Pediatric Clinical Classification System for use in Canadian inpatient settings.

BACKGROUND: A classification system that categorizes International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) diagnosis codes into clinically meaningful categories is important for pediatric clinical and health services research using administrative data. While a Pediatric Clinical Classification System (PECCS) is available for the United States ICD-10 system (i.e, ICD-10-CM), differences in the ICD-10 system between countries limits PECCS use in Canada. OBJECTIVE: To translate PECCS from ICD-10-CM to ICD-10-CA for use in Canada (PECCS-CA), and examine the utility of PECCS-CA in administrative data of pediatric hospital encounters in Ontario, Canada. METHODS: PECCS was translated by mapping each ICD-10-CA code to its corresponding ICD-10-CM code, based on code description and alphanumeric code, using automated functions in Microsoft Excel. All unmatched ICD-10-CA codes were manually matched to an ICD-10-CM code. The ICD-10-CA codes were mapped to a PECCS category based on the placement of the corresponding ICD-10-CM code. Finally, in this cross-sectional study, the utility of PECCS-CA was examined in pediatric hospital encounters in children <18 years of age with an inpatient or same day surgery encounter, between April 1, 2014 to March 31, 2019 in Ontario. RESULTS: In total, 16,992 ICD-10-CA diagnosis codes were mapped to 781 mutually exclusive condition categories that included pediatric specific conditions and treatments in PECCS-CA. From the 781 categories, 777 (99.5%) were derived from the original PECCS, 3 (0.4%) from merging the original PECCS categories, and 1 (0.1%) was newly developed. The PECCS-CA was applied to health administrative data of 911,732 hospital encounters in children. The most prevalent condition in children was low birth weight (n = 54,100 encounters). CONCLUSION: The PECCS-CA is an open-source classification system which maps ICD-10-CA codes into 781 clinically important pediatric categories. The PECCS-CA can be used for pediatric health services and outcomes research in Canada.

Prevalence, Cost, and Variation in Cost of Pediatric Hospitalizations in Ontario, Canada.

Importance: Identifying conditions that could be prioritized for research based on health care system burden is important for developing a research agenda for the care of hospitalized children. However, existing prioritization studies are decades old or do not include data from both pediatric and general hospitals. Objective: To assess the prevalence, cost, and variation in cost of pediatric hospitalizations at all general and pediatric hospitals in Ontario, Canada, with the aim of identifying conditions that could be prioritized for future research. Design, Setting, and Participants: This population-based cross-sectional study used health administrative data from 165 general and pediatric hospitals in Ontario, Canada. Children younger than 18 years with an inpatient hospital encounter between April 1, 2014, and March 31, 2019, were included. Main Outcomes and Measures: Condition-specific prevalence, cost of pediatric hospitalizations, and condition-specific variation in cost per inpatient encounter across hospitals. Variation in cost was evaluated using (1) intraclass correlation coefficient (ICC) and (2) number of outlier hospitals. Costs were adjusted for inflation to 2018 US dollars. Results: Overall, 627 314 inpatient hospital encounters (44.8% among children younger than 30 days and 53.0% among boys) at 165 hospitals (157 general and 8 pediatric) costing $3.3 billion were identified. A total of 408 003 hospitalizations (65.0%) and $1.4 billion (43.8%) in total costs occurred at general hospitals. Among the 50 most prevalent and 50 most costly conditions (of 68 total conditions), the top 10 highest-cost conditions accounted for 55.5% of all costs and 48.6% of all encounters. The conditions with highest prevalence and cost included low birth weight (86.2 per 1000 encounters; $676.3 million), preterm newborn (38.0 per 1000 encounters; $137.4 million), major depressive disorder (20.7 per 1000 encounters; $78.3 million), pneumonia (27.3 per 1000 encounters; $71.6 million), other perinatal conditions (68.0 per 1000 encounters; $65.8 million), bronchiolitis (25.4 per 1000 encounters; $54.6 million), and neonatal hyperbilirubinemia (47.9 per 1000 encounters; $46.7 million). The highest variation in cost per encounter among the most costly medical conditions was observed for 2 mental health conditions (other mental health disorders [ICC, 0.28] and anxiety disorders [ICC, 0.19]) and 3 newborn conditions (intrauterine hypoxia and birth asphyxia [ICC, 0.27], other perinatal conditions [ICC, 0.17], and surfactant deficiency disorder [ICC, 0.17]). Conclusions and Relevance: This population-based cross-sectional study of hospitalized children identified several newborn and mental health conditions as having the highest prevalence, cost, and variation in cost across hospitals. Findings of this study can be used to develop a research agenda for the care of hospitalized children that includes general hospitals and to ultimately build a more substantial evidence base and improve patient outcomes.

