NAMCS is conducted annually by the National Center for Health Statistics (NCHS) to collect information about patient, clinician, and office visit characteristics. Data from the 2012 NAS supplement(12) to NAMCS were released in 2017. Clinician eligibility for NAMCS was determined by responses to the Physician Induction Interview (PII).(11, 13) Participating clinicians who responded affirmatively to the PII asthma screener question (“Do you treat patients with asthma?”), regardless of specialty, were included in the NAS. Since 2006, NAMCS has included visits to office-based physicians and a panel of CHCs with up to three physicians and/or midlevel clinicians sampled per CHC. Sample selection is designed to produce nationally representative estimates for both NAMCS and CHC visits. Starting in 2012, the office-based component of NAMCS was split from the CHC component to produce separate data files for visits to private physician offices and CHCs to increase flexibility in use of patient visit data.(11) However, patient visit data from either component could not be linked to physician responses to the NAS supplement. This analysis utilized the 2012 NAS file released by NCHS that included NAMCS office-based physicians, CHC physicians and CHC mid-level clinicians. The 2012 NAS file included specific physician/clinician survey weights provided by NCHS. The NAMCS physician sampling frame included non-federally employed physicians who were classified as being engaged primarily in office-based patient care by the American Medical Association or the American Osteopathic Association and included general/family practitioners, internists, pediatricians and obstetricians. Specialty physicians in anesthesiology, radiology, and pathology, and those over age 85 years were excluded. Allergists and pulmonologists were oversampled to provide a sufficient sample size of asthma specialists for the NAS supplement. The CHC sampling frame included physicians and mid-level practitioners (i.e., physician assistants, nurse practitioners, and nurse midwives) from sampled CHC delivery sites. No asthma specialists were sampled in the CHC-based portion of the NAS. The NCHS Institutional Review Board approved NAS and informed consent was obtained from participating clinicians.

The unweighted and weighted response rates for the overall combined NAS sample were 38% and 28%, respectively, similar or higher than other national physician surveys.(14) The weighted response rate was higher for CHC providers (73%) than for NAMCS physicians (26%). Of the 1726 respondents, 49 were specialists unlikely to directly manage asthma, leaving 1677 eligible participants. Seventeen records were missing demographic data and were excluded from the sample. Clinician race/ethnicity was not included in the NAMCS questionnaire used in CHCs. Non-clinical respondents were also excluded (n=15). The final sample of 1645 included 1412 primary care clinicians (primary care physicians from the office-based and CHC samples and CHC mid-level practitioners from the CHC sample) and 233 asthma specialists (allergists and pulmonologists from the office-based sample). Available information on demographic characteristics included specialty, age, sex, and Census practice region, level of practice urbanization, practice ownership, and age of patient population.

Outcome variables were categorized into the four EPR-3 cornerstones of care: assessment and monitoring, patient education, environmental control, and pharmacologic treatment (Table E1). Clinician agreement with and adherence to specific EPR-3 recommendations and self-efficacy defined as clinician confidence in their ability to competently implement specific EPR-3 recommendations were determined by self-report (Table E2). Missing responses were low (0.01%-2.0%) and were excluded for individual outcomes.(15)

The association between adherence and overall agreement and overall self-efficacy with providing guideline-based care were assessed using two index variables. An agreement index variable was defined dichotomously as a response of “strongly agree” versus all other responses to all 5 questions about agreement (Table E2). Similarly, a self-efficacy index variable was defined as a response of “very confident” versus all other responses for all 5 questions about self-efficacy (Table E2).

Descriptive statistics were used to summarize clinician characteristics and outcome measures using provided sample weights to calculate nationally representative estimates. Differences between primary care clinicians and asthma specialists in distributions across response categories for guideline adherence outcomes were assessed using chi-square tests, with P<.05 (2-sided) considered statistically significant. Thus, P values in the text and tables reflect differences across the range of Likert scale response categories (e.g., “never (0% of the time)” to “almost always (75% to 100% of the time)” in Table II) rather than differences for one particular response category. Separate logistic regression models assessed the impact of agreement and self-efficacy on guideline adherence for each outcome that could be dichotomized to “almost always” versus all other responses. For some covariate categories, there were zero cells for asthma specialists. Therefore, the simplest models with the control indices for agreement and self-efficacy as the only independent variables are reported. Multivariate results that include additional covariates but that omit covariates with zero counts for any category are shown in the supplement. National estimates were calculated using NAS sample weights that accounted for the probability of clinician selection and non-response. Estimation of standard errors (SE) and statistical inference took the complex survey design into account. Analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC) and SUDAAN 11.0 (RTI, Research Triangle Park, NC). Estimates with a relative standard error >30% (SE/estimate) are flagged to indicate that these estimates have lower precision.

