OBJECTIVES:

Examine the degree of seasonal variation in nonrecommended resource use for bronchiolitis management subsequent to publication of the American Academy of Pediatrics (AAP) 2014 guidelines.

METHODS:

We performed a multicenter retrospective cohort study using the Pediatric Health Information System database, examining patients aged 1 to 24 months, diagnosed with bronchiolitis between November 2015 and November 2018. Exclusions included presence of a complex chronic condition, admission to the PICU, hospital stay >10 days, or readmission. Primary outcomes were use rates of viral testing, complete blood count, blood culture, chest radiography, antibiotics, albuterol, and systemic steroids. Each hospital’s monthly bronchiolitis census was aggregated into hospital bronchiolitis census quartiles. Mixed-effect logistic regression was performed, comparing the primary outcomes between bronchiolitis census quartiles, adjusting for patient age, race, insurance, hospitalization status, bacterial coinfection, time since publication of latest AAP bronchiolitis guidelines, and clustering by site.

RESULTS:

In total, 196 902 bronchiolitis patient encounters across 50 US hospitals were analyzed. All hospitals followed a similar census pattern, with peaks during winter months and nadirs during summer months. Chest radiography, albuterol, and systemic steroid use were found to significantly increase in lower bronchiolitis census quartiles, whereas rates of viral testing significantly decreased. No significant variation was found for complete blood count testing, blood culture testing, or antibiotic use. Overall adherence with AAP guidelines increased over time.

CONCLUSIONS:

Resource use for patients with bronchiolitis varied significantly across hospital bronchiolitis census quartiles despite adjusting for potential known confounders. There remains a need for greater standardization of bronchiolitis management.

Bronchiolitis, the most common acute lower respiratory system disease in infants, is the leading cause of hospitalization in children within the first 12 months of life.1  In 2006, the American Academy of Pediatrics (AAP) published clinical practice guidelines recommending against the routine use of non–evidence-based interventions commonly used to diagnose and treat bronchiolitis (including bronchodilators, corticosteroids, antiviral and antibacterial agents, and chest physiotherapy), instead recommending primarily supportive care with diagnosis and severity-of-illness assessment based on history and physical examination.2  In 2014, the AAP updated the 2006 guidelines, supporting the earlier guidelines recommending radiographic and laboratory studies not be obtained routinely.1 

Although several studies have revealed modest reductions in use of nonrecommended diagnostic tests and therapies since the publication of the AAP bronchiolitis guidelines, wide variation among hospitals in diagnosis and management persists, with authors of studies continuing to find considerable overuse of many nonrecommended therapies.35  Such overuse of non–evidence-supported resources may result in unnecessary costs to both patients and the health care system, and has led researchers interested in high-quality and cost-efficient care to attempt to explain persistent variations in guideline adoption.

One such focus has been on the possible role of seasonal variation in bronchiolitis and how that might influence the differential adoption of bronchiolitis guidelines. Respiratory syncytial virus (RSV), which has been implicated as the causative pathogen for acute bronchiolitis in 50% to 80% of cases, is a seasonal virus typically beginning during the fall season and continuing through the winter and spring in temperate climates.6  A previous study of patients hospitalized with bronchiolitis revealed that during the winter months, infants received better quality care in terms of reduced probability of receiving steroids, radiographs, and laboratory tests.7  However, this study was before the 2014 guidelines and was limited to patients hospitalized with bronchiolitis, who only constitute a small proportion of patients with bronchiolitis.

The primary goal of this study is to examine the degree of seasonal variation in nonrecommended resource use for diagnostic tests and therapeutic treatments of bronchiolitis subsequent to publication of the 2014 AAP bronchiolitis guidelines. Secondarily, in this study, we seek to understand the underlying factors that may explain any observed differences. In particular, by comparing the month before to the month after the peak bronchiolitis season at an institutional level, we hope to understand whether differences are more likely driven by bronchiolitis prevalence or by other institutional or provider factors. We hypothesize that, if use is similar in the pre- and post-month, variation is more likely attributable to bronchiolitis census, whereas variation in the pre- and post-month that is more similar to previous months would reflect other temporal factors, such as physician short-term memory.