Rates in Bronchiolitis Hospitalization, Intensive Care Unit Use, Mortality, and Costs From 2004 to 2018.

IMPORTANCE: Over the last 2 decades, bronchiolitis guidelines and improvement efforts focused on supportive care and reducing unnecessary tests, treatments, and hospitalization. There have been limited population-based studies examining hospitalization outcomes over time. OBJECTIVE: To describe rates and trends in bronchiolitis hospitalization, intensive care unit (ICU) use, mortality, and costs. DESIGN, SETTING, AND PARTICIPANTS: This cohort study used population-based health administrative data from April 1, 2004, to March 31, 2018, to identify bronchiolitis encounters using hospital discharge diagnosis codes in Ontario, Canada. Children younger than 2 years with and without bronchiolitis hospitalization were included. Data were analyzed from January 2020 to July 2021. MAIN OUTCOMES AND MEASURES: Bronchiolitis hospitalization per 1000 person-years, ICU use per 1000 hospitalizations, mortality per 100 000 person-years, and costs per 1000 person-years adjusted to 2018 Canadian dollars and reported in 2018 US dollars. RESULTS: Among 2 336 446 included children, 1 199 173 (51.3%) were male. During the study period, 43 993 children (1.9%) younger than 2 years had 48 058 bronchiolitis hospitalizations at 141 hospitals. Bronchiolitis accounted for 48 058 of 360 920 all-cause hospitalizations (13.3%) and 215 654 of 2 566 348 all-cause hospital days (8.4%) in children younger than 2 years. Bronchiolitis hospitalization was stable over time, at 14.0 (95% CI, 13.6-14.4) hospitalizations per 1000 person-years in 2004-2005 and 12.7 (95% CI, 12.2-13.1) hospitalizations per 1000 person-years in 2017-2018 (annual percent change [APC], 0%; 95% CI, -1.6 to 1.6; P = .97). ICU admission increased significantly from 38.1 (95% CI, 32.2-44.8) per 1000 hospitalizations in 2004-2005 to 87.8 (95% CI, 78.3-98.0) per 1000 hospitalizations in 2017-2018 (APC, 7.2%; 95% CI, 5.4-8.9; P < .001). Over the study period, bronchiolitis mortality was 2.8 (95% CI, 2.3-3.4) per 100 000 person-years and remained stable (APC, 1.1%; 95% CI, -8.4 to 11.7; P = .85). Hospitalization costs per 1000 person-years increased from $49 640 (95% CI, $49 617-$49 663) in 2004-2005 to $58 632 (95% CI, $58 608-$58 657) in 2017-2018 (APC, 3.0%; 95% CI, 1.3-4.8; P = .002). CONCLUSIONS AND RELEVANCE: From 2004 to 2018, bronchiolitis hospitalization and mortality rates remained stable; however, ICU use and costs increased substantially. This represents a major increase in high-intensity hospital care and costs for one of the most common and cumulatively expensive conditions in pediatric hospital care.

Identifying Conditions With High Prevalence, Cost, and Variation in Cost in US Children's Hospitals.

