Association of Asthma Illness Representations and Reported Controller Medication Adherence Among School-Aged Children and Their Parents

Jennifer Sonney, PhD, APRN, PPCNP-BC; Kathleen C. Insel, PhD, RN; Chris Segrin, PhD; Lynn B. Gerald, PhD, MSPH; Ida M. Ki Moore, DNS, RN, FAAN


J Pediatr Health Care. 2017;31(6):703-712. 

In This Article



The demographic characteristics of the sample are summarized in Table 1. Mothers made up 94% of the parent sample, and child participants were 64% male. All parent/caregiver participants were parents. Using a visual analog scale of 0 through 100, parents reported bearing most of the responsibility for the child's asthma care (mean [M] = 85, standard deviation [SD] = 3.44). Child participants ranged in age from 6 to 11 years (M = 8.7, SD = 1.67). Given the average age of child participants, as expected the average grade in school was third. Mean parent-reported child age at diagnosis was 31 months (SD = 23.2). All child participants in this study had insurance coverage for routine and urgent/emergent health care and for prescription medications.

Asthma status. Parent and child participants jointly reported on the child's asthma control using the C-ACT. The Cronbach's α for the total C-ACT is reported in Table 2. The total C-ACT scores (M = 22.4, SD = 3.1) indicate that 79% of the children had well-controlled asthma (C-ACT score ≥ 20) and that 21% had not well-controlled asthma (C-ACT score = 13–19). There were no children in this sample with very poorly controlled asthma (C-ACT score < 13). Although there was not a significant association between C-ACT total score and parent-reported oral corticosteroid prescriptions in the past year (r(34) = −0.16, p = .39), there was a negative association with C-ACT and number of parent-reported asthma exacerbations (r(34) = −0.64, p < .01).

Asthma illness representations. Both parent and child participants reported on asthma illness representations using the BIPQ and BMQ instruments. The Cronbach's α for the BMQ questionnaire and subscales are reported in Table 2. A Cronbach's α was not calculated for BIPQ items, because items were not expected to correlate because they measure different illness representation domains (Broadbent et al., 2006).

Paired t tests were used to compare the means for parent and child BIPQ and BMQ items (see Table 3). There was a significant difference in the consequences (BIPQ 1) domain. Parents (M = 4.8, SD = 2.44) rated the child's asthma as significantly more disruptive to the child's life than did the children (M = 2.7, SD = 2.44, t(33) = 3.16, p < .01). Similarly, parent reports on timeline or duration of asthma (BIPQ 2) also shows a significant difference with parents believing asthma will last longer (M = 6.4, SD = 2.98) than children believe (M = 5.0, SD = 3.95, t(33) = 2.18, p < .05). There were no significant differences, however, between parent and child reports of perceived personal control (BIPQ 3), treatment control or efficacy (BIPQ 4), or symptoms (BIPQ 5; see Table 3). Parent and child responses to the BMQ subscales (perceived medication necessity, medication concerns) and necessity–concern differential were similar. The overall necessity–concern differential for parents (M = 6.5, SD = 4.99) and children (M = 6.6, SD = 5.58) indicates that perceived medication necessity outweighs medication concerns.

Medication adherence. Both parent and child participants self-reported on the child's asthma controller medication adherence using the MARS-A. The Cronbach's α for parent and child MARS-A are presented in Table 2. Parent-reported controller medication adherence (M = 4.3, SD = 0.6) was significantly higher than child-reported adherence (M = 3.5, SD = 0.7, t(33) = 6.39, p < .001).

ICCs were calculated using two-way random single measures (Shrout & Fleiss, 1979). Moderate agreement was evident between the parent and child timeline domains (ICC(2,1) = 0.414), suggesting that parents and children are more consistent in agreement on timeline or duration of asthma. There was also a weak agreement between the parent and child symptoms domains (ICC(2,1) = 0.129). The remaining personal control, treatment efficacy, and consequences domains showed no significant agreement between parents and children (Table 4).

Hierarchical regression analyses were calculated first with parent-reported medication adherence as the dependent variable (Table 5). Parent beliefs about medication necessity versus concerns was a significant predictor of parent-reported treatment adherence (β = .55, p < .01), with an overall model fit of R 2 equal to 0.30. A positive BMQ differential occurs when medication necessity scores outweigh medication concern scores. The addition of child treatment control was also predictive of parent-reported medication adherence (β = −.50, p < .01) and a change in R 2 equal to 0.30, indicating that children's perceived treatment efficacy is predictive of parental MARS-A above and beyond what is predicted by parental BMQ. When child-reported medication adherence was the dependent variable (Table 6), child beliefs about medication necessity versus concerns was the only significant predictor (β = .50, p < .01), with an overall model fit of R 2 equal to 0.25, and none of the parent variables reached significance.