We performed a multicenter retrospective cohort study using the Pediatric Health Information System (PHIS) database (Children’s Hospital Association, Overland Park, KS). The PHIS database is an administrative and billing database collaborative consisting of 51 freestanding tertiary care children’s hospitals coordinated by the Children’s Hospital Association.

Patients who presented to a study hospital with a primary diagnosis of bronchiolitis (C.A., S.M., E.K., R.M., B.A., unpublished observations) between November 1, 2015, and October 31, 2018, were included in this study, with the start point chosen such that the 2014 AAP bronchiolitis guidelines had been widely disseminated at that time, given their publication 1 year before. Inclusion criteria included all patients aged 1 to 24 months presenting to the hospital for the emergency department, ambulatory surgery, the observation unit, or general inpatient wards encounters. A subset analysis of hospitalized patients included both observation unit and general inpatient wards encounters. Exclusion criteria included the presence of a chronic complex condition, PICU admission, a length of stay >10 days, or any readmission during the study period (primary visits were not excluded). Chronic complex condition was defined by using International Classification of Diseases, Ninth Revision and International Classification of Diseases, 10th Revision billing codes, as previously described.8,9  Patients with asthma or reactive airways disease as a diagnosis were not excluded; however, patients with other chronic lung diseases were. Because RSV seasonality in Florida is different from the rest of the United States,10  1 hospital from Florida was excluded, with 50 remaining hospitals used as part of our analysis.

For individual hospitals, the calendar month’s total bronchiolitis census (all outpatient and inpatient encounters with a primary diagnosis of bronchiolitis) was collected and analyzed across the 3-year study period. First, the individual hospital monthly census (eg, December 2015, 2016, and 2017) was averaged across the 3 years to create a mean individual hospital monthly census for the 3-year time period. Of note, there was a predictable seasonal variation across all hospitals, with higher census months in the winter and lower census months in the summer across the 3 years (see Fig 1). Next, individual hospital average monthly bronchiolitis census was rank ordered into quartiles. Quartile 4 was designated first as the busiest consecutive 3 average monthly censuses for an individual hospital, after which the remaining 3 quartiles were designated, with quartile 1 defined as the least busy consecutive 3 average monthly censuses. Afterward, all individual hospital data were aggregated by their quartiles for analysis (ie, hospital 1’s quartile 4 grouped with hospital 2’s quartile 4, etc). Because monthly census aggregation was done at the hospital-level, quartiles encompass different months for different hospitals. However, most hospitals fit a similar pattern of monthly census throughout the 3-year study window (see Fig 1, Supplemental Fig 4 for more detail).

FIGURE 1

Census of study hospitals, by month, for patients seen with bronchiolitis. The dark line represents the mean number for the entire cohort.

FIGURE 1

Census of study hospitals, by month, for patients seen with bronchiolitis. The dark line represents the mean number for the entire cohort.

Close modal

Patient characteristics collected from the database were age, sex, race or ethnicity, payer type, and whether the patient was admitted or not. Outcomes collected were the rates of viral testing, complete blood count, blood culture, chest radiography, antibiotics, albuterol, and systemic steroids. Because of the concern for confounding by bacterial coinfection, all secondary diagnoses listed for each bronchiolitis encounter were independently analyzed by 2 authors (C.A. and S.M.) and annotated as to whether they represented bacterial infections, with 99% concordance in annotation. The discordant annotations (1%) were settled in discussion with a third author (B.A.), and a dichotomized variable reflecting this assessment (the presence of a bacterial coinfection or not) was generated. In addition, the number of months between each patient encounter and the date of the November 2014 AAP bronchiolitis guidelines publication was calculated to account for possible increased guideline adherence over time. Finally, the total estimated adjusted cost, as provided by the PHIS database, which is adjusted for differences in labor costs between the different hospitals on the basis of geographic location by using the Health Care Financing Administration wage price index11  for each encounter, was also collected and averaged over its length to derive adjusted average cost per day.

