Association of Asthma Exacerbation Risk and Physician Time Expenditure With Provision of Asthma Action Plans and Education for Pediatric Patients

Abstract

OBJECTIVE

To provide information about factors underlying provision of asthma action plans (AAPs) to a minority of pediatric patients with asthma, assess whether risk of exacerbation acts on provision of AAP and asthma education directly, suggesting targeting to highest-risk patients, or indirectly by influencing physician-patient interaction time.

METHODS

This study was a retrospective cross-sectional analysis of a nationally representative sample of physician office visits that consisted of patients aged 2 to 18 years with asthma. Exacerbation risk comprised proxy indicators of control and severity. Direct and time-mediated effects of exacerbation risk on provision of AAP and education were calculated from logistic regression models.

RESULTS

Asthma action plans were provided in 14.3% of visits, education in 23.9%. Total direct effects of exacerbation risk (ORs = 3.88–4.69) far exceeded indirect, time-mediated effects (both ORs = 1.03) on AAPs. Direct effects on education were similar but smaller. After adjusting for risk, physician time expenditure of ≥30 minutes was associated with nearly doubled odds of providing AAP or education (ORs = 1.90–1.99). Visits that included allied health professionals alongside physician care were significantly associated with all 4 outcomes in multivariate analyses (ORs = 3.06–5.28).

CONCLUSIONS

Exacerbation risk has a strong, direct association with AAP provision in pediatric asthma, even controlling for physician time expenditure. Provision of AAP and education to pediatric patients with asthma may be facilitated by increasing available time for office visits and involving allied health professionals.

Introduction

Asthma is a chronic inflammatory disorder of the pulmonary airways that affects approximately 7.0 million US children and adolescents,¹ with elevated rates of disease prevalence and severity among those who are Black or African American, or of lower socioeconomic status.^1–3 Education of patients and parents on environmental trigger avoidance, medication delivery, and exacerbation management is a core component of evidence-based asthma care^4,5 but is typically used in only a minority of patients.^2,3,6–8

Asthma action plans (AAPs), which provide written, structured guidance on recognition and management of symptom progression, are among the least frequently used strategies. Among US children or adolescents with asthma, the lifetime rate of AAP provision increased from 42% in 2002 to 51% in 2013.^2–9 However, AAP provision rates for single office visits or years are much lower: 22% to 25% of visits among urban children aged 2 to 12 years with persistent asthma^6–10 and 2% to 9% per year among primary care patients aged 5 to 18 years, including those not receiving regular care.⁸ Consistent with these low rates of AAP provision, 16% of primary care physicians and 31% of asthma specialists reported in the 2012 National Asthma Survey of Physicians that they “almost always” provide an AAP in office visits for asthma,¹¹ and 37% of asthma specialists reported “sometimes” or “never” using AAPs.¹²

Available information about factors that facilitate AAP provision is limited but suggests positive associations of AAP use with Northeast US region¹³ and care by a pediatrician.^7,14 Evidence about the effects of computerized decision support (CDS) guideline-compliance reminders on AAP provision is inconsistent.^15–19 Evidence about racial disparities is also inconsistent. In one national survey, parents of non-Hispanic Black children were more likely than those of non-Hispanic White children to report “ever” receiving an AAP (58% vs 47%, respectively), even after controlling for asthma severity.⁹ In contrast, in a sample of asthma hospitalizations for patients aged 2 to 17 years, non-Hispanic Black race was associated with a 48% reduction in the odds of receiving an AAP at discharge, compared with non-Hispanic White race.²⁰

One possible reason for these inconsistencies is that previous research on AAP provision has not controlled for available physician time. A commonly raised concern is that “competing demands” of numerous mandated or recommended disease-state management or preventive medicine interventions^21,22 make it difficult to find time for AAP or other asthma education.²³ Lack of time was reported as a barrier to smoking-cessation counseling for parents or children with asthma by 50% of pediatricians in one survey.²⁴ Longer visit duration was associated with provision of asthma education in an analysis of asthma office visits made from 2001–2006.²³

