Over the last decade, children 2 years of age and younger hospitalized with bronchiolitis have seen substantial increases in treatment cost and admission into intensive care units (ICUs), but those trends did not coincide with an increase in illness severity in this population, according to results from a cross-sectional study.
In order to better understand trends in bronchiolitis hospital resource use and diagnostic coding among children’s hospitals, Robert J. Willer, DO, of the University of Utah School of Medicine Primary Children’s Hospital in Salt Lake City, and colleagues performed a retrospective analysis of bronchiolitis cases in the Pediatric Health Information System (PHIS) database to determine the association of patient-level factors and coding practices with bronchiolitis hospitalization costs.
Their findings were published in JAMA Network Open.
“Data from this cohort of children’s hospitals revealed increasing costs per hospitalization for patients with bronchiolitis from 2010 to 2019,” they found. “Costs per hospitalization were higher when including patients with [complex chronic conditions] CCCs, although the rate of increase was similar regardless of medical complexity. Furthermore, costs increased among patients without a CCC or mechanical ventilation, suggesting that complexity of condition and mechanical ventilation are unlikely explanations for the observed trends. The only group that did not see increased costs across the period studied was patients who received care outside the ICU, suggesting that increasing costs for bronchiolitis are associated with higher ICU use.”
During the study period, the proportion of ICU-treated patients more than doubled, and the use of non-invasive ventilation rose even further, which would seem to suggest “increasing illness severity among hospitalized children may be responsible for increasing resource use,” they noted. This, along with increased use of the diagnostic code for respiratory failure in these patients, would seem to imply higher patient acuity—however, “decreasing total hospital and ICU [length of stay] LOSs, ICU use, mechanical ventilation rates, and mortality for the respiratory failure [All Patient Refined Diagnosis Related Group] APR-DRG cohort do not support this conclusion,” the study authors argued.
For their analysis, Willer and colleagues conducted a retrospective cross-sectional study of 385,883 infants ages 24 months or younger who were hospitalized with bronchiolitis at 39 hospitals in the PHIS database from Jan. 1, 2010 to Dec. 31, 2019.
The primary study outcome was trends in inflation-adjusted standardized unit cost per hospitalization over time; the outcome was not a direct measure of hospital dollar cost, but rather “a measure of the volume of resources expended, expressed in dollar units,” they explained. Patient-level outcomes included total hospital and ICU LOS, ICU admission rates, mechanical ventilation rates, and mortality rates. They also collected data on the proportion of patients with CCCs and evaluated severity of illness according to APR-DRG severity of illness (SOI) index scores.
Of the 385,883 bronchiolitis hospitalizations included in the analysis, patient mean age was 7.5 months, 227,309 were boys (58.9%), and 253,870 were publicly insured (65.8%).
“Among patients hospitalized with bronchiolitis, the median standardized unit cost per hospitalization increased significantly during the study period (from $5,636 [95% CI, $5,558-$5,714] in 2010 to $6,973 [95% CI, $6,915-$7,030] in 2019; P<0.001 for trend),” they found. “Similar increases in cost were observed among subgroups of patients without a complex chronic condition and without the need for mechanical ventilation. However, costs for patients without a complex chronic condition or mechanical ventilation, who received care outside the intensive care unit did not change in an economically significant manner (from $4,803 [95% CI, $4,752-$4,853] in 2010 to $4,853 [95% CI, $4,811-$4,895] in 2019; P<0.001 for trend), suggesting that intensive care unit use was a primary factor associated with cost increases.”
The study authors also noted substantial changes in coding practices over the study period.
“Among patients hospitalized with bronchiolitis, 1.2% (95% CI, 1.1%-1.3%) were assigned an APR-DRG (All Patient Refined Diagnosis Related Group) for respiratory failure in 2010, which increased to 21.6% (95% CI, 21.2%-21.9%) in 2019 (P<0.001 for trend),” they wrote. “Increased costs and coding intensity were not accompanied by objective evidence of worsening illness severity.”
So, if illness severity does not necessarily account for increasing costs and treatment intensity, what factors might explain these trends?
One possibility, the authors wrote, is that widespread adoption of high-flow nasal cannula (HFNC) could be increasing ICU utilization and raising SOI index scores, in which case physician’s subjective assessment of respiratory distress may be leading to increased HFNC use that is not necessarily correlated to the patient’s condition.