Mixed-effect logistic regression was used to test the association between the hospital bronchiolitis census quartiles and the measured outcomes (use rates of viral testing, complete blood count, blood culture, chest radiography, antibiotics, albuterol, and systemic steroids). Models were adjusted for patient age, ethnicity or race, insurance, hospitalization status, bacterial coinfection, time since publication of the latest 2014 AAP bronchiolitis guidelines, and clustering by site by using a random intercept. Odds for each primary outcome were compared between the different quartile census months by using the highest bronchiolitis census quartile as the reference. The above was repeated on the subset of patients who were hospitalized to evaluate for any discrepancies between the inpatient and outpatient settings. Finally, as a secondary analysis, bronchiolitis cases occurring in the month immediately before quartile 4 and those occurring in the month immediately after quartile 4 were compared, adjusting for the same potential confounders as above, to see how other institutional or provider factors may affect outcomes.

This research was deemed non–human subjects by the University of Nebraska Medical Center Institutional Review Board. P values for association between bronchiolitis census quartiles and use measures, as well as distribution of demographics other than age by bronchiolitis census quartile, were generated by using χ2 tests. The P value for distribution of age by bronchiolitis census quartile was determined by Fischer’s exact test. All analyses were conducted by using the software R, version 3.6.4 (The R Foundation, Vienna, Austria).

Fifty hospitals were included as part of the analysis after excluding 1 hospital from Florida. There were 275 636 total encounters identified in the PHIS database meeting our inclusion criteria. Of these, 78 734 were excluded on the basis of our exclusion criteria. Characteristics of the remaining 196 902 patients are included in Table 1.

TABLE 1

Characteristics of Patients With Diagnosis of Bronchiolitis, by Hospital Bronchiolitis Census Quartile

