Comparison Between Physicians and Nurse Practitioners’ Resource Utilization in the Diagnosis and Management of Bronchiolitis in the Pediatric Emergency Department

INTRODUCTION

Nurse practitioners (NPs) are increasingly incorporated into emergency department (ED) staffing models to ensure high-quality care, improve patient outcomes, satisfaction, wait-times, and institutional costs.^1–4 They frequently care for children, as pediatric visits constitute roughly 20% of all visits to the ED.^5–6 Teams following the emergency NP service model have significantly higher levels of adherence to clinical practice guidelines (CPG) along with heightened diagnostic accuracy compared with a non-NP team following the standard of care model.⁷ NPs are well equipped to manage respiratory diseases within both primary and acute care settings.⁸ While respiratory diseases in children such as bronchiolitis are commonly treated by NPs in the ED, there are limited studies examining the resource utilization between advanced practice providers (APP) and physicians. Therefore, this study aimed to describe and compare the diagnostic and therapeutic management of bronchiolitis among NP and physicians in the pediatric ED, and to describe and compare medical practices before and after an institutional CPG implementation for the diagnosis and management of bronchiolitis.

MATERIALS AND METHODS

Study design and data source:

This retrospective cross-sectional study analyzed the de-identified standard of care quality metrics for pediatric ED medical visits of children with a diagnosis of bronchiolitis. The setting was a Level 1 academic pediatric trauma center where ~60,000 pediatric patients are evaluated in the ED annually. This study was approved by the Office of Human Subjects Protection.

Identification of sample:

We included patients aged 1 to 24 months evaluated in the ED between January 1, 2014 and November 30, 2018. We included the International Classification of Diseases, Ninth and Tenth Revisions (ICD-9-CM and ICD-10) codes for the following: acute bronchiolitis due to respiratory syncytial virus (RSV) (466.11 or J21.0), acute bronchiolitis unspecified (466.1 or J21.9), acute bronchiolitis due to other infectious organisms (466.19 or J21.8), and wheezing (786.07 or R06.2).

Study variables:

Data included date of ED visit, age stratified into groups (1–3, 4–12, and 13–24 months), ED length of stay (LOS, minutes), ED disposition (discharge or admission), ED diagnosis, diagnostic tests (chest radiography, viral testing, RSV tests), treatments (bronchodilator, corticosteroid, antibiotics), respiratory score (RS), RS scoring location, smoking status of parent or caregiver, and medical provider (physician, NP, physician in combination with NP). The combination of physician and NP occurred for patients evaluated/managed by a NP requiring higher levels of treatment based on patient severity or clinical impression. Smoking status (never smoker, passive smoke exposure-never smoker, and current daily smoker) was collected from the parent or guardian upon inpatient admission. Testing or treatment (ToT) was defined by any of the following ED orders: chest radiography, viral testing, RSV test, bronchodilator, corticosteroid, and antibiotic. Total ToT was defined as the sum of all ToT performed in the ED during a single medical visit (range 0 to 6). We defined the following timelines: before the institutional CPG implementation as early era, from January 1, 2014, to October 31, 2016; and after the institutional CPG implementation as late era, from November 1, 2016, to November 30, 2018. The institutional CPG was based on the American Academy of Pediatrics’ CPG recommendations, which emphasizes supportive care and discourages the use of radiographies, viral testing, bronchodilators, and steroids in the diagnosis and management of bronchiolitis.⁹

Outcome measures:

The primary outcome measures were to 1) compare the distribution of diagnostic tests and treatment modalities among providers evaluating and managing children with bronchiolitis, 2) describe the effect variation in resource utilization before and after the institutional CPG implementation on the diagnosis and management of bronchiolitis in the ED. Secondary outcomes of interest included age, ED LOS, diagnosis, disposition, and use of an RS.

Statistical analysis:

All categorical variables were presented as counts and percentages and continuous variables as median with interquartile range (IQR [Q1–Q3]). We calculated the Total ToT as a conceptually continuous variable. Comparisons between groups were done through chi-squared analysis, Fisher’s exact test, or Kruskal-Wallis, as appropriate. Comparisons were made for each provider (physician, NP, or both) as well as providers between eras. RS medians were calculated for providers, testing, disposition, and diagnosis classification (wheezing, bronchiolitis). Smoking status was compared with the ED RS, and Total ToT. All statistical analyses were conducted in IBM SPSS Statistics 25 (IBM Corp., Armonk, NY, USA) with p-values less than 0.05 considered statistically significant.

