Of 1,016 patients screened, 667 were excluded primarily owing to medical and psychiatric comorbidity, particularly emphysema, coronary artery disease, active major depression, and somatization disorder (). Of the remaining 349 eligible patients, 169 were not enrolled mostly because they were missed by study personnel or they did not attend their scheduled office visit. Fifteen patients refused participation; the main reason was because they had no time to participate in research. The remaining 180 patients were enrolled, and 90 were randomized to each group. Only 1 patient from each group was lost to any repeated assessment of the AQLQ and the SF-36. All the patients had at least 1 contact by telephone to record ED visits and hospitalizations.
Flow diagram of the study participants.
At enrollment, the groups were comparable with respect to most demographic and asthma characteristics (). Approximately one-third of the patients had a history of smoking, and this rate was higher in the intervention group. Body mass index also was higher in the intervention group, and, overall, 44% of patients were obese. Approximately 21% of participants had any major physical comorbidity, mostly diabetes mellitus, and 30% had a positive screen for depression. Approximately 26% of patients cited depression as a comorbidity, and half of these had a positive screen for depression. Of the remaining 74% of patients who did not cite depression as a comorbidity, approximately one-quarter had a positive screen for depression (κ = 0.29). Not all the patients reported taking asthma maintenance medications, and most reported that they were somewhat adherent to their asthma medications. Approximately 24% of patients expected that it was somewhat to very likely that they could be cured of asthma. Asthma self-efficacy was moderate, but asthma knowledge was low. The AQLQ score reflected a moderate impact of asthma, and the SF-36 physical component score reflected limitations in function that, given the low overall comorbidity, were most likely due to asthma.
Characteristics of the Study Groups at Enrollment
Mean time in the study was 27 months. The number of contacts to measure the AQLQ and the SF-36 during the study period was comparable between groups, with approximately 50% from each group receiving 3 or more interviews to assess these outcomes (). In addition, approximately 50% of participants were contacted 4 or more times by telephone to record any ED visits or hospitalizations. For intervention patients, 87% had at least 1 contact with study personnel during the first 12 weeks to reinforce the workbook and the contract; most had 3 or more contacts. Intervention patients who were not contacted by telephone during any particular week were sent information by postal mail or email for that week. The most common contracts were to “take medications” and “use a peak flow meter.” At the conclusion of the trial, 72% reported that they fulfilled their contract mostly or completely.
Longitudinal Components of the Trial
At 1 follow-up visit, patients in both groups again completed the Asthma Self-efficacy Scale and the knowledge domain of the Knowledge, Attitude, and Self-efficacy Asthma Questionnaire. Compared with enrollment scores, repeated self-efficacy scores improved for both groups, but more so for intervention patients (0.38, P < .001) than for controls (0.20, P = .06). Knowledge scores improved minimally for intervention patients but not for controls (2.12, P = .15 and −1.49, P = .33, respectively).
For the first repeated measurement at approximately 5 months, intervention patients had greater improvements in AQLQ scores than did controls (1.2 vs 0.3, P = .003). For the entire follow-up period, mean AQLQ scores were 5.1 for controls and 5.1 for intervention patients, which corresponded to within-patient improvements from enrollment scores of approximately 1.0 (P < .001) for each group. In a mixed-effects model accounting for the fixed time variable, there was no difference between randomization groups for the main outcome, ie, repeated measurements of AQLQ scores (P = .98). Several demographic and asthma variables were associated with improvements in AQLQ scores in bivariate analyses (). In a multivariate model accounting for the fixed time variable and controlling for variables found at enrollment to differ between groups (ie, body mass index, smoking history, and medication adherence), the variables that remained associated with improvement in AQLQ scores were younger age, more education, better enrollment AQLQ score, more asthma self-efficacy, more asthma knowledge, and fewer depressive symptoms.
Variables Associated With Improvement in Asthma Quality of Life Questionnaire Scores During Follow-Up
Similar analyses were conducted for SF-36 scores, and no differences were found between randomization groups. For the physical component, younger age, male sex, white race, more education, more social support, less comorbidity, fewer depressive symptoms, and a better enrollment physical component score were associated with improvement. In multivariate analysis, fewer depressive symptoms and a better enrollment score remained associated (P < .001 for both). For the mental component, more education, more social support, less comorbidity, fewer depressive symptoms, and a better enrollment mental component score were associated with improvement. In multivariate analysis, fewer depressive symptoms (P = .002), less comorbidity (P = .01), and a better enrollment mental component score (P = .003) remained associated.
In total, 56 patients (31%) had an ED visit for asthma during the study period, with no difference between groups (30% of controls and 32% of intervention patients, P = .84 [accounting for time in the study]). In multivariate analysis, female sex, expecting to be cured of asthma, less asthma knowledge, and a positive screen for depression were associated with having an ED visit (). In total, 16 patients (9%) were hospitalized for asthma during the study period, with no difference between groups (7% of controls and 11% of intervention patients, P = .30 [accounting for time in the study]). In multivariate analysis, previous hospitalization for asthma (P = .04) and a positive screen for depression (P = .02) were associated with having a hospitalization for asthma during the study period.
Variables Associated With Having an Emergency Department Visit During the Study Period
Given that depressive symptoms were associated with all outcomes, relationships between this variable and other variables known to be associated with poor clinical asthma were assessed. For enrollment characteristics, patients with a positive screen for depression were more likely to be nonwhite, to have less education, to be current smokers, to have less asthma self-efficacy, and to have less asthma knowledge (P ≤ .001 for all). In addition, patients with a positive screen had worse enrollment AQLQ and physical and mental component scores (P < .001 for each comparison). These baseline relationships were similar for the control and intervention groups. There were differences, however, between randomization groups in outcomes according to depressive symptoms. For controls, patients with a positive screen had less improvement in quality of life and were more likely to have had an ED visit and to have been hospitalized for asthma compared with those with a negative screen (). For intervention patients, however, those with a positive screen did not have less improvement in quality of life and were not more likely to have had an ED visit or to have been hospitalized. This suggests that the intervention may have been beneficial for patients with more depressive symptoms. Similar results were found for the SF-36 physical component scores where depressive symptoms were less of a determinant of outcome for patients in the intervention group.
Effects of Depression Screen on Outcomesa
At each telephone and in-person follow-up visit, patients were asked about adverse events. No adverse events were associated with this trial. At the conclusion of the trial, patients were asked how participation affected their asthma; 46% of controls and 68% of intervention patients (P = .005) reported that it made their asthma better, and the remaining patients reported that it had no effect; none reported that it made their asthma worse.