self-injury is not among the top 10 diagnostic categories, although two of the top 10 involve poisoning by medications (5.4%) or psychotropics (2%). In contrast, suicide and self-injury is the top diagnostic category of patients aged 10–16 years (15.7%) and 17–20 years (15.8%), with poisoning by medication also in the top 10 diagnoses for these adolescents and young adults. A full list of primary diagnostic categories is listed in Table 3, with individual diagnostic codes listed in Table S5.

Table 3:

Primary discharge diagnosis (grouped by PECCS categories) for hospitalizations involving delirium. The top 15 diagnostic categories are listed. Diagnosis are given overall, and for the age brackets of 1–9, 10–16, and 17–20.

All ages (n=43,138)	n	%
Suicide and intentional self-inflicted injury	4,413	10.3%
Transient alteration of awareness	3,190	7.5%
Seizures w and w/o intractable epilepsy	2,467	5.8%
Septicemia (except in labor)	2,290	5.3%
Respiratory failure; insufficiency; arrest	2,125	5.0%
Poisoning by other medications and drugs	1,681	3.9%
Other nervous system disorders	1,520	3.6%
Diabetic ketoacidosis	1,514	3.5%
Substance-related disorders	1,236	2.9%
Poisoning by psychotropic agents	1,031	2.4%
Partial epilepsy w with w/o intractable epilepsy	957	2.2%
Epilepsy; convulsions	783	1.8%
Pneumonia	704	1.6%
Mood disorders (major depressive disorder)	682	1.6%
Intracranial injury	672	1.6%
Ages 1 to 9 (n=14,407)
Transient alteration of awareness	1,742	12.3%
Seizures w and w/o intractable epilepsy	1,318	9.3%
Respiratory failure; insufficiency; arrest	1,010	7.1%
Poisoning by other medications and drugs	762	5.4%
Septicemia (except in labor)	690	4.9%
Partial epilepsy w with w/o intractable epilepsy	470	3.3%
Other nervous system disorders	462	3.3%
Epilepsy; convulsions	373	2.6%
Pneumonia	357	2.5%
Poisoning by psychotropic agents	278	2.0%
Substance-related disorders	274	1.9%
Other convulsions	271	1.9%
Encephalitis (except that caused by tuberculosis or sexually transmitted disease)	233	1.6%
Diabetic ketoacidosis	176	1.2%
Infantile spasms w/o intractable epilepsy	173	1.2%
Ages 10 to 16 (n=13,644)
Suicide and intentional self-inflicted injury	1,933	15.7%
Transient alteration of awareness	971	7.9%
Seizures w and w/o intractable epilepsy	548	4.4%
Septicemia (except in labor)	521	4.2%
Respiratory failure; insufficiency; arrest	515	4.2%
Other nervous system disorders	453	3.7%
Diabetic ketoacidosis	447	3.6%
Poisoning by other medications and drugs	314	2.5%
Mood disorders (major depressive disorder)	309	2.5%
Partial epilepsy w with w/o intractable epilepsy	291	2.4%
Epilepsy; convulsions	229	1.9%
Encephalitis (except that caused by tuberculosis or sexually transmitted disease)	199	1.6%
Substance-related disorders	197	1.6%
Poisoning by psychotropic agents	177	1.4%
Schizophrenia and other psychotic disorders	176	1.4%
Ages 17–20 (n=15,087)
Suicide and intentional self-inflicted injury	1,931	15.8%
Septicemia (except in labor)	904	7.4%
Diabetic ketoacidosis	779	6.4%
Substance-related disorders	667	5.5%
Poisoning by other medications and drugs	510	4.2%
Poisoning by psychotropic agents	480	3.9%
Other nervous system disorders	464	3.8%
Respiratory failure; insufficiency; arrest	457	3.7%
Seizures w and w/o intractable epilepsy	436	3.6%
Schizophrenia and other psychotic disorders	362	3.0%
Alcohol-related disorders	280	2.3%
Intracranial injury	279	2.3%
Mood disorders (major depressive disorder)	274	2.2%
Transient alteration of awareness	235	1.9%
Mood disorders (bipolar disorder)	171	1.4%

In order to explore the association between delirium diagnosis and overall cost of care, the log-transformed overall hospital charges were modeled using a general linear model incorporating the sampling methodology of the KID. In this model, delirium diagnosis was associated with an adjusted increased hospital cost of $8,648 per hospitalization, or $373 million overall given the 43,138 observed cases. Older age, male sex, elective admission, lack of injury diagnosis, presence of a major surgical procedure, and hospital length of stay were all independently associated with higher overall hospital charge, while hospital region, primary service line, and primary payor were not associated with a different hospital cost (Table S6).