Quartile 1, n = 11 819 (6.0%)Quartile 2, n = 24 583 (12.5%)Quartile 3, n = 52 182 (26.5%)Quartile 4, n = 108 318 (55.0%)P
Median age (IQR), mo 8.20 (8.2) 7.60 (8.6) 6.90 (8.0) 6.60 (8.3) <.001 
Ethnicity, n (%)     <.001 
 White 3699 (31.3) 8041 (32.7) 19 901 (38.1) 43 655 (40.3) — 
 African American 3935 (33.3) 8014 (32.6) 14 021 (26.9) 24 790 (22.9) — 
 Hispanic 2591 (21.9) 5381 (21.9) 12 188 (23.4) 26 769 (24.7) — 
 Other 1281 (10.8) 2466 (10.0) 4742 (9.1) 10 239 (9.5) — 
 Missing 313 (2.7) 681 (2.8) 1330 (2.6) 2865 (2.6) — 
Sex, n (%)     <.001 
 Female 4513 (38.2) 9446 (38.4) 21 194 (40.6) 46 012 (42.5) — 
 Male 7306 (61.8) 15 136 (61.6) 30 983 (59.4) 62 298 (57.5) — 
 Missing 0 (0.0) 1 (0.0) 5 (0.0) 8 (0.0) — 
Payer, n (%)     <.001 
 Public 8318 (70.4) 17 276 (70.3) 35 819 (68.6) 71 809 (66.3) — 
 Private 2768 (23.4) 5807 (23.6) 13 329 (25.5) 29 741 (27.5) — 
 Other 733 (6.2) 1500 (6.1) 3034 (5.8) 6768 (6.3) — 
Hospitalized, n (%)     .01 
 Yes 3573 (30.2) 7070 (28.8) 15 004 (28.8) 31 214 (28.8) — 
 No 8246 (69.8) 17 513 (71.2) 37 178 (71.3) 77 103 (71.2) — 
Bacterial diagnosis, n (%)     <.001 
 Yes 1366 (11.6) 2962 (12.1) 7318 (14.0) 16 574 (15.3) — 
 No 10 453 (88.4) 21 601 (87.9) 44 787 (86.0) 91 676 (84.7) — 
Quartile 1, n = 11 819 (6.0%)Quartile 2, n = 24 583 (12.5%)Quartile 3, n = 52 182 (26.5%)Quartile 4, n = 108 318 (55.0%)P
Median age (IQR), mo 8.20 (8.2) 7.60 (8.6) 6.90 (8.0) 6.60 (8.3) <.001 
Ethnicity, n (%)     <.001 
 White 3699 (31.3) 8041 (32.7) 19 901 (38.1) 43 655 (40.3) — 
 African American 3935 (33.3) 8014 (32.6) 14 021 (26.9) 24 790 (22.9) — 
 Hispanic 2591 (21.9) 5381 (21.9) 12 188 (23.4) 26 769 (24.7) — 
 Other 1281 (10.8) 2466 (10.0) 4742 (9.1) 10 239 (9.5) — 
 Missing 313 (2.7) 681 (2.8) 1330 (2.6) 2865 (2.6) — 
Sex, n (%)     <.001 
 Female 4513 (38.2) 9446 (38.4) 21 194 (40.6) 46 012 (42.5) — 
 Male 7306 (61.8) 15 136 (61.6) 30 983 (59.4) 62 298 (57.5) — 
 Missing 0 (0.0) 1 (0.0) 5 (0.0) 8 (0.0) — 
Payer, n (%)     <.001 
 Public 8318 (70.4) 17 276 (70.3) 35 819 (68.6) 71 809 (66.3) — 
 Private 2768 (23.4) 5807 (23.6) 13 329 (25.5) 29 741 (27.5) — 
 Other 733 (6.2) 1500 (6.1) 3034 (5.8) 6768 (6.3) — 
Hospitalized, n (%)     .01 
 Yes 3573 (30.2) 7070 (28.8) 15 004 (28.8) 31 214 (28.8) — 
 No 8246 (69.8) 17 513 (71.2) 37 178 (71.3) 77 103 (71.2) — 
Bacterial diagnosis, n (%)     <.001 
 Yes 1366 (11.6) 2962 (12.1) 7318 (14.0) 16 574 (15.3) — 
 No 10 453 (88.4) 21 601 (87.9) 44 787 (86.0) 91 676 (84.7) — 

P value significance testing by Fisher’s exact tests. IQR, interquartile range; —, not applicable.

Among our sample, 59% of the patients were male, 68% were publicly insured, 75% were between 1 and 12 months of age, and the median age was 7 months (Table 1). Patient median age decreased as quartile census increased (8.2 months in quartile 1 to 6.6 months in quartile 4; P < .001), white infants increasingly presented as quartile census increased (31.3% of patients in quartile 1 to 40.3% of patients in quartile 4; P < .001), and bacterial coinfection rates also increased by bronchiolitis census quartile (11.6% in quartile 1 to 15.3% in quartile 4; P < .001). Although there were statistically significant differences in sex, payer mix, and rates of hospitalization by quartile, these differences were small and not clinically significant (shown in Table 1). All 50 hospitals followed a similar bronchiolitis census pattern, with peaks during winter months and nadirs during summer months (Fig 1).

There were significant differences in unadjusted rates of both testing and treatment variables by quartile (Table 2). Chest radiography (30.2%–20.3%), albuterol use (34.7%–21.5%), and systemic steroid use (12.1%–5.7%) significantly decreased as quartile census increased (P < .001). In contrast, more patients received viral testing (6.3%–17.2%) and antibiotics (8.7%–10.1%) as quartile census increased (P < .001).