EPR-3 recommends that clinicians assess asthma impairment (symptom frequency and asthma control) and risk for adverse outcomes (hospitalizations, ED visits, and courses of systemic corticosteroids) during clinic visits. A higher percentage of asthma specialists almost always assessed all measures of asthma impairment than primary care clinicians, including a patient’s ability to engage in normal daily activities (84.5% versus 48.4%), frequency of daytime symptoms (91.1% versus 56.0%) and nighttime awakenings (81.7% versus 53.4%), and patient’s perception of control (70.7% versus 50.7%) (P<0.001 for distributional comparison across the Likert scale for all comparisons) (Table II). Most primary care clinicians (72.3%) asked about frequency of rescue inhaler use, albeit less frequently than asthma specialists (90.6%; P<.001). In contrast, use of a control assessment tool (e.g., Asthma Control Test, Asthma Control Questionnaire, Asthma Therapy Assessment Questionnaire, or similar tool) was low in both groups. More than half of the specialists (51.0%) and 69.7% of the primary care clinicians reported that they either sometimes or never used such tools. Indeed, 39.7% of primary care clinicians and 28.9% of specialists reported that they never used control assessment tools (data not shown).

The assessment of risk also differed between the two groups (Table II). While 86.8% and 81.9% of asthma specialists reported almost always asking patients about oral steroid use and emergency department (ED)/urgent care visits, fewer primary care clinicians (52.9% and 56.2%, respectively; P<.001 for comparison between clinician groups) almost always assessed these risk factors for adverse outcomes.

Both groups reported low frequencies of objective asthma assessment and monitoring. Only 11.2% of primary care clinicians and 12.8% of specialists almost always asked about home peak flow results (P=.34). Specialists were more likely to report almost always performing spirometry than primary care clinicians (e.g., 44.7% versus 10.8%; P<.001). The 64.1% of primary care clinicians who reported that they sometimes or never performed spirometry was made up of 36.3% who sometimes and 27.8% who never performed spirometry.

Asthma specialists were more likely to report assessing daily controller use for persistent asthma than primary care clinicians (91.7% versus 59.5% respectively, P<.001). Repeated assessment of inhaler technique was less frequently reported by both groups: 39.7% of asthma specialists reported almost always assessing technique versus 16.8% of primary care clinicians (P<.001).

Guideline-recommended patient education items covered in the survey included the frequency of providing asthma action plans (Likert scale), and whether or not the clinician provided trigger and risk education, inhaler use observation and advised changing home and work environment (Table III). Although written asthma action plans can improve asthma-related outcomes,(1, 16, 17) only 30.6% (SE 3.6) of specialists and 16.4% (SE 1.6) of primary care clinicians used them almost always (P <.001), and 6.1% (SE 2.3, RSE>30%) and 17.6% (SE 1.8) never used them, respectively (P <.001) (Table III, Panel A). Almost all specialists reported providing patient education regarding asthma symptom recognition, avoiding risk factors, inhaler technique and changing the home/work environment. Primary care providers also reported providing patient education in these areas with a high frequency but with a lower frequency than specialists, especially for inhaler use observation (Table III, Panel B).

Both groups reported patient concerns and misunderstandings about asthma pharmacologic therapies (Figure 1). Both groups reported patients sometimes or often misunderstood medication risks, were concerned about the side effects of inhaled corticosteroid (ICS) therapy, and were confused between rescue and controller medications. Specialists, however, more often than primary care clinicians reported that patients were almost always concerned about long-term ICS effects (8.6% versus 5.1%; P=.002) while primary care clinicians more often reported that patients were almost always confused rescue and controller medications, although the estimate for asthma specialists had an RSE>30% (11.7% versus 4.0%; P=.004).

Control of environmental factors can reduce asthma morbidity.(18) Asthma specialists were more likely to assess environmental triggers at home, school and/or workplace than primary care clinicians (Table IV). Specialists more often (67.3%) reported performing allergy testing “almost always” or “often” versus 24.8% of primary care clinicians (P<.001). Nearly a third (30.4%, SE 2.2) of primary care clinicians reported never performing these tests versus 5.3% (SE 2.1) of specialists (P<.001) (data not shown).