Another is that “an increased emphasis on accurate documentation and coding in hospitals could be associated with shifts in the APR-DRG SOI index score over time that are not dependent on patient-level factors… as an example, an APR-DRG SOI index score of 2 for bronchiolitis carries a state-specific service intensity weight of 0.6303, whereas an APR-DRG SOI index score of 2 for respiratory failure carries a state-specific service intensity weight of 0.9234, leading directly to higher facility reimbursement. Given the discrepancies in the formal definition of respiratory failure and how it is commonly used in clinical settings, we cannot determine from our data whether undercoding predominated early in our study period or whether overcoding predominated in later years.”
In an editorial accompanying the study, Christopher M. Horvat, MD, MHA, and Jonathan H. Pelletier, MD, both of the University of Pittsburgh School of Medicine in Pittsburgh, acknowledged both of these possibilities.
When it comes to HFNC, they argued that the therapy can potentially help maintain positive airway pressure and facilitate breathing among children with bronchiolitis—in other words, the goal of HFNC therapy in these patients is not to reduce mortality but rather to support treatment and reduce the need for mechanical ventilation. However, “to our knowledge, there are no well-established objective respiratory distress scales for bronchiolitis, nor do administrative databases harbor the necessary vital sign data and respiratory assessments to conduct such analyses.”
What’s more, analyzing whether the increase in ICU admissions for bronchiolitis is associated with an actual increase in cost or simply a measured increase of a database metric is also unclear, “because it would require a breakdown of both the fixed and variable costs attributable to the care of inpatient bronchiolitis,” they added, something that the PHIS database, which doesn’t include vital sign data, could not capture.
“Before therapies such as HFNC and patterns of use such as increasing ICU admissions become the focus of medical reversal and deimplementation initiatives, there is a need for more detailed data that accurately reflect the patient’s response to these aspects of care and that are uninfluenced by changes in coding practices designed to maximize reimbursements,” they concluded. “… Willer et al have highlighted the need for validated measures of respiratory distress; well-designed, prospective observational studies; and randomized clinical trials that will help our field refine the management of bronchiolitis.”
Willer and colleagues came to a similar conclusion, noting that their findings have implications for future studies, particularly when studying trends over time.
“Owing to significant shifts in coding practices, inclusion of only patients with a primary diagnosis of bronchiolitis or those who have been assigned to an APR-DRG for bronchiolitis will exclude a increasing proportion of patients hospitalized with bronchiolitis and may lead to biased results,” they wrote. “This is particularly true if patients with a diagnosis of respiratory failure are excluded because this diagnosis seems to be replacing bronchiolitis as the primary diagnosis for some patients with bronchiolitis. Furthermore, if the patient SOI index score is changing because of diagnostic coding practices rather than patient-level factors, using it in models as a covariate is likely to be associated with an inaccurate adjustment for the SOI index score. Such coding bias has been demonstrated among the adult Medicare population, and pediatric health services researchers should be cautious in assuming that diagnoses obtained from claims data are accurate and useful in reducing confounding.”
Study limitations included a lack of detailed clinical information in the PHIS database; the primary outcome measure represents an approximation of costs rather than actual health care costs or spending; and data from the PHIS database are limited to children’s hospitals and cannot account for shifts in the care of child patients from community hospitals to children’s hospitals.
Over the last decade, children 2 years of age and younger hospitalized with bronchiolitis have seen substantial increases in treatment cost and admission into intensive care units (ICUs), but those trends did not coincide with an increase in illness severity in this population.
Over the study period, diagnostic coding practices changed substantially, with a major increase in the number of patients hospitalized for bronchiolitis who were assigned an All Patient Refined Diagnosis Related Group (APR-DRG) for respiratory failure from 2010-2019.
John McKenna, Associate Editor, BreakingMED™
Study coauthor Coon reported receiving grants from Intermountain Stanford Collaboration, which funded a randomized clinical trial (NCT03354325) for patients hospitalized with bronchiolitis, outside the submitted work. Coauthor Harrison reported receiving grants from Health Resources and Services Administration during the conduct of the study.
Horvat reported receiving grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the National Institute of Neurological Disorders and Stroke. Pelletier reported receiving a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development.
Cat ID: 138
Topic ID: 85,138,501,728,791,730,138,192,195,925