Discussion

This study reports a nationally representative analysis of the diagnosis of delirium among children and adolescents in the general hospital setting. Delirium was uncommonly diagnosed among pediatric hospitalizations in 2019, with delirium codes present among the discharge diagnosis list for 43,138 hospitalizations (95% CI: 41,170 to 45,106), or 2.3% of the overall hospitalizations for patients aged 1 to 20 years. While delirium occurrence varies widely by population, the observed diagnostic coding rate is far lower than prevalence of delirium found in prior cross-sectional and retrospective studies of specific pediatric populations. These include 8.2% of patients seen by a pediatric psychiatry consult-liaison service,³⁸ a pooled prevalence of 34% in the pediatric intensive care unit (PICU) setting based on a systematic review,²² and 40–66% among pediatric cardiothoracic ICU patients.²² Notably, this pooled delirium prevalence among PICU patients is similar to the 31% found in a systematic review of adult ICU patients.³⁹ The observed rate in this study, however, is similar to prior studies in adults claims data: one prior study looking at billing data from a single state found that 2.1% of adult hospitalizations contained a billing code for delirium, whereas the literature prevalence from prior prospective studies and meta-analysis found a prevalence of 23.6%.²³

This difference in delirium rates among research studies and billing claims could reflect a failure to diagnose, a failure to code, or a difference in incidence rates in the community practice settings well represented in claims data but often not present in research studies. As the KID does not differentiate between hospitalizations involving ICU admission compared to those without, one explanation for the lower rate of delirium observed here is due to a lower prevalence in a mixed hospital sample vs. prior ICU samples. However, given an estimated 250,000 PICU admissions annually in the US,¹⁸ if the 34% delirium prevalence among PICU hospitalizations (based on pooled meta-analysis)²² was extrapolated nationally, this alone would represent nearly twice as many delirium cases as identified in this study, unless prior studies were systematically biased towards institutions with higher than average delirium incidence. As a result, our study suggests that there is likely significant under recognition of delirium or failure to code for identified cases.

One possibility is that delirium is clinically diagnosed at rates similar to the literature prevalence, but that appropriate billing codes corresponding to the clinical diagnoses made are not submitted. This could be a side effect of bundled billing practices or minimal financial incentive to make secondary or associated diagnoses. One study using natural language processing of the free text of clinical notes in adults suggest that delirium is being clinically recognized up to twice as often as it is billed for,³⁴ and undercoding may be more common for mild or more subacute presentations.³³ Systematic quality improvement interventions can increase the accuracy of delirium billing codes,⁴⁰ but these are not typical in routine practice. A complimentary possibility is that delirium is being underrecognized in routine clinical practice. Multiple delirium rating scales exist for pediatric patients, with sensitivities varying among patient age and ability to participate in clinical examination.^25–28 Prospective and cross-sectional studies in the PICU population have found that delirium identification is poor without systematic assessment,⁴¹ and that identification of delirium in pediatric oncology patients increased five-fold with systematic screening.⁴² As a result, 2016 European critical care guidelines now recommend universal screening for delirium in PICU patients,⁴³ but it is unclear whether this has been undertaken in hospitals included in the 2019 KID. Moreover, it is unclear if screening tools designed for PICU use are generalizable to non-critical care settings, and so further research is needed into tools for optimal delirium recognition across a range of hospital settings.

Hospitalizations which include delirium diagnosis were associated with increased mortality and cost. Mortality for delirium cases was seven times higher than for non-delirium hospitalizations (2.1% vs. 0.3%), and involved more long hospitalization (75% LOS 8 days for delirium cases vs. 4 days for non-delirium), although the median length of hospitalization was the same at 3 days. The $8,648 of additional cost associated with a delirium diagnosis in this sample is less than the $14,000 cost seen in the PICU setting.¹⁸ Given the far higher number of patients identified in our sample (464 in the PICU study vs 43,138 here) the adjusted additional aggregate cost of $373 million for the delirium cases identified here represents a substantial area of healthcare spending. Importantly this is cost within the episode of care, not the one year cost or other cumulative cost estimates which are not possible in the KID, but in adult studies are much larger than the cost of the index hospitalization.⁴⁴ Notably, however, these cost estimates, while adjusted for demographics, in-hospital procedure utilization, and length of stay, do not control for other diagnostic and procedural differences that may be present between hospitalizations with and without delirium, which may confound analysis of the cost of delirium itself.