A related possibility is that physicians “triage” AAPs to patients with the greatest need for them.²¹ Although inconsistent with Centers for Disease Control and Prevention (CDC) guidance that AAPs should be used by “everyone with asthma,”²⁵ this possibility is consistent with a National Asthma Education and Prevention Program guideline indicating that, although recommended for all those with asthma, written AAPs are “particularly recommended” for patients whose asthma is moderate or severe persistent, has caused severe exacerbations, or is poorly controlled.⁵ One study of pediatric visits by Yee et al⁶ found that AAPs and other preventive measures were more likely to be provided to children whose symptoms were more frequent, debilitating, or required oral steroid use, because these patterns of symptoms indicate reduced asthma control, worsening lung function, and greater exacerbation risk.²⁶

An important question not addressed in previous research is the degree to which the effects of exacerbation risk observed by Yee et al⁶ are mediated by the amount of time the patient spends with the physician. A significant direct association of exacerbation risk with AAP or education provision in a statistical model that includes visit time as a mediating variable would indicate that risk is an independent determinant of preventive service provision, suggesting a possible triaging effect. In contrast, a finding of minimal direct effect and a significant mediated effect would suggest that risk increases the likelihood of receiving an AAP or other education primarily by increasing visit duration. Quantifying these effects could inform efforts to increase use of AAPs and education. To address this gap in available information, the purpose of this study was to identify predictors of AAP and education provision for US children and adolescents with asthma, focusing on the direct and time-mediated indirect effects of asthma exacerbation risk.

Methods

Study Design and Data Source. The study was a retrospective, cross-sectional analysis of office visits made in 2013–2016 by pediatric patients aged 2 to 18 years who were diagnosed with asthma, either at the sampled visit or previously; were seen by a non-surgical physician in primary care (i.e., excluding obstetricians/gynecologists, cardiologists, dermatologists, urologists, psychiatrists, neurologists, and ophthalmologists); and were not sent from the office to emergency or hospital care. Data were derived from the National Ambulatory Medical Care Survey (NAMCS), a publicly available, nationally representative assessment of US office-based physician visits.²⁷ The NAMCS has been used previously to study care provided to pediatric patients with asthma^14,28,29 and by the CDC to quantify the national prevalence and content of asthma care, including provision of AAPs and other asthma education.³⁰ The National Center for Health Statistics (NCHS), which conducts the survey annually, uses a multistage, stratified cluster-randomized sampling design.²⁷ In the first stage, the NCHS identifies geographic areas (e.g., counties, cities) and samples physicians from those areas, using the national master lists of the American Medical Association and the American Osteopathic Association.²⁷ These lists, which are comprehensive and not limited to organizational members, are stratified by specialty. For each chosen physician, the NCHS samples 1 week of visits from the 52 weeks of the year.²⁷ In the final stage, the NCHS samples visits for the chosen physician and week: 100% of visits if the office is small, or a random sample if the office is large.²⁷ The NCHS excludes visits made to federal facilities, telephone or administrative contacts (e.g., prescription refills), and physicians who do not provide direct patient care (e.g., radiologists).²⁷ During the present study period, survey response rates ranged from 46% to 59%.^31–34 After accounting for ability to provide data, 30% to 39% of sampled physicians contributed at least 1 office visit record. Weights provided in the dataset adjust for the multistage clustered sampling design, stratification, and survey non-response, producing nationally representative estimates.²⁷

Study Variables Collected. Data are collected annually from patient medical records, either paper or electronic, by US Bureau of the Census field representatives using an automated, laptop-based tool and standardized instructions.^31–34 Collected data fields include characteristics of the physician and office, the patient, and the visit. Characteristics of physicians and offices accessed for the present study were use of CDS reminders for guideline-based care; office geographic region; physician specialty; and whether the patient also saw an allied health professional, including a nurse practitioner or physician's assistant. The measure of visit time recorded in the NAMCS reflects time spent with the physician. No measure of time spent with allied professionals was available.