Recommendations for control of environmental factors differed between specialists and primary care clinicians. Primary care clinicians were less likely to recommend dust mite, mold and pest control measures than specialists but almost 60% of all clinicians recommended removing pets from homes with pet-sensitive patients. While both clinician groups provided recommendations on pollen avoidance, recommendations regarding cooking appliances were infrequent (i.e., 44.7%-53.1% of clinicians sometimes or never gave recommendations). The majority of specialists and primary care clinicians (>83%) recommended environmental tobacco smoke (ETS) avoidance, whereas air pollution avoidance was more frequently recommended by specialists. The biggest differences between primary care clinicians and specialists were observed for recommendations regarding air pollution avoidance. Specialists more often recommended air pollution avoidance for most patients than primary care clinicians (63.2 % vs 44.8%; P<.001). In addition, more primary care clinicians sometimes/never recommended air pollution avoidance as compared to specialists (21.6% vs 6.8%; P<.001).

Self-reported medication prescription for both groups was consistent with EPR-3 recommendations (e.g., short-acting beta agonists for symptom relief, and ICS for difficult to control asthma, add-on daily control and long-term control) (Figure 2). Asthma specialists reported greater use of other asthma medications than primary care clinicians (e.g., the percentage reporting never using specific medications was 16.6% of specialists versus 42.2% of primary care providers for long-course oral steroids, 14.1% versus 83.3% for Omalizumab, 45.9% versus 75.3% for methylxanthines, and 13.7% versus 30.6% for anticholinergics). Both groups prescribed short courses of oral steroids for asthma exacerbations, but specialists were less likely to prescribe short course oral steroids for symptom relief (43.2% versus 56.9%), and more likely to prescribe this medication for difficult-to-control asthma (60.7% versus 45.1%) and as add-on daily therapy (12.1% versus 5.3%) (P<0.05 for all comparisons).

More asthma specialists than primary care clinicians agreed strongly with the EPR-3 recommendations regarding spirometry for asthma diagnosis (77.6% versus 35.5%), the effectiveness of inhaled corticosteroids (ICS) for persistent asthma (76.0% versus 48.3%), twice yearly follow-up visits for persistent asthma (68.8% versus 48.9%) and assessment of asthma severity for initial treatment (79.3% versus 50.3%) (P<0.001 for all comparisons) (Table V, Panel A). In contrast, less than half of specialists and primary care clinicians strongly agreed that asthma action plans are effective, but strong agreement was still higher among specialists (41.0% versus 30.6%, P=.026). Strong agreement with all five guideline recommendations included in the survey was higher among specialists than primary care clinicians (27.9% versus 12.1%, P<.001).

Asthma specialists reported higher self-efficacy with providing guideline-recommended care than primary care clinicians (Table V, Panel B). Specialists were very confident in using spirometry (92.8%), in assessing severity (81.3%), in prescribing inhaled corticosteroids (91.1%), and in stepping up or down therapy (89.5%, 87.0%), while percentages were lower among primary care clinicians (37.0%, 49.4%, 65.2%, 64.5%, and 49.8%, respectively; P <.001). Accordingly, self-efficacy for all five recommendations was higher among specialists than primary care clinicians (72.3% versus 21.5%, P<.001).

The association between guideline agreement and self-efficacy with self-reported adherence to guideline recommendations was analyzed for outcomes for which responses could be dichotomized as “almost always” performed versus other responses (Table V, Panel C). Strong agreement was not associated with almost always performing any guideline component for either clinician group with three exceptions. In unadjusted models, among primary care clinicians, agreement with guidelines was significantly associated with almost always providing a written asthma action plan (OR 2.0, 95% CI 1.1, 3.6), almost always asking about the frequency of rescue inhaler use (OR 2.9, 95% CI 1.6, 5.5) and assessing daily controller use (OR 1.9, 95% CI 1.1, 3.2), but not among specialists. For primary care clinicians, higher self-efficacy in providing guideline components was associated with higher adherence to most guideline recommendations. For specialists, high self-efficacy was associated with higher odds of performing spirometry (OR 4.3, 95% CI 2.0, 9.0). In models adjusted for clinician age, sex, and region (Table E3), guideline agreement was associated with asking about rescue inhaler use, home peak flow monitoring, and assessing daily controller use for persistent asthma, but not with providing a written asthma action plan for primary care clinicians. Self-efficacy continued to be associated with adherence to most guideline recommendations for primary care clinicians in adjusted models but not for specialists in whom only high self-efficacy regarding spirometry and testing for allergic sensitivity were associated with guideline agreement (Table E3).