RESULTS

A total of 5,311 visits to the ED for wheezing and bronchiolitis were evaluated. We excluded 51 medical visits (1%) due to missing variables. The median age was 295 days (IQR 162–474). The median length of stay was 144 minutes (IQR 91–206). The highest proportion of patients were treated by a physician (65.3%), followed by NP (30.3%), and a combination of NP/physician (4.3%). Infants younger than 12 months represented 60% of all wheezing and bronchiolitis ED visits. There was a significant difference in the distribution of age groups among providers (P<0.001). The 1–3 month age group was more frequently evaluated by the NP/physician combination (15%), followed by a physician (14%). The age groups 4–12 months (53%) and 13–24 months (43%) were more frequently evaluated by an NP. Over half (58%) of ED visits took place between the months of December through March (P=0.009). Patient LOS was significantly longer when a physician was working alone (149 minutes, IQR 61–226 minutes) or with an NP (182 minutes, IQR 93–285 minutes) (P<0.001). NPs more frequently discharged patients home compared with physicians or the NP/physician combination (98%, 69% and 64%, P<0.001, respectively). We found no difference in the admission location between providers. NPs and the combination of NP/physician more frequently assigned a wheezing diagnosis (52% and 56%, respectively) compared with physicians (43%, P<0.001). Physicians and the combination of NP/physician more frequently assigned a bronchiolitis diagnosis (63% and 57%, respectively) when compared with the NP group (54%, P<0.001). Subgroup analysis on the ICD code identified 2,233 clinical impressions for bronchiolitis. Physicians, when working alone or in combination, more frequently assigned an acute bronchiolitis diagnosis due to RSV (47% and 56%, respectively, P<0.001) and acute bronchiolitis due to other infectious organisms (27% and 27%, respectively, P<0.001). NPs more frequently assigned an acute bronchiolitis diagnosis, unspecified (35%, P<0.001).

There was a significant difference among NP, physicians, and the NP/physician combination for ordering chest radiographs, RSV testing, bronchodilators, and corticosteroids (P≤0.001) (Table 1). There was no difference in viral testing, antibiotics usage or ordering pattern of any ToT among providers. The Total ToT was significant among providers (P<0.001), with physicians ordering a median of 2 ToT (IQR 1–3) compared with the NP group (1 ToT, IQR 1–2) or the NP/physician combination (1 ToT, IQR 1–3). Overall, NPs were less likely to give a corticosteroid (OR 0.327 [0.271–0.395, P<0.001]) or admit a patient (OR 0.035 [0.023–0.052, P<0.001] than physicians (Table 2). These relationships remained significant when adjusted by era.

Table 1.

Distribution of demographics, tests, treatments and disposition of patients with bronchiolitis among nurse practitioners and physicians in the pediatric emergency department.

Variable	Total (N=5,260) n (%)	Nurse practitioner (N=1,598) n (%)	Physician (N=3,435) n (%)	Both (N=227) n (%)	P-value
Age group (months)
1–3	581 (11)	59 (4)	489 (14)	33 (15)	<0.001
4–12	2593 (49)	845 (53)	1633 (48)	115 (51)
13–24	2086 (40)	694 (43)	1313 (38)	79 (35)
Diagnostic test
Chest radiography	1959 (37)	588 (37)	1259 (37)	112 (49)	0.001
Viral testing	1145 (22)	350 (22)	739 (22)	56 (25)	0.530
RSV test	1020 (19)	205 (13)	762 (22)	53 (23)	<0.001
Treatment
Bronchodilator	2970 (57)	874 (55)	1943 (57)	153 (67)	0.001
Corticosteroid	1013 (19)	147 (9)	812 (24)	54 (24)	<0.001
Antibiotics	455 (9)	123 (8)	305 (9)	27 (12)	0.079
Any ToT	3284 (80)	946 (80)	2158 (80)	180 (86)	0.105
Total ToT*	2 (1–3)	1 (1–2)	2 (1–3)	1 (1–3)	<0.001
Disposition
Admission	1178 (22)	25 (2)	1072 (31)	81 (36)	<0.001
Discharge	4070 (77)	1571 (98)	2353 (69)	146 (64)
LOS (minutes)*	144 (91–206)	137 (105–170)	149 (61–226)	182 (93–285)	<0.001

RSV=Respiratory syncytial virus; LOS=Length of stay; ToT=test or treatment

^*Median (interquartile range [Q1–Q3])

Table 2.

