A total of 6,065 visits with a principal diagnosis of asthma were identified on the basis of ICD-9-CM codes, and the charts were reviewed. A total of 2,015 visits were excluded because of age of 13 years or less (n = 712), age of 55 years or greater (n = 490), history of chronic obstructive pulmonary disease (n = 177), no history of chronic asthma (n = 377), visits not prompted by asthma exacerbation (n = 289), and missing information on inclusion/exclusion criteria (n = 7). The final cohort comprised 4,053 patients who presented to 63 EDs between 2003 and 2006, with 88% of the visits made in 2004. The median number of patients treated per ED was 69 (IQR, 64–70). Participating EDs had high annual visit volumes and cared for high numbers of asthmatic patients annually (). Seventy-six percent were affiliated with an emergency medicine residency program (ie, academic EDs). Participating EDs were all urban but located in different geographic regions of the country. Seventy-eight percent of the EDs used electronic ED visit notes.
The median age of the patients was 34 years (IQR, 24–43 years); 64% were women, and 47% were black. Disease burden was high, with 9% admitted for asthma in the previous year. Nine percent of the patients had been intubated for asthma, and 23% visited the ED for acute asthma in the previous year. On presentation to the ED, the median respiratory rate was 20 breaths/min, oxygen saturation was 97%, and initial PEF was 240 L/min. On the basis of initial PEF results, the majority of patients (77%) were classified as having at least a moderate exacerbation. Concomitant bacterial infections were rare; only 3% had pneumonia. Most patients (79%) were discharged from the ED. Sixteen percent were admitted to the ward or observation unit, and 2% were admitted to the ICU.
shows the item-by-item guideline-recommended treatments prescribed to the patients, as well as the overall concordance scores. At the patient level, the overall concordance with guideline recommendations was moderate, with a median score of 67 (IQR, 63–83; ). The level of concordance varied considerably by quality measure. By using 70% as a criterion, the asthma care was suboptimal in several areas: PEF assessment (52%), prescription of oral corticosteroid at discharge (66%), and timeliness-related measures (all <70%). The level of concordance also differed by the strength of evidence outlined in the NIH guidelines. The concordance was significantly higher with level A recommendations than with level A and B recommendations (100% vs 67%, P < .001).
Performance on quality measures both at the patient and ED level
Distribution of composite guideline concordance score at the patient level. The scores are slightly negatively skewed, with more extreme values to the left.
At the ED level, the number of EDs included for each quality measure ranged from 61 to 63. Similar to the findings at the patient level, the overall ED concordance with guideline recommendations was moderate, with a mean ED composite score of 71 (SD, 7; ). The quality of care delivered in the ED also varied considerably. The best-performing ED scored 90, whereas the worst-performing ED scored 53. Item-by-item measures revealed that the variation in ED performance was greatest in the following areas: PEF assessment (SD, 24), inhaled anticholinergics (SD, 22), oral corticosteroids at discharge (SD, 19), oral antibiotics at discharge (SD, 14), and all timeliness-related measures (all SDs >10). The composite concordance was significantly higher with level A recommendations than with level A and B recommendations (89% vs 71%, P < .001).
Distribution of composite guideline concordance score at the ED level. The scores are normally distributed. The superimposed curve represents the normal curve based on sample mean and SD.
The ED characteristics associated with ED-level guideline concordance are shown in . Although EDs with higher asthma volume and those affiliated with an emergency medicine residency program tended to have higher unadjusted composite concordance scores, the only significant finding on multivariable analysis was that southern EDs were less likely to deliver guideline-concordant care compared with northeastern EDs (β-coefficient, −28.2; 95% CI, −13.8 to −2.7). Further adjustment for aggregate patient mix, including racial composition, reduced the South-Northeast quality gap, but it remained statistically significant (β-coefficient, −6.5; 95% CI, −12.8 to −0.2).
Unadjusted and multivariable predictors of higher ED composite guideline concordance score (level A + B)
About three quarters of patients received care that was fully concordant with the 4 level A recommendations in the guidelines (). These patients also had a significantly lower risk of admission compared with others (17% vs 25%, P <.001). Multivariable logistic regression was performed to assess the association between concordance with processes of care and patient outcomes (ie, hospital admission). After adjustment for patient and ED characteristics, the risk of admission remained significantly lower (adjusted OR, 0.54; 95% CI, 0.41–0.71; P < .001) among patients who received all level A guideline-recommended care in the ED compared with other patients. Also, the risk of admission increased with age (P < .001) and was greater in women compared with men (P = 5 .001). The c-statistic for the model was 0.81, and the Hosmer-Lemeshow test demonstrated a good fit (P = 5 .72).
Univariable and multivariable associations between guideline-concordant care and hospital admission among ED patients with acute asthma
First, using the “opportunity-based” method for the ED-level analysis, southern EDs remained associated with a lower concordance score compared with northeastern EDs after adjusting for patient mix (β-coefficient, −6.4; 95% CI, −12.3 to −0.5). Second, for the admission model, the magnitude of the protective effect of guideline-concordant care on admission was similar by using the hierarchic generalized linear model approach (adjusted OR, 0.50; 95% CI, 0.39–0.65; P < .001). Finally, excluding patients admitted to the ICU or admitted within 1 hour of ED arrival (6% of the study population) did not materially change the results (adjusted OR, 0.51; 95% CI, 0.37–0.69; P < .001).