Patient data accessed for the present study included demographics: age, race or ethnicity, sex, and Medicaid coverage.³⁵ Visit data included diagnoses (up to 3 in 2013, up to 5 in 2014–2016) in the International Classification of Diseases (ICD)-9 format in 2013–2015 and ICD-10 format in 2016; ^{31–34,36,37} reasons for the visit; whether the major reason was routine; a complete list of medications newly prescribed or continued at the visit coded in Lexicon Plus, a proprietary system licensed to the NCHS by Cerner Multum; and asthma-related metrics.^31–34

The asthma-related metrics included level of disease control (well controlled, not well controlled, or very poorly controlled); severity (intermittent, mild persistent, moderate persistent, or severe persistent); and the study outcomes, provision of AAP and education. Asthma control and severity ratings were obtained from the medical record. Asthma education was defined for data collectors as provision of information regarding inhaler use and elimination of asthma triggers, including allergens. Beginning in 2015, AAP was defined for data collectors as a written plan that provides instructions on medications, asthma control, and guidance for seeking care for asthma attacks.^31–34

Classification of Asthma Symptoms. The construct of asthma exacerbation risk in visits by patients aged 2 to 18 years with asthma diagnosis was defined by using proxy indicators associated with risk and/or provision of an AAP or education in previous research.^6,26 These included any one or more of the following: 1) asthma as the sole diagnosis for the visit; 2) asthma as the primary diagnosis coupled with an indicator that this was the first visit or a flare-up of a chronic condition; 3) a breathing problem, such as shortness of breath, recorded as the primary reason for the visit; 4) a recording of asthma as moderate persistent or severe persistent (rather than intermittent or mild persistent), not well controlled, or very poorly controlled; or 5) use of a medication indicating symptomatic asthma (systemic glucocorticoids, leukotriene modifiers, bronchodilator combinations, inhaled corticosteroids, omalizumab, or antiasthmatic combination)³⁸ in a patient for whom severity and control were not recorded, a common occurrence in pediatric asthma.²⁸

For bivariate analyses, a measure of highest risk was added, defined as either asthma that was not well controlled or poorly controlled, or severity recorded as severe persistent.⁶ This measure was not used in multivariate analyses because the number of patients meeting this criterion (n = 99) was too small to be statistically reliable.

Statistical Analyses. Bivariate analyses assessed the association of each predictive factor identified in the literature review with each of 4, non–mutually exclusive outcomes, all measured as binomials representing services received at the visit. The first two, AAP (with or without asthma education) and education (with or without AAP), are reported as measures of guideline-concordant care by the CDC.³⁰ Additionally, to provide a complete picture of educational activities documented in the medical record, 2 additional outcomes were both AAP and education and either AAP or education. Bivariate associations were assessed for statistical significance by using the Pearson chi-square test.

Multiple logistic regressions were run to produce adjusted estimates. Direct and time-mediated indirect effects of exacerbation risk on the outcomes were calculated from the models. For the calculations of direct and time-mediated effects, physician time was defined as a binomial, <30 minutes vs >30 minutes. This categorization was based on observed associations of visit time with the outcomes in bivariate analysis.

In logistic regression mediation analysis, the direct effect odds ratio is the exponentiated coefficient for the independent variable of interest (exacerbation risk in this study) in a covariate-adjusted model that includes the mediator (time in this study).^39,40 Indirect (mediated) effects may be calculated by using a variety of methods.^39,40 In this study, time-mediated effects were calculated by using the potential outcomes framework method described by Rijnhart et al⁴⁰ (2019; Supplemental Table S1). The technique assumes no confounding from missing covariates.⁴⁰

All statistical analyses were performed with an a priori significance level of 0.05 using IBM SPSS for Complex Samples (version 25.0; Armonk, NY) to adjust statistical tests for the homogeneity of variance associated with the NAMCS multistage, cluster-randomized design. Estimates were tested for statistical reliability, using standards published by the NCHS,^31–34,41 and meet these standards except where otherwise noted in the tables. Direct and indirect effects were calculated by entering coefficient values into Microsoft Excel (Redmond, WA) and applying the formulas for calculation of direct and indirect effects.⁴⁰

Results

Sample Selection. From 2013–2016, a total of 1901 visits made by US pediatric patients with asthma were included in the NAMCS sample (Table 1). After application of weights for the multistage sampling design and non-response, these represented more than 48 million office visits, or 10% of all visits made by those aged 2 to 18 years. After excluding patients seen by non–primary care physicians or sent directly from the office to the emergency department or hospital, the number of visits remaining in the sample was 1582, representing a weighted US population estimate of 41,635,031 nationwide. Of those, 636 (n = 16,274,799 weighted) met >1 criteria for elevated exacerbation risk. Mean (median) physician time was 23.0 (20.0) minutes for the sample overall; 22.5 (18.0) minutes for those with no exacerbation risk indicators; 23.7 (20.0) for those with elevated risk; and 31.6 (30) for the subset of patients at highest risk (i.e., asthma poorly controlled, not well controlled, or severe persistent).

