displays the participants’ baseline characteristics by ethnic group. A greater proportion of the African American and Puerto Rican families was poor (73% and 60%, respectively) and received health care for their child at a hospital-based clinic (79% and 90%, respectively). White parents (14 years) had greater number of years of education than African American and Puerto Rican parents (12 years). Puerto Rican children were older, had longer duration of asthma, and more acute visits in the past year compared to White and African American children. African American families (43%) reported more smokers in the household compared to white (27%) and Puerto Rican (24%) families. Fewer white children (30%) had moderate to severe persistent asthma, based on clinical assessment, compared to African American (54%) and Puerto Rican (51%) children. Interestingly, Puerto Rican parents perceived their child’s asthma to be more severe than the clinical assessment classification. Seventy-four percent of parents rated their child’s asthma severity as moderate or severe compared to 51% classified as such by clinical assessment. African American and White parents tended to underestimate their child’s severity. Forty-one percent of African American children were rated as having moderate to severe asthma by their parents compared to 54% per clinical assessment. Twenty-three percent of white parents perceived their child as having moderate to severe asthma compared to 30% based on clinical assessment. Puerto Rican and African American parents had lower illness representation scores compared to White parents indicating closer alignment with the lay model. White parents reported the highest quality of their relationship with the health care provider, followed by African American parents, and Puerto Rican parents. There were no differences in beliefs about when to contact the child’s health care provider.
Baseline characteristics of study participants by ethnic group (N=309)
Model Fit (Research Question 1)
presents the results of the model fit and trimming analyses. The first model analyzed included all variables as identified in the conceptual model (). Ethnicity was included in the analyses as a grouping variable, thus is not represented as a path variable in these analyses. Based on H0/ε=0.05, HA/ε=0.08, α=0.05, df=15, and N=309, power was 0.52 for rejecting the hypothesis of close fit. This model demonstrated good fit statistics (χ2/df=0.83, CFI=1.0, TLI=1.1, RMSEA<0.0001, and SRMR=0.02) but low power; thus, results were examined for non-significant paths that could be trimmed. Five subsequent model trimming analyses were conducted removing paths that were not statistically significant for any ethnic group or whose influence on the fit statistics was negligible. The final model for the illness representation equation was comprised of parental education, poverty, parent–health care provider relationship, asthma duration, and parental perception of severity. The equation for acute visits consisted of illness representation, parental perception of severity, clinical assessment of severity, smoke exposure, and health care setting. The final model exhibited almost identical fit statistics to the original model but the increase in df (from 15 to 21) resulted in an increase in power to 0.67.
Test of model fit for social/contextual factors, perception of severity, and parental illness representations on children’s acute visits.
Prior to conducting the SEM mediation analyses, regression analyses were run, which paralleled the model fitting analyses. The initial model contained all variables from the model fitting analyses plus interactions of ethnic group with each co-variate. Model trimming was carried out to maximize percent of variance accounted for and elimination of non-significant interactions. The final model retained the interaction of ethnic group with parent-health care provider relationship for the illness representation equation. The final equation for acute visits kept the interactions of ethnic group with illness representation, parental perception of severity, and clinical assessment of severity. This model accounted for 45% of the variance in asthma illness representations and 30% of the variance in acute visits. presents the results for the final regression model, which are summarized below.
Regression results for differences in asthma illness representations and acute visits
Effects on Asthma Illness Representations
Statistically significant differences in illness representations were observed for ethnicity(β=−0.36, SE=0.17, 95% CI=−0.65,−0.08, p=0.03), poverty(β=−0.14, SE=.04, 95% CI=−0.21,−0.07, p=0.001), parental perception of severity(β=−0.06, SE= 0.02, 95% CI=−0.09,−0.03, p=0.001), and parental education( β=0.03, SE=0.01, 95% CI=0.02, .05, p≤<0.0001). All of these relationships were consistent with previous results demonstrating that poor, lesser educated, ethnic minority parents who perceive greater disease severity in their children hold illness beliefs aligned with the lay model. The interaction of ethnic group and parent–health care provider relationship on illness representation demonstrated a trend towards statistical significance (β=0.09, SE=0.05, 95% CI=0.01, 0.17, p=0.06).
Effects on Acute Asthma Visits
There were only two statistically significant predictors of acute visits: clinical assessment of asthma severity (β=0.32, SE=0.15, 95% CI=0.07, 0.57, p=0.04) and health care setting (β=−0.47, SE=0.12, 95% CI=−0.66, −0.27, p≤0.0001). Greater asthma severity led to increased acute visits. Interestingly, children who received their asthma care at the hospital-based ambulatory clinics had fewer acute visits. There was a trend for asthma illness representation and parental perception of severity towards significance. None of the ethnic group interactions on acute visits was statistically significant.
Ethnic Group Differences in Asthma Illness
Representations and Acute Visits (research question 2)
Direct Effects on Asthma Illness Representations
The final model accounted for a greater percentage of variance of illness representations for the white parents (32%) compared to African American (23%) and Puerto Rican (26%) parents (). As illustrated in , , and , differences in predictors of illness representations were observed by ethnic group. Examination of individual path coefficients revealed that only quality of the parent–health care provider relationship was a significant predictor of illness representations across all ethnic groups. Higher quality ratings were associated with higher illness representation scores, indicating alignment with the professional model. Higher education was indicative of higher illness representation scores for white and Puerto Rican parents; the African American parents demonstrated a trend towards significance in the same direction. Living in poverty and shorter asthma duration were associated with illness representation scores congruent with the lay model only for white parents. Parental perception of greater asthma severity resulted in lower illness representation scores, indicating alignment with the lay model only for the African American parents.
Model direct, indirect, and total effects of variables on illness representations and acute asthma visits by ethnic group
Final path analysis: acute asthma visits on asthma illness representation score-White.
Final path analysis: acute asthma visits on asthma illness representation score- African American.
Final path analysis: acute asthma visits on asthma illness representation score-Puerto-Rican
Direct Effects on Acute Asthma Visits
Results for acute visits demonstrated that the final model accounted for a substantially larger proportion of the variance among African American (35%) and Puerto Rican (34%) parents than for white parents (19%) (). There were also ethnic group differences in predictors of acute asthma visits ( and , and ). Higher illness representation scores (congruent with the professional model) were associated with fewer acute visits but only for the African American sample. Parental perception of greater asthma severity was associated with more acute visits among the African American and Puerto Rican children only. Among white and African American children, greater asthma severity per clinical assessment was linked with more acute visits. African American and Puerto Rican families who received their child’s asthma care in a hospital-based clinic had fewer acute visits, but health care setting was not significantly related to acute visits for the white sample. Smoke exposure was not related to acute visits for any ethnic group.
Indirect Effects on Acute Asthma Visits
Among the African American sample, there was a trend toward significance for the mediating effect of illness representation on the relationship between parent–health care provider relationship and acute asthma visits (β=−0.10, SE=0.06, 95% CI=−0.21,−0.02, p=<0.09). Illness representations did not mediate the effect of any social/contextual variables on acute visits for the white or Puerto Rican samples ().