TABLE 2

Crude Primary Outcome Measure, by Hospital Bronchiolitis Census Quartiles

Quartile 1, n (%)Quartile 2, n (%)Quartile 3, n (%)Quartile 4 (Reference), n (%)P
Total 11 819 (6.0) 24 583 (12.5) 52 182 (26.5) 108 318 (55.0) — 
Blood culture 639 (5.4) 1209 (4.9) 2927 (5.6) 6482 (6.0) <.001 
Complete blood cell count 985 (8.3) 1870 (7.6) 4162 (8.0) 8975 (8.3) .002 
Viral testing 750 (6.4) 2319 (9.4) 8223 (15.8) 18 587 (17.2) <.001 
Chest radiograph 3569 (30.2) 6444 (26.2) 11 984 (23.0) 21 977 (20.3) <.001 
Albuterol 4099 (34.7) 7410 (30.1) 12 789 (24.5) 23 276 (21.5) <.001 
Antibiotics 1023 (8.7) 2064 (8.4) 4837 (9.3) 10 992 (10.1) <.001 
Systemic steroids 1434 (12.1) 2517 (10.2) 3897 (7.5) 6156 (5.7) <.001 
Quartile 1, n (%)Quartile 2, n (%)Quartile 3, n (%)Quartile 4 (Reference), n (%)P
Total 11 819 (6.0) 24 583 (12.5) 52 182 (26.5) 108 318 (55.0) — 
Blood culture 639 (5.4) 1209 (4.9) 2927 (5.6) 6482 (6.0) <.001 
Complete blood cell count 985 (8.3) 1870 (7.6) 4162 (8.0) 8975 (8.3) .002 
Viral testing 750 (6.4) 2319 (9.4) 8223 (15.8) 18 587 (17.2) <.001 
Chest radiograph 3569 (30.2) 6444 (26.2) 11 984 (23.0) 21 977 (20.3) <.001 
Albuterol 4099 (34.7) 7410 (30.1) 12 789 (24.5) 23 276 (21.5) <.001 
Antibiotics 1023 (8.7) 2064 (8.4) 4837 (9.3) 10 992 (10.1) <.001 
Systemic steroids 1434 (12.1) 2517 (10.2) 3897 (7.5) 6156 (5.7) <.001 

—, not applicable.

After adjustment for age, ethnicity or race, insurance status, time since publication of the AAP bronchiolitis guidelines, clustering, and bacterial coinfection, this census-dependent decrease in the use of chest radiography, albuterol use, and systemic steroids in higher census months as compared to lower census months persisted (P < .001; Fig 2, the vertical line represents the highest census quartile 4), as did the higher use of viral testing in higher census months as compared to lower census months (P < .001; Fig 2). Significant differences in antibiotic use across censuses were no longer seen. No significant differences between quartiles were found for obtaining a complete blood count or a blood culture in unadjusted or adjusted analyses. Of note, the adjustment for time since publication of the AAP guidelines significantly impacted the results, with each year since publication equating to 7% reduced odds of chest radiography (odds ratio [OR]: 0.93; 95% confidence interval [CI]: 0.92–0.94), 4% reduced odds of blood culture (OR: 0.96; 95% CI: 0.94–0.99), 2% increased odds of viral testing (OR: 1.02; 95% CI: 1.01–1.04), 15% reduced odds of albuterol (OR: 0.85; 95% CI: 0.84–0.86), 5% reduced odds of systemic steroids (OR: 0.95; 95% CI: 0.93–0.97), and 5% reduced odds of antibiotics (OR: 0.95; 95% CI: 0.93–0.97).

FIGURE 2

Crude and adjusted ORs for testing and treatment of bronchiolitis (quartile 1, lowest bronchiolitis census quartile; quartile 2, second-lowest bronchiolitis census quartile; quartile 3, second-highest bronchiolitis census quartile; individual bars), as compared to the busiest bronchiolitis census quartile (quartile 4, vertical dashed line). CBC, complete blood cell count.

FIGURE 2

Crude and adjusted ORs for testing and treatment of bronchiolitis (quartile 1, lowest bronchiolitis census quartile; quartile 2, second-lowest bronchiolitis census quartile; quartile 3, second-highest bronchiolitis census quartile; individual bars), as compared to the busiest bronchiolitis census quartile (quartile 4, vertical dashed line). CBC, complete blood cell count.