Table 1.

Sample Selection Flowchart, Counts of US Physician Office Visits, 2013–2016

Characteristics	Unweighted Sample No.	Weighted Population*
Patients aged 2–18 yr	18,771	488,734,405
Those with asthma diagnosed either at the sampled visit or previously	1901	48,068,069
Whole sample after excluding inappropriate specialists† and patients sent to emergency department or hospital	1582	41,635,031
Elevated exacerbation risk subsample—meets ≥1 of the following in addition to asthma diagnosis‡
Asthma is the sole diagnosis at the visit	205	—
Asthma is the primary diagnosis, and this is a new problem or flare-up of chronic problem	249	—
Primary reason for visit is described as a breathing problem (e.g., shortness of breath)	44	—
Asthma severity recorded as moderate or severe	101	—
Asthma control is not well controlled or poorly controlled	88	—
Persistency and severity are both unrecorded and patient uses a medication indicating higher risk§	279	—
Unduplicated elevated exacerbation risk visits	636	16,274,799

^* US population estimates after application of sampling weights, accounting for complex clustered sampling design and non-response.

^† Surgeon, cardiologist, obstetrician or gynecologist, dermatologist, psychiatrist, or neurologist, or no physician was seen at the visit.

^‡ Counts shown are not mutually exclusive; patients could meet >1 criterion.

^§ Systemic glucocorticoids, leukotriene modifiers, bronchodilator combinations, inhaled corticosteroids, omalizumab, or antiasthmatic combinations.

Bivariate Findings. Asthma action plans were provided in 14.3% of visits overall and were accompanied by asthma education in 11.9% of visits (Table 2). Education was provided in 23.9% of visits, and either education or AAP in 26.3%. Receipt of care from an allied health professional alongside physician care was significantly associated with all 4 outcomes in bivariate analyses, with AAP and/or education provision rates of 42% to 56%. Additional significant predictors of receiving an AAP were Northeast region (27.6%); Hispanic ethnicity (21.3%); greater exacerbation risk (23.5% all elevated risk, 56.6% highest risk); and physician time of ≥30 minutes (26.1% for 30–39 minutes, 31.8% for ≥40 minutes). Like AAP provision, asthma education provision was significantly associated with greater exacerbation risk and physician time expenditure. The use of CDS reminders for guideline-based care was not significantly associated with any outcome.

Table 2.

Percentage of Visits Including Asthma Education and/or Action Plan, Patients Aged 2–18 yr With Asthma, NAMCS 2013–2016