Comparison between nurse practitioners and physicians in the use of resources for the testing and treatment of bronchiolitis in the pediatric emergency department.

Variable	Odds ratio (CI 95%)	P-value
Chest radiography	1.006 (0.890–1.138)	0.925
Viral testing	1.023 (0.886–1.181)	0.769
Bronchodilator	0.927 (0.823–1.045)	0.222
Corticosteroid	0.327 (0.271–0.395)	<0.001
Antibiotics	0.856 (0.688–1.065)	0.175
Any ToT	0.996 (0.839–1.182)	0.965
Disposition	0.035 (0.023–0.052)	<0.001

ToT=test or treatment

We identified 3,306 medical visits in the early and 1,954 in the late eras. There was a significant change in the distribution of age groups in the early and late (P=0.018) eras. However, there was no difference in the distribution of age groups among NP and the NP/physician combination after the CPG implementation. Among physicians, the 4–12 months age group was the most frequently seen (48%) followed by the 13–24 months and 1–3 months age groups (38% and 14%, respectively). We found no statistical difference between eras in the LOS, or disposition among providers. A decrease in the assigned wheezing diagnosis among NPs (from 60% to 41%, P<0.001) and physicians (47% to 36%, P<0.001) was noted in the late era. No statistical difference was noted in the diagnosis of wheezing for the NP/physician combination across eras. An increase in the assigned diagnosis of bronchiolitis was noted among NPs (from 48% to 63%, P<0.001) and physicians (from 61% to 67%, P<0.001), but not in the NP/physician combination (P=0.697). Subgroup analysis on the ICD code between eras showed a significant difference among NPs (20%), physicians (75%), and the NP/physician combination (5%). NPs decreased the diagnosis of acute bronchiolitis due to other infectious organism (from 31% to 8%), with a concomitant increase in acute bronchiolitis, unspecified (from 25% to 49%) (P<0.001). The NP/physician combination behaved similarly across eras, with a significant decrease in the diagnosis of acute bronchiolitis due to RSV and due to other infectious organism (P=0.002). Physicians, however, had an increase in the acute bronchiolitis due to RSV (45% to 49%) and acute bronchiolitis unspecified (from 17% to 41%), and a decrease in acute bronchiolitis due to other infectious organism (from 37% to 10%) (P<0.001).

There was a significant decrease in chest radiograph orders by physicians between the early and late era (39% and 33%, P=0.001, respectively), but there was no statistical change among NPs or the NP/physician combination (Table 3). NPs and physicians significantly reduced RSV testing and bronchodilators across eras. Only physicians showed a statistical change in the use of corticosteroids (27% to 19%, P<0.001). There was no statistical difference among providers between eras in the use of viral testing or antibiotics. The ordering pattern of any ToT significantly decreased among NPs and physicians across eras (P<0.001). The Total ToT was significant among NPs and physicians (P<0.001), with a decrease among physicians, ordering a median of 1 ToT (IQR 0–2) in the late era. Overall resource utilization in the NP/physician combination was not statistically different across eras (86% to 92%). NPs and physicians were less likely to perform any ToT or administer a bronchodilator after the CPG implementation (Table 4). Physicians were also less likely to order a chest radiography (OR 0.772 [0.667–0.894, P=0.001]) or corticosteroids (OR 0.630 [0.531–0.749, P<0.001]). The Breslow-Day homogeneity of odd’s ratio was used to determine if the change in practice by era was a confounder in the analysis between providers, which was found to be non-significant. Median RS analysis was significantly different among NPs (1, IQR 0–2), physicians (3, IQR 2–4), and the NP/physician combination (2, IQR 1–3) (P<0.001). Overall, patients who were admitted had a median RS of 3 (IQR 2–5, P<0.001). RS analysis of the diagnoses of wheezing or bronchiolitis was non-significant. The use of a RS was found to be significant in patients with a chest radiography (RS 2, IQR 1–4, P=0.042), viral testing (RS 2, IQR 1–4, P=0.016), bronchodilator (RS 2, IQR 1–3, P<0.001), and corticosteroid (RS 3, IQR 2–4, P<0.001). The RS was non-significant in the use of antibiotics. Smoking status was significant for Total ToT for never smoker (2, IQR 1–2, P<0.001), passive smoke exposure-never smoker (2, IQR 1–2, P<0.001) and current daily smoker (1, IQR 1–2, P<0.001). Smoking status was not significant in the RS or use of a bronchodilator.