Close modal

Quartile 1 patients incurred significantly higher costs per day on average compared with quartile 4 patients (quartile 1: $944; quartile 4: $848; P < .001). A subset analysis of only hospitalized patients (inpatient or observation status) revealed similar significant census-graded differences across quartiles among the same outcome measures (data not shown). Moreover, the percentage of patients hospitalized for bronchiolitis across each bronchiolitis census quartile was not meaningfully different (quartile 1: 30.2%; quartile 2: 28.8%; quartile 3: 28.8%; quartile 4: 28.8%).

A secondary analysis was conducted comparing the month directly before the highest bronchiolitis census quartile (quartile 4) to the month directly after. The cumulative number of patients seen across hospitals in the month before the high-census period was similar to the number of patients seen in the month after the high-census period (average number of patients seen across hospitals in the month before and after quartile 4 = 346.9 [SD: 257.0] vs 403.2 [SD: 291.6] [P = .304]; cumulative totals before and after quartile 4 across years and hospitals = 17.621 vs 20 561). By using the month directly after the highest bronchiolitis census quartile as the reference, the month directly before quartile 4 had greater use of albuterol and systemic steroids and significantly less use of viral testing, as compared to the month after quartile 4 (Fig 3).

FIGURE 3

Crude and adjusted ORs for testing and treatment of bronchiolitis during the month before the busiest bronchiolitis census quartile (individual bars) compared to the month after the busiest bronchiolitis census quartile (quartile 4, vertical dashed line). CBC, complete blood cell count.

FIGURE 3

Crude and adjusted ORs for testing and treatment of bronchiolitis during the month before the busiest bronchiolitis census quartile (individual bars) compared to the month after the busiest bronchiolitis census quartile (quartile 4, vertical dashed line). CBC, complete blood cell count.

Close modal

In this large multisite cohort of infants diagnosed with uncomplicated bronchiolitis between 1 and 24 months of age, we observed significant variation in resource use across high and low bronchiolitis census months. Although the 2014 AAP evidence-based bronchiolitis guidelines suggest a minimization of unnecessary diagnostic testing and treatment of bronchiolitis, our results reveal that there is significant variation across seasons that is not recommended by AAP guidelines. In particular, we observed significant increases in the use of chest radiography, albuterol, and systemic steroids during low census months, indicating lower physician compliance with AAP guidelines during times of low bronchiolitis census on these measures. In contrast, viral testing was significantly higher in high-census months. No significant trends were observed across census months for complete blood count, blood culture, or antibiotic use.

To further understand whether our observed trends were more likely to be due to bronchiolitis census or temporally driven by physician behavior, we compared the month before the highest bronchiolitis census quartile to the month after, which did not have a statistically significant difference in census. If census is the driving force behind behavior, we would expect these 2 months to have similar use rates. However, if temporal experience and short-term memory are the driver, we would expect that use rates in the month after the peak bronchiolitis quartile would be more similar to those during the peak season, whereas use rates in the month before the peak quartile would be more similar to those during low season months.

We found that there were significantly higher rates of albuterol and steroids in the month before the bronchiolitis surge compared to the month after, implying that physician behavior may be influenced by recent patient care expectations, as opposed to individual patient differences. It is possible that this greater use of albuterol and steroids in the early bronchiolitis season may reflect provider uncertainty regarding a diagnosis of bronchiolitis versus asthma, whereas in the month after the bronchiolitis peak, there is greater comfort among providers to minimize the use of non–evidence-based measures for bronchiolitis. There was, however, no difference in the rates of chest radiography use between these months. As such, it seems that physicians do not expect seasonal variation in community-acquired pneumonia or other diseases notable on chest radiography, but they do expect seasonal variation in bronchiolitis. Greater viral testing in high bronchiolitis census months may be a result of local hospital practices, such as policies to cohort patients in the hospital with similar viral pathogens. Alternatively, it may reflect provider beliefs that viral pathogen testing is more appropriate in times when there may be a higher pretest probability of viral coinfection. Regardless, a higher pretest probability of viral pathogens does not justify greater testing and is not recommended by AAP bronchiolitis guidelines.