Characteristics	Number of Cases		Percentage of Visits, %
	Unweighted	Weighted	Action Plan	Education	Both	Either
All	1582	41,635,031	14.3	23.9	11.9	26.3
Visits in 2013	738	10,889,701	12.7	18.5*	9.1*	22.1*
Visits in 2014	579	10,727,778	9.3	15.6*	6.3*	18.5*
Visits in 2015 or 2016	265	20,017,550	17.9†	31.4*,†	16.5*,†	32.8*,†
Sex
Female	667	18,380,308	16.5	27.2	14.2	29.5
Male	915	23,254,721	12.6	21.3	10.1	23.8
Age group, yr
0–5	369	8,963,542	12.5	28.7	10.8	30.5
6–12	732	19,802,481	15.1†	24.4	12.8†	26.7
13–18	481	12,869,007	14.3	19.8	11.4	22.7
Geographic region
Northeast	352	11,667,727	27.6**,†	37.5*†	24.1**,†	41.0*†
Midwest	411	7,390,870	12.3**	22.7*	11.4**	23.7*
South	438	13,601,578	7.8**,†	17.6*	6.6**,†	18.8*
West	381	8,974,855	8.5**,†	16.8*	4.6**,†	20.7*
Race or ethnicity‡
White, non-Hispanic	954	21,219,732	10.4**	20.6	8.7*	22.3
Black, non-Hispanic	252	7,956,935	8.9**	27.1	7.1*	28.8
Hispanic	276	10,104,285	21.3**	23.7	16.5*	28.5
Payment source
Private or other	940	24,061,601	12.3	20.3	9.7*	22.9
Medicaid	549	14,749,016	20.0	28.5	17.5*	31.0
Type of provider seen
Non-pediatrician	428	12,394,126	11.8	25.3	9.5	27.5
Pediatrician	1154	29,240,904	15.4	23.3	12.9	25.8
Non-physician§	120	5,626,163	43.2**,†	55.0**,†	42.4**,†	55.7**,†
Medications
None	593	17,366,936	13.4†	20.8†	11.2†	23.0†
Adrenergic bronchodilators	877	21,473,771	15.4	26.4	13.1	28.6
Corticosteroid¶	310	7,608,355	21.8	34.4	18.9	37.2
Leukotriene modifiers	217	5,241,742	18.5	24.3	13.1	29.7
Indicates potential exacerbation#	434	10,639,488	22.2*	32.2	17.9	36.5*
>1 therapy class	441	11,270,760	20.7	34.3*	17.6	37.4*
Exacerbation risk††
Lower	946	25,360,231	8.4**	18.5**	6.4**	20.5**
High but not poorly controlled	537	13,207,121	15.8**	25.5**	12.2**	29.2**
High and poorly controlled	99	3,067,677	56.6**,†	61.8**,†	56.5**,†	61.9**,†
Time spent with physician, min
<15	240	5,147,306	8.1*,†	10.5**	4.0**,†	14.6**
15–19	593	15,150,661	8.7*	16.3**	6.2**	18.8**
20–29	362	10,123,966	9.9*,†	22.3**	9.1**,†	23.1**
30–39	237	6,120,740	26.1*	34.8**	21.2**	39.8**
≥40	150	5,092,357	31.8*,†	50.3**,†	31.4**,†	50.6**,†
Electronic reminder system
No	358	7,373,300	14.5	22.7	10.7†	26.5
Yes	1159	33,166,588	14.6	24.9	12.5	27.0

NAMCS, National Ambulatory Medical Care Survey

^* p < 0.05.

^† Comparison does not meet 1 or more statistical reliability standards for proportional estimates; interpret result cautiously.

^‡ Imputed to account for missing data using a model-based regression method; those with other non-Hispanic race (n = 100, e.g., Asian, Native Hawaiian, American Indian) were excluded from race or ethnicity analysis.

Table 2.

Percentage of Visits Including Asthma Education and/or Action Plan, Patients Aged 2–18 yr With Asthma, NAMCS 2013–2016

^§ Nurse practitioner or physician's assistant.

^¶ Systemic glucocorticoids, inhaled corticosteroids, or antiasthmatic combination products including steroids.

^# Systemic glucocorticoids, leukotriene modifiers, inhaled corticosteroids, or omalizumab.

^** p < 0.01, Pearson chi-square test.

^†† Indicates any one or more of the following: asthma is the sole diagnosis for the visit; asthma is the primary diagnosis coupled with an indicator that this was the first visit or a flare-up of a chronic condition; primary reason for the visit is a breathing problem; asthma is moderate persistent or severe persistent (rather than intermittent or mild persistent); asthma is not well controlled or very poorly controlled; or patient for whom severity and control were not recorded uses a medication indicating higher risk. “Poorly controlled” (n = 99) indicates that asthma control is marked as “not well controlled” or “very poorly controlled” or asthma severity is marked as “severe persistent.”

Adjusted and Time-Mediated Analyses. Logistic regression models of AAP and education were consistent with bivariate findings (Table 3; Supplemental Table S2). Receipt of care from an allied health professional and asthma exacerbation risk were significant positive predictors of all outcomes. Hispanic ethnicity was a significant positive predictor of AAP provision but not of education. Physician time and Northeast region were significant positive predictors of all outcomes except the provision of both AAP and education. Neither non-Hispanic Black race or ethnicity nor the use of CDS reminders was a significant predictor in either model for any of the 4 outcomes. Measures of model quality suggested modest predictive accuracy and fit, with a Nagelkerke R² of 0.231 to 0.243 when predicting AAP and 0.147 to 0.164 when predicting education. Additionally, the relative standard error (standard error divided by the coefficient) for many adjusted estimates exceeded 30%, indicating limited statistical reliability and reflected in the wide confidence intervals for these estimates.