Importance: Identifying high priority pediatric conditions is important for setting a research agenda in hospital pediatrics that will benefit families, clinicians, and the health care system. However, the last such prioritization study was conducted more than a decade ago and used International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. Objectives: To identify conditions that should be prioritized for comparative effectiveness research based on prevalence, cost, and variation in cost of hospitalizations using contemporary data at US children's hospitals. Design, Setting, and Participants: This retrospective cohort study of children with hospital encounters used data from the Pediatric Health Information System database. Children younger than 18 years with inpatient hospital encounters at 45 tertiary care US children's hospitals between January 1, 2016, and December 31, 2019, were included. Data were analyzed from March 2020 to April 2021. Main Outcomes and Measures: The condition-specific prevalence and total standardized cost, the corresponding prevalence and cost ranks, and the variation in standardized cost per encounter across hospitals were analyzed. The variation in cost was assessed using the number of outlier hospitals and intraclass correlation coefficient. Results: There were 2 882 490 inpatient hospital encounters (median [interquartile range] age, 4 [1-12] years; 1 554 024 [53.9%] boys) included. Among the 50 most prevalent and 50 most costly conditions (total, 74 conditions), 49 (66.2%) were medical, 15 (20.3%) were surgical, and 10 (13.5%) were medical/surgical. The top 10 conditions by cost accounted for $12.4 billion of $33.4 billion total costs (37.4%) and 592 815 encounters (33.8% of all encounters). Of 74 conditions, 4 conditions had an intraclass correlation coefficient (ICC) of 0.30 or higher (ie, major depressive disorder: ICC, 0.49; type 1 diabetes with complications: ICC, 0.36; diabetic ketoacidosis: ICC, 0.33; acute appendicitis without peritonitis: ICC, 0.30), and 9 conditions had an ICC higher than 0.20 (scoliosis: ICC, 0.27; hypertrophy of tonsils and adenoids: ICC, 0.26; supracondylar fracture of humerus: ICC, 0.25; cleft lip and palate: ICC, 0.24; acute appendicitis with peritonitis: ICC, 0.21). Examples of conditions high in prevalence, cost, and variation in cost included major depressive disorder (cost rank, 19; prevalence rank, 10; ICC, 0.49), scoliosis (cost rank, 6; prevalence rank, 38; ICC, 0.27), acute appendicitis with peritonitis (cost rank, 13; prevalence rank, 11; ICC, 0.21), asthma (cost rank, 10; prevalence rank, 2; ICC, 0.17), and dehydration (cost rank, 24; prevalence rank, 8; ICC, 0.18). Conclusions and Relevance: This cohort study found that major depressive disorder, scoliosis, acute appendicitis with peritonitis, asthma, and dehydration were high in prevalence, costs, and variation in cost. These results could help identify where future comparative effectiveness research in hospital pediatrics should be targeted to improve the care and outcomes of hospitalized children.

The impact of the diabetes management incentive on diabetes-related services: evidence from Ontario, Canada.

Financial incentives have been introduced in several countries to improve diabetes management. In Ontario, the most populous province in Canada, a Diabetes Management Incentive (DMI) was introduced to family physicians practicing in patient enrollment models in 2006. This paper examines the impact of the DMI on diabetes-related services provided to individuals with diabetes in Ontario. Longitudinal health administrative data were obtained for adults diagnosed with diabetes and their family physicians. The study population consisted of two groups: DMI group (patients enrolled with a family physician exposed to DMI for 3 years), and comparison group (patients affiliated with a family physician ineligible for DMI throughout the study period). Diabetes-related services was measured using the Diabetic Management Assessment (DMA) billing code claimed by patient's physician. The impact of DMI on diabetes-related services was assessed using difference-in-differences regression models. After adjusting for patient- and physician-level characteristics, patient fixed-effects and patient-specific time trend, we found that DMI increased the probability of having at least one DMA fee code claimed by patient's physician by 9.3% points, and the probability of having at least three DMA fee codes claimed by 2.1% points. Subgroup analyses revealed the impact of DMI was slightly larger in males compared to females. We found that Ontario's DMI was effective in increasing the diabetes-related services provided to patients diagnosed with diabetes in Ontario. Financial incentives for physicians help improve the provision of targeted diabetes-related services.