Although our analysis adjusts for demographic variation and the presence of coinfection, our unadjusted analysis revealed notable demographic variation across bronchiolitis census quartiles. Notably, patients were slightly younger and consisted of a greater proportion of white patients on high-census months as compared to low census months for reasons that are unclear. Likewise, the proportion of bacterial coinfection was also slightly greater in high-census months, which may reflect the greater incidence of certain bacterial infections during winter months.

Although our data do not allow us to impute causality behind observed trends or to determine if there is any therapeutic benefit to the observed seasonal variation, to the best of our knowledge, there are no studies that indicate there should be differences in the diagnostic or medical management of bronchiolitis across seasons or the local disease prevalence. Consistent with Van Cleve and Christakis,7  we speculate that the higher rate of chest radiography, albuterol, and steroid use may be due to a higher degree of diagnostic uncertainty or discomfort among providers to diagnose bronchiolitis on the basis of history and examination alone during lower bronchiolitis census months. However, respiratory viruses other than influenza and RSV that cause bronchiolitis circulate year-round; hence, a clinical presentation of bronchiolitis during low census periods should not necessarily lead a physician to workup and treat for other nonbronchiolitis causes.12  It has been shown that local practice guidelines can reduce unnecessary resource use.13  Further emphasis of use of these guidelines during low census months may further reduce unnecessary testing and therapies and reduce cost. In future studies, researchers should focus on understanding the underlying factors associated with seasonal variation in resource use for bronchiolitis.

It is notable that, in our study, we report a lower overall use of resources across all quality metrics as compared to similar studies in past years, suggesting that there has been continued improvement in the adoption of national guidelines across children’s hospitals.3,7  Indeed, we find this to be the case in our data set, with our data revealing significant yearly decreases in the use of chest radiography, blood cultures, albuterol, systemic steroids, and antibiotics. Interestingly, and in contrast, viral testing slightly increased yearly (1.02 OR), perhaps reflecting increased availability of respiratory viral testing panels over time.

Despite this overall improved adherence to the guidelines over time, in our results we identify potentially important areas for quality improvement, particularly because our data suggest that patients with bronchiolitis in lower census months incur significantly higher costs on average per day. Although the number of patients with bronchiolitis seen in high-census months greatly outnumbers the number of those in low census months, suggesting that quality improvement interventions could have the greatest magnitude of effect if enacted during high-census periods, the numbers in low census months are not insignificant, and meaningful cost and quality impacts can still be made during them. In addition, improving adherence to best practice guidelines should always be sought, and our findings reveal that low census months are most ripe for improvement.

This study has several limitations, including its inherent use of a large, deidentified billing and demographic database that is dependent on diagnosis and procedure codes, and thus does not take into account more nuanced clinical information that may have explained the use of ancillary testing. As such, we cannot exclude the possibility that specific tests or therapies may have been appropriately ordered for reasons that are obscured by this analysis and/or not addressed in the national guidelines. Another limitation includes our analysis of the pre- and post–high season months, with which we sought to further understand whether testing differences are due to temporally driven physician behavior or census. Although census differences between these months do not meet statistical significance, there is still a larger census in the month after (16% higher), which does not allow us to exclude the possibility that these differences may have been ascribed to census rather than temporal-driven behavior. Nevertheless, regardless of whether these differences are due to census or temporally driven by physician behavior, they reflect important seasonal decreases in adherence to AAP bronchiolitis guidelines. Lastly, although, in our study, we used a nationally representative sample involving 50 tertiary care children’s hospitals, our results may not be generalizable to the majority of infants who are treated for bronchiolitis at smaller community pediatric hospitals, which may have different physician practice trends.

There is significant variation in resource use rates for bronchiolitis across high and low census months within this large cohort of patients. Our findings reveal the need for a greater focus on the standardization of decisions regarding diagnostic and therapeutic interventions, with particular attention to lower bronchiolitis census months when deviance from the guidelines is greatest.