Hospitalizations with delirium involved a diverse range of organ system dysfunction, including intentional self-injury, seizures, sepsis, respiratory failure, diabetic ketoacidosis, and injuries. The principal discharge diagnosis among patients with a delirium diagnoses varied based on the age of the patient, but notably intentional and non-intentional ingestions and poisonings caused a substantial proportion of delirium cases. Some of this may be due to the specific billing codes used in this study, which includes multiple codes for encephalopathy including “toxic encephalopathy.” At present the literature on delirium and acute encephalopathy are largely independent, although 2020 consensus guidelines recommend against the use of many encephalopathy terms (e.g. “acute confusional state”) as lacking construct validity.⁴⁵ Thus the altered mental status that may occur following certain ingestions (which may meet DSM-5 criteria for delirium) may be diagnosed as G92 “toxic encephalopathy,” thereby classifying as delirium for this study. Given increasing rates of youth suicide in the US,⁴⁶ with ingestion of prescribed and over the counter medications as a common means of suicide attempt,⁴⁷ the inclusion of pediatric consult liaison psychiatrists along with the primary medical or surgical teams may enhance the care of patients with delirium.⁴⁸

Strengths and limitations of this study both derive from its utilization of large-scale heath claims data. The design of the KID, including comprehensive coverage of nearly all general hospitals serving pediatric patients in the US, allows nationally-representative analysis of delirium claims. This minimizes bias that may come from data sources deriving from a single healthcare system, region, or payment source. This reliance on claims data, however, means this analysis is necessarily limited to the billing for delirium, as opposed to a case definition rooted in structured active screening and diagnosis. As described above, both the diagnosis and coding for delirium are known to be limited in the adult delirium literature, and it is likely that these same constraints apply to younger patients, and so this study likely underestimates the true burden of delirium in youth. Moreover, the choice of specific codes used to define delirium may impact results. While there is only a single ICD-10 code titled delirium (F05), prior research in adult patients has identified a broader set of codes that are, in practice, diagnosed in adult delirium patients. The codes used in this study matched those in the adult literature in order to allow for comparisons between the age ranges, however this may introduce error as these codes have not been specifically validated in the pediatric population. Some, including F01.51 (vascular dementia with behavioral disturbance) that was diagnosed in 33 patients in the KID, may lack face validity in this population. Further research is needed to define codes associated with delirium in the pediatric population, which will likely require cross-sectional studies utilizing the free text of clinical notes or delirium assessment by expert raters. Moreover, as this is a study of hospitalized patients, it is unable to comment on delirium diagnosis among patients who are not ultimately admitted to the hospital. This may particularly undercount emergence delirium, a phenomenon that occurs most commonly in preschool-aged children in the early postanesthetic period.⁴⁹ As many common pediatric surgeries are day procedures not involving admission, such cases would not be counted in this dataset. Furthermore, since the KID samples hospitalizations and not individual patients, it is possible that some individuals may be counted more than once in the database for distinct hospitalizations (at the same or different healthcare facilities), which may bias the demographic information presented here. Finally, as the KID only samples general hospitals, it does not provide information about delirium diagnosis at alternative facilities, such as long-term rehabilitation hospitals or federally-operated hospitals.

Conclusions

Billing codes for delirium were recorded for 43,138 hospitalizations (95% CI: 41,170 to 45,106), or 2.3% of the overall hospitalizations for patients aged 1 to 20 years in 2019 in the United States. Delirium diagnosis was made in patients across the age span of pediatrics, and was associated with seven-fold increased mortality and adjusted additional healthcare costs of $373 million dollars. Billing codes for delirium were present at a rate lower than expected from prior studies of active delirium assessment, suggesting that gaps may remain in the identification of this serious pediatric neuropsychiatric condition.

Significant Outcomes:

• Delirium billing codes were present for 2.3% of hospitalizations for patients aged 1 to 20 years in general hospitals in the United States, for a total of 43,138 hospitalizations in 2019

• Suicide and self-inflicted injury was the most common primary discharge diagnosis among hospitalizations involving a delirium diagnosis

• Delirium diagnosis was associated with seven-fold increased in-hospital mortality relative to hospitalizations without a delirium diagnosis, and $373 million in aggregate increased cost

Limitations:

• Analysis is based on claims data for delirium diagnosis, rather than systematic, prospective assessment of delirium

• Low occurrence of delirium diagnosis may reflect under recognition of delirium, a failure to code recognized cases, and/or actual differences in rate, but claims data cannot distinguish among these possibilities

Data Availability Statement:

Publicly available datasets were analyzed in this study. The dataset analyzed for this study can be obtained from the Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality at: https://www.hcup-us.ahrq.gov/db/nation/kid/kiddbdocumentation.jsp