Table 3.

Distribution of selected variables between nurse practitioners and physicians during the early and late era in the testing and treatment of bronchiolitis in the pediatric emergency department.

Variable	Nurse practitioner			Physician
	Early era (N=910) n (%)	Late era (N=688) n (%)	P-value	Early era (N=2,199) n (%)	Late era (N=1,236) n (%)	P-value
Diagnostic test
Chest radiography	350 (39)	238 (35)	0.116	853 (39)	406 (33)	0.001
Viral testing	207 (23)	143 (21)	0.36	474 (22)	265 (21)	0.966
RSV testing	153 (17)	52 (8)	<0.001	514 (23)	248 (20)	0.026
Treatment
Bronchodilator	588 (65)	286 (42)	<0.001	1372 (62)	571 (46)	<0.001
Corticosteroid	90 (10)	57 (8)	0.295	583 (27)	229 (19)	<0.001
Antibiotics	71 (8)	52 (8)	0.925	187 (9)	118 (10)	0.317
Any ToT	755 (83)	191 (71)	<0.001	1804 (82)	354 (72)	<0.001
Total ToT*	1 (1–2)	1 (0–2)	<0.001	2 (1–3)	1 (0–2)	<0.001

RSV=Respiratory syncytial virus; ToT=test or treatment

^*Median (interquartile range [Q1–Q3])

Table 4.

Comparison between nurse practitioners and physicians during the late vs early era in the testing and treatment of bronchiolitis in the pediatric emergency department.

Variable	Nurse practitioners OR (CI 95%)	Physicians OR (CI 95%)
Chest radiography	0.846 (0.689–1.040)	0.772 (0.667–0.894)*
Viral testing	0.891 (0.700–1.134)	0.993 (0.838–1.177)
Bronchodilator	0.390 (0.318–0.478)*	0.518 (0.449–0.596)*
Corticosteroid	0.823 (0.581–1.165)	0.630 (0.531–0.749)*
Antibiotics	0.966 (0.666–1.402)	1.136 (0.892–1.446)
Any ToT	0.490 (0.358–0.670)*	0.558 (0.446–0.698)*
Disposition	2.012 (0.898–4.506)	1.022 (0.879–1.188)

ToT=test or treatment

^*P-value <0.05 as statistically significant.

DISCUSSION

Our study showed significant differences in the use of ToT among providers and across eras after the institutional CPG implementation for the diagnosis and management of children with bronchiolitis. Our findings indicate that there was a significant difference in the rate of use of chest radiography, RSV test, bronchodilator, and corticosteroids among providers; but overall, only the NP group was less likely to use corticosteroids in the treatment of patients with bronchiolitis. The implementation of a clinical pathway in the pediatric ED was successful in decreasing the levels of resource utilization in both NPs and physicians when diagnosing and managing bronchiolitis patients.^10,11 After the CPG implementation, NPs were less likely to use bronchodilators, while physicians were less likely to use chest radiographies, bronchodilators, and corticosteroids.

These findings are consistent with earlier observations, which report the improvement of patient outcomes in emergency and critical care settings in cases that incorporated NP services.^12–14 However, a study by Pines et al.¹⁵ found a broadly consistent pattern regarding emergency physicians and APPs within the same ED, with similar levels of observed care and test utilization across both groups. This overall comparison in practice patterns may be more accurate due to larger samples and confounder controls, but no ED studied contained the independent APP practice free of an emergency physician’s supervision, a limitation that our study overcomes through the inclusion of the NP as a separate group for comparison.

Our findings confirm a prior study showing NPs order fewer diagnostic tests than physicians in the ED setting.¹⁶ The unchanged practice by NPs in the use of chest radiographs after the CPG implementation may be due to lack of competence regarding the knowledge of respiratory sounds associated with bronchiolitis or need for confirmatory testing of the absence of pneumonia.