Dr Andrews contributed to the analysis and interpretation of data and drafted the initial abstract and Methods, Results, and Discussion sections; Dr Maxwell contributed to the analysis and interpretation of data and drafted the introduction; Ms Kerns collected the data from the Pediatric Health Information System database, conducted the data analysis, and contributed to and helped revise the Methods section; Drs Alverson and McCulloh conceptualized and designed the study and supervised the analyses; and all authors reviewed and revised the manuscript, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.

FUNDING: Dr McCulloh and Ms Kerns receive support from the Office of the Director of the National Institutes of Health under award UG1OD024953. Funded by the National Institutes of Health (NIH).

1
Ralston
SL
,
Lieberthal
AS
,
Meissner
HC
, et al
;
American Academy of Pediatrics
.
Clinical practice guideline: the diagnosis, management, and prevention of bronchiolitis [published correction appears in Pediatrics. 2015;136(4):782]
.
Pediatrics
.
2014
;
134
(
5
).
2
American Academy of Pediatrics Subcommittee on Diagnosis and Management of Bronchiolitis
.
Diagnosis and management of bronchiolitis
.
Pediatrics
.
2006
;
118
(
4
):
1774
1793
3
Parikh
K
,
Hall
M
,
Teach
SJ
.
Bronchiolitis management before and after the AAP guidelines
.
Pediatrics
.
2014
;
133
(
1
).
4
McCulloh
RJ
,
Smitherman
SE
,
Koehn
KL
,
Alverson
BK
.
Assessing the impact of national guidelines on the management of children hospitalized for acute bronchiolitis
.
Pediatr Pulmonol
.
2014
;
49
(
7
):
688
694
5
Florin
TA
,
Byczkowski
T
,
Ruddy
RM
,
Zorc
JJ
,
Test
M
,
Shah
SS
.
Variation in the management of infants hospitalized for bronchiolitis persists after the 2006 American Academy of Pediatrics bronchiolitis guidelines
.
J Pediatr
.
2014
;
165
(
4
):
786
792.e1
6
Miller
EK
,
Gebretsadik
T
,
Carroll
KN
, et al
.
Viral etiologies of infant bronchiolitis, croup and upper respiratory illness during 4 consecutive years
.
Pediatr Infect Dis J
.
2013
;
32
(
9
):
950
955
7
Van Cleve
WC
,
Christakis
DA
.
Unnecessary care for bronchiolitis decreases with increasing inpatient prevalence of bronchiolitis
.
Pediatrics
.
2011
;
128
(
5
).
8
Feudtner
C
,
Christakis
DA
,
Connell
FA
.
Pediatric deaths attributable to complex chronic conditions: a population-based study of Washington State, 1980–1997
.
Pediatrics
.
2000
;
106
(
1 pt 2
):
205
209
9
Feudtner
C
,
Feinstein
JA
,
Zhong
W
,
Hall
M
,
Dai
D
.
Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation
.
BMC Pediatr
.
2014
;
14
:
199
10
Pitzer
VE
,
Viboud
C
,
Alonso
WJ
, et al
.
Environmental drivers of the spatiotemporal dynamics of respiratory syncytial virus in the United States
.
PLoS Pathog
.
2015
;
11
(
1
):
e1004591
11
Freeland
MS
,
Chulis
GS
,
Brown
AP
, et al
.
Measuring hospital input price increases: the rebased hospital market basket
.
Health Care Financ Rev
.
1991
;
12
(
3
):
1
14
12
Price
OH
,
Sullivan
SG
,
Sutterby
C
,
Druce
J
,
Carville
KS
.
Using routine testing data to understand circulation patterns of influenza A, respiratory syncytial virus and other respiratory viruses in Victoria, Australia
.
Epidemiol Infect
.
2019
;
147
:
e221
13
Mittal
V
,
Hall
M
,
Morse
R
, et al
.
Impact of inpatient bronchiolitis clinical practice guideline implementation on testing and treatment
.
J Pediatr
.
2014
;
165
(
3
):
570
576.e3

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data