Table 3.

Logistic Regression-Adjusted Models of Asthma Action Plan Provision and Education

Equation	Equation 1. Dependent Variable Is Asthma Action Plan						Equation 2. Dependent Variable Is Education
	Without Mediator (Time)			With Time as Mediator			Without Mediator (Time)			With Time as Mediator
No. of cases		1333			1333			1333			1333
Nagelkerke R2		0.231			0.243			0.147			0.164
Model χ2(P)*		182.7 (.000)			192.9 (.000)			135.4 (.000)			151.7 (.000)
C-statistic		0.647			0.665			0.580			0.591
	Exp (B)	CI-L	CI-U	Exp (B)	CI-L	CI-U	Exp (B)	CI-L	CI-U	Exp (B)	CI-L	CI-U
Race or ethnicity†
White, non-Hispanic	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Black, non-Hispanic	1.13‡	0.59	2.18	1.15‡	0.59	2.22	1.31‡	0.81	2.12	1.32‡	0.81	2.16
Hispanic	2.52‡	1.42	4.48	2.64‡	1.43	4.84	1.04‡	0.56	1.94	1.08‡	0.57	2.02
Region
Northeast	2.38‡	1.20	4.73	2.11‡	1.02	4.38	2.05‡	1.12	3.76	1.87‡	1.01	3.44
All others	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Insurance§
Medicaid	0.74‡	0.37	1.46	0.72‡	0.35	1.44	0.90‡	0.53	1.53	0.88‡	0.52	1.51
All others	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Electronic reminders
No	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Yes	0.89‡	0.42	1.89	0.79‡	0.37	1.69	1.14‡	0.61	2.10	1.03‡	0.55	1.93
Non-physician¶
No	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Yes	4.39	1.94	9.92	3.50‡	1.37	8.91	4.69	2.03	10.85	3.78‡	1.59	9.01
Exacerbation risk
Lower	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Higher	4.06	2.06	7.98	3.88	1.95	7.72	1.95‡	1.23	3.09	1.87‡	1.18	2.94
Time, min
<30	Mediator—not included			REF	REF	REF	Mediator—not included			REF	REF	REF
≥30	Mediator—not included			1.90‡	1.04	3.48	Mediator—not included			1.90‡	1.25	2.91

CI-L, 95% confidence interval, lower limit; CI-U, 95% confidence interval, upper limit; LL, log likelihood; Ref, reference

^* χ² change value calculated by subtracting –2LL value at previous model step. All significance levels and confidence intervals account for the complex clustered sampling design.

^† Imputed to account for missing data using a model-based regression method; those with other non-Hispanic race (n = 100, e.g., Asian, Native Hawaiian, American Indian) were excluded from race or ethnicity analysis.

^‡ Relative standard error of the estimate exceeds 30%; interpret cautiously.

^§ Based on primary payer for the visit. Other payer type includes Medicare, private insurance, worker's compensation, self-pay, no charge, or other/unknown.

^¶ Nurse practitioner or physician assistant.

Coefficients for models that predicted AAP, either alone or with education, changed only slightly in models that included the time mediator, compared with those that did not (Table 3; Supplemental Table S2). This finding suggests only modest shared variance (i.e., statistical correlation) between exacerbation risk and visit time, and therefore minimal time-mediated effects of risk on provision of AAP or education. Consistent with this finding, the total direct effects of exacerbation risk (ORs = 3.88–4.69) far exceeded the indirect, time-mediated effects (both ORs = 1.03) in both AAP models (Figure, left side). For models that predicted education, either overall or with AAP, the direct effects again exceeded the indirect, time-mediated effects, but by a smaller amount (1.87–1.89 vs 1.03–1.04, respectively; Figure, right side). Nonetheless, after adjusting for exacerbation risk, visit time of >30 minutes was associated with a near doubling (ORs = 1.90–1.99) in models of all outcomes except both AAP and education.

Figure.

Mediation models of associations between asthma severity, visit time, and provision of asthma action plan and education.

Discussion

In this assessment of a national sample of office visits made by pediatric patients with asthma, uses of AAP and asthma education were uncommon, with only about one quarter of visits including the provision of either service. Covariate-adjusted modeling suggested that the primary effect of exacerbation risk on provision of AAPs and, to a lesser extent, education was direct, indicating that these services were targeted to patients with the greatest need for them. Physician time did not appear to mediate the association between risk and AAP or education; however, it was an independent predictor after accounting for the effects of risk.