Separately, the NP and physician groups saw further changes in the forms of a decrease in diagnosis of wheezing and an increase in diagnosis of bronchiolitis. The adjustment in diagnosis could reflect the institutional CPG implementation. A standardized approach for diagnosis and clear clinical definitions could have improved the providers’ ability to diagnose patients, subsequently leading to better billing and coding. Our study indicates that the physician group and the NP/physician combination were more likely to diagnose bronchiolitis, with the NP group demonstrating a low frequency of such assignment. Subgroup analysis based on ICD code of the bronchiolitis type attributed the acute bronchiolitis due to RSV and the acute bronchiolitis due to other infectious organisms diagnoses to the two former groups. This distribution can be ascribed to the differing ordering patterns among NP, physicians, and the combined group. NPs more frequently diagnosed acute bronchiolitis, unspecified, for which they were least likely to perform RSV testing, further emphasized in the late era. Physicians and the NP/physician combination most frequently ordered RSV testing, which explains the type of diagnosis assigned by them. The cause of increased RSV bronchiolitis diagnoses in the later era is unclear because it was not accompanied by a significant increase in RSV or viral testing, but patients seen by a physician were most likely to be admitted. A study from our group previously noted that an influencing factor affecting physician practice compliance with viral testing was the requirement for epidemiologic surveillance and/or cohorting¹⁰. The specific use of viral testing as an evaluation technique was negligible both among providers and across the eras. However, a cross-reference analysis on the use of an RSV test in correlation with the bronchiolitis ICD codes between eras showed an inconsistency between the number of RSV tests performed and the assigned ICD code, suggesting incongruous coding/billing or missing information in the database. The performance of a viral test outside of the hospital setting may also have afforded the etiologic agent for bronchiolitis, but we lacked the results of such tests, and therefore cannot make further conclusions.

The use of any ToT was found to be significantly increased among the NP/physician combination group in cases with higher severity and/or complexity. The physicians’ significantly higher Total ToT, therefore, may be related to the Emergency Severity Index (ESI).¹⁷ Acutely ill patients classified in more urgent groups (ESI 1–3) could have been more likely to be treated by a physician; therefore the increased resource utilization may be linked to the patients’ inferred augmented severity.¹⁸ The lower-acuity patients classified into higher ESI levels (ESI 4–5) are typically seen by an NP and require less resource utilization¹⁹; hence the decreased likelihood of an NP to order ToTs or admit a patient. Thus, hospitalization is an indirect indicator of bronchiolitis severity.²⁰ This trend was consistent across eras, and analysis reflecting the physicians’ elevated admission rate as a result of an inferred higher level of acuity also remained unvaried in light of the CPG. However, changes in provider efficacy and efficiency, as measured by LOS, may be impacted by hospital capacity, crowding in the ED, and disease seasonality, which were not accounted for in this study.^21–22

After the CPG implementation, the NP and combined groups had no changes to their patients’ age group distribution, but the physicians had a growth in their patients within the 4–12 months and 13–24 months age groups. This shift in age distribution could be a result of pathway standardization, leading to the treatment of younger patients by the NPs or a higher level of acuity in the younger population. The RS distribution suggests that the physicians’ patients exhibited higher degrees of respiratory distress. The RS did not play a role in the providers’ assignment of diagnosis for wheezing or bronchiolitis. Subgroup analysis showed that patients with a wider range and higher RS received more chest radiographies, viral testing, corticosteroids, and bronchodilator use. This is consistent with our previous work¹⁰, where we noted that another influencing factor that affects physician practice compliance with the use of the above-mentioned ToT is disease severity. The RS in relation to the ToT among groups was non-significant, which may be provider-dependent or coincidental and does not drive practice or causality.

Limitations

As with any retrospective analysis, there are several limitations. We performed a secondary analysis of aggregated data to evaluate the standard of care within the pediatric ED. We used the admission-data level based on ICD-9 and ICD-10 codification, which may have overestimated or underestimated bronchiolitis episodes. The lack of individual patient characteristics, demographics, and clinical information, such as ESI or cost savings, prevents us from further analysis. Our study lacks information about the appropriateness of the diagnostic testing or treatment which may be related to disease severity. Our study assumes that the changes seen in the early and late era were due to the CPG implementation, but cannot exclude that they may have been due to changes in pathophysiology of disease over time. Furthermore, we lack information on patient/parental satisfaction, return rate, and parental preference of provider, which may have affected patient outcomes. Lastly, we lack the NP level of training or years of experience that may have affected resource utilization or the criteria by which they partner with a physician, and may only make inferences based on their practice.

Conclusion

Resource utilization in the diagnosis and treatment of bronchiolitis in the pediatric ED remains variable. The presence of NPs in the ED permits specialized, cost-effective care, as exemplified by reduced diagnostic testing and LOS. The implementation of a clinical pathway on the diagnosis and management of bronchiolitis successfully decreased the use of non-recommended tests and therapies among medical providers.