Pediatric Clinical Classification System for Use in Inpatient Settings

Abstract

This cross-sectional study assesses the specificity of the Pediatric Clinical Classification System and Healthcare Cost and Utilization Project diagnosis categories for clinical classification in pediatric inpatient settings.

The transition from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) to the International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) system in 2015 in the United States limits the use of previously developed ICD-9-CM clinical classification systems to evaluate health care use, costs, variation, and trends. The ICD-10-CM Healthcare Cost and Utilization Project (HCUP) Clinical Classifications Software (CCS)¹ aggregates more than 70 000 ICD-10-CM diagnosis codes into categories. However, it lacks categories for important inpatient pediatric conditions (eg, bronchiolitis, mental health²), thereby limiting its use. Therefore, we developed the open source Pediatric Clinical Classification System (PECCS), which categorizes all ICD-10-CM diagnosis codes into mutually exclusive, clinically meaningful conditions for use in inpatient settings. To assess the ability of the PECCS to identify pediatric categories, we compared detection of conditions in the PECCS with the HCUP-CCS (Agency for Healthcare Research and Quality) using a national pediatric hospitalization database.

Methods

In this cross-sectional study, all 283 HCUP-CCS diagnosis categories were reviewed manually and were either preserved or further subdivided into clinically meaningful pediatric categories (Table 1). Categories and diagnosis codes present in the ICD-9-CM pediatric diagnosis code grouper, developed by Keren et al,³ served as a reference for the clinically meaningful pediatric categories that needed to be included in the PECCS. This resource, along with manual review, guided decisions to preserve or subdivide the HCUP-CCS categories. All ICD-10-CM diagnosis codes within each CCS category were independently reviewed by one of us (M.R.A.) and then by 2 others (P.J.G. and S.M). The category list was finalized over multiple meetings, and all disagreements were resolved through consensus. The Children’s Hospital Association⁴ provides the full set of PECCS codes. Ethical approval for the study was obtained from The Hospital for Sick Children, which also waived the need for informed patient consent as the data were anonymized before analysis.

Table 1. Examples of Increased Specificity of Categories for Pediatric Use With PECCS Compared With HCUP-CCS^a.

HCUP clinical classification systemb		PECCS
Category name	No. of encounters	Category name	No. (%) of encountersc
Miscellaneous mental health disorders	13 261	Miscellaneous mental health disorders	5357 (40.4)
		Anorexia nervosa	4709 (35.5)
		Conversion disorder	2979 (22.5)
		Bulimia nervosa	216 (1.6)
SLE and connective tissue disorder	15 108	Kawasaki disease	7796 (51.6)
		SLE and connective tissue disorder	5186 (34.3)
		Juvenile dermatomyositis	1719 (11.4)
		Dermatopolymyositis	407 (2.7)
Intestinal obstruction without hernia	21 092	Impaction of intestine	6354 (30.1)
		Intussusception	6183 (29.3)
		Intestinal obstruction without hernia	5535 (26.2)
		Intestinal or peritoneal adhesions with obstruction	2172 (10.3)
		Volvulus	701 (3.3)
		Paralytic ileus	147 (0.7)
Esophageal disorders	85 124	Gastroesophageal reflux and esophagitis	41 334 (48.6)
		Eosinophilic esophagitis	32 144 (37.8)
		Stricture and stenosis of esophagus	7411 (8.7)
		Esophageal disorders	4139 (4.9)
		Barrett esophagus with or without dysplasia	96 (0.1)

Abbreviations: HCUP, Healthcare Cost and Utilization Project; HCUP-CCS, HCUP Clinical Classifications Software; PECCS, Pediatric Clinical Classification System; SLE, systemic lupus erythematosus.

^aData are from 45 US hospitals participating in the Pediatric Health Information System between January 1, 2016, and December 31, 2019.

^bHCUP-CCS beta version 2019.1 software was used for analysis.

^cThe data reflect the percentage of encounters for each PECCS category within the corresponding HCUP-CCS category.

To compare the PECCS with the existing HCUP-CCS, both were applied to data from the Pediatric Health Information System, an administrative database containing hospitalization data from 45 US children’s hospitals.⁵ The sample included children younger than 18 years with inpatient, ambulatory surgery, or observation unit encounters at a hospital participating in the Pediatric Health Information System hospital between January 1, 2016, and December 31, 2019. To illustrate the improved utility of PECSS for identifying important pediatric categories, Table 1 presents select categories from the PECCS and HCUP-CCS with encounter frequencies. Table 2 lists the 10 most prevalent PECCS categories at Pediatric Health Information System hospitals. Data were analyzed using SAS version 9.4 (SAS Institute Inc).

Table 2. Top 10 Clinical Conditions Identified by Applying PECCS to Administrative Data From 45 US Children’s Hospitals^a.

Rank	PECCS category	Encounters, No. (%) (N = 5 041 752)b
1	Otitis media	275 418 (5.5)
2	Hypertrophy of tonsils and adenoids	257 996 (5.1)
3	Dental caries	180 472 (3.6)
4	Acute bronchiolitis	143 503 (2.9)
5	Asthma	119 654 (2.4)
6	Chemotherapy	93 296 (1.9)
7	Pneumonia	84 474 (1.7)
8	Respiratory failure, insufficiency, or arrest	80 285 (1.6)
9	Redundant prepuce and phimosis	63 461 (1.3)
10	Seizures with and without intractable epilepsy	62 142 (1.2)

Abbreviation: PECCS, Pediatric Clinical Classification System.

^aData are from 45 US hospitals participating in the Pediatric Health Information System between January 1, 2016, and December 31, 2019.

^bPercentages do not total 100% because only the top 10 conditions are presented.

Results

The final classification system categorized all 72 446 ICD-10-CM diagnosis codes into 834 clinically distinctive categories. Of the 283 groups outlined in the current HCUP-CCS, 177 (62.5%) were maintained and 106 (37.5%) were revised to create 657 categories. Of the 5 041 752 hospital encounters studied, 4 159 642 (82.5%) were classified into these revised PECCS categories.

Table 1 presents 4 selected conditions in HCUP-CCS, which correspond to 19 conditions in the PECCS, demonstrating the specificity of the PECCS as a classification system for pediatric health conditions. For example, there were 13 261 encounters of miscellaneous mental health disorders using the HCUP-CCS. However, the PECCS provides 3 additional specific and clinically meaningful categories: anorexia nervosa, conversion disorder, and bulimia nervosa.

Table 2 presents the 10 most prevalent conditions at Pediatric Health Information System hospitals identified by using the PECCS. If the HCUP-CCS were used, the categories bronchiolitis and redundant prepuce and phimosis would have been identified as broader, nonpediatric specific categories of bronchitis and other male genital disorders, respectively.

Discussion

The PECCS builds capacity in child health research by providing an open-source classification system to identify clinically meaningful, pediatric categories in inpatient settings using ICD-10-CM administrative data. The HCUP-CCS does not identify several important pediatric conditions owing to its broad categories and adult focus. The PECCS can be used for prioritization of comparative effectiveness research, understanding health services use and trends, and outcomes research. A limitation of the PECCS is that it does not differentiate between acute and chronic conditions. However, it can still be used with different data sources, in different pediatric settings, and by researchers in diverse pediatric fields. The PECCS can be modified for different country-specific ICD-10 versions to be used internationally. The PECCS will be updated annually to coincide with future ICD-10-CM revisions.