The finding of a strong independent association of exacerbation risk with AAP provision, to a greater extent than asthma education provision, may reflect the purpose of the AAP, which is to teach patients and family members to recognize and appropriately respond to warning signs and asthma emergencies.⁵ A related possibility is that physicians perceive little benefit of an AAP for lower-risk patients and may therefore diverge from the recommendation to provide AAPs for all patients with asthma.^24,25,42,43 Consistent with this possibility are findings of physician surveys indicating low rates of agreement that AAPs are effective tools.^7,11,12 For lower-risk patients, physicians may perceive more benefit from general education on trigger avoidance and inhaler use, explaining the greater effect of exacerbation risk in the AAP models than in the education models. Detailed examination of these questions would be an important area for future research.

Independent of exacerbation risk, the associations of physician time expenditure and allied health professional involvement with AAP or education provision may suggest a need for restructuring of primary care provision or guidelines to provide realistic expectations for education to be provided during office visits. For example, in a time-allotment mathematical model, Yarnall et al²² found that delivery of all guideline-recommended prevention would require physicians to work 22-hour days and proposed sustainable, multidisciplinary team-based models. A study of one such program randomized pediatric practices to education by non-physician clinical staff or to a 12-month intervention delay and found the initiative increased education (56% vs 20% control) and AAP provision (29% vs 5%).⁴⁴ These results are consistent with the present study's finding of higher rates of AAP and education provision in visits that included allied health professionals. Although not included in the definition of allied health professionals used by NAMCS, pharmacists play an important role in the provision of asthma education and AAPs given their clinical expertise and ability to educate patients on medications.⁴⁵ Pharmacists are regarded as enablers of the provision of AAP and education,⁴⁶ positively impacting clinical outcomes in patients with asthma.^47,48

The finding that Hispanic ethnicity predicted provision of AAP, but not of education, is seemingly inconsistent with a previous study, also using NAMCS data, which found a positive association of Hispanic or Latino ethnicity and asthma education in pediatric and adult office visits from 2007–2010.¹⁴ That study was conducted before the NAMCS began measuring AAP provision. It is likely that at that time, when no AAP field was available, data collectors viewed AAP as one component of asthma education. Because education on triggers and medication usage is a language-based activity, the present study finding may indicate the benefit of the AAP color-coded (green, yellow, red) format when English is not the primary language.⁵ This possibility is consistent with ongoing studies of pictorial or video-based asthma education or AAPs in populations with language barriers to self-management.^49,50

Important limitations of this study should be noted. Foremost, analyses based on medical record information indicate factors potentially involved in medical decision-making but do not directly measure the underlying explanations for those decisions. Surveys or qualitative measurement, such as interviews or focus groups, would provide more detail on factors physicians take into consideration when making determinations about AAP and asthma education. Additionally, the possibility of confounding by unmeasured characteristics of physicians or patients exists. In particular, confounding by indication may partly explain the association between AAP or education provision and use of allied health professionals, as it is possible physicians referred patients to nurse practitioners specifically for the purpose of providing education. These study findings highlight the need for further research on use of allied health professionals in the provision of asthma care education. An additional potential limitation was lack of statistical reliability for some adjusted estimates, reflected in relatively wide confidence intervals in the multivariate models. As evidence on AAP and asthma education provision accumulates with additional years of NAMCS data collection, the estimates obtained in this study should be confirmed with a larger sample size. Despite these limitations, findings provide information about medical decisions that have been relatively unexplored in recent work.

Conclusion

Assessment of a national sample of pediatric office visits suggested physicians may target AAPs to higher-risk pediatric patients with asthma. Assessing and reporting the benefits of AAPs for lower-risk pediatric patients may facilitate AAP provision. The independent association of physician time expenditure with provision of AAP or education suggests the potential value of initiatives to expand team-based preventive care.

ABBREVIATIONS

AAP

asthma action plan

CDC

Centers for Disease Control and Prevention

CDS

computerized decision support

ICD

International Classification of Diseases

NAMCS

National Ambulatory Medical Care Survey

NCHS

National Center for Health Statistics