Table gives sample characteristics for the N
884 respondents who had analyzable PAM scores (see Methods
). US born Latinos tended to be younger with a greater proportion ages 18–29 than the foreign born. Foreign born Latinos had appreciably lower levels of education and income and were less likely to have health insurance than US born Latinos. The vast majority of foreign born Latinos (88%) interviewed in Spanish or mostly Spanish, whereas only 19% of the US born did. English reading proficiency was high in the US born (94%) but low (31%) in the foreign born. Self-reported health status differed as well with 55% of the US born reporting excellent or very good health compared to only 22% of the foreign born.
US born Latinos had a significantly (P
0.001) greater mean value on the PAM than foreign born Latinos (75 versus 70; see Table ). There were no significant differences on the PAM by sex or age, and differences by education were only significant among the foreign born. US born Latinos with low levels of education (0–8 years) had a low mean PAM, but this small subpopulation, comprises only 5% of the US born population. Household income was associated with activation only among the US born, with those in the second quintile ($15,000–24,999) having the lowest mean PAM; the other income quintiles had greater mean PAM scores. Self-reported health status, however, was highly associated with PAM among both the US born and foreign born, with those reporting better health generally having higher PAM scores.
No significant difference in PAM scores by years of residence in the US was shown for foreign born Latinos. Both US and foreign born Latinos who interviewed in Spanish had lower mean PAM scores than those interviewed in English; however, differences in PAM scores by interview language did not reach statistical significance among the US born.
Persons classified as bilingual had the greatest mean PAM scores, although differences by verbal language proficiency were only significant among the foreign born. To differentiate possible language proficiency effects from language of interview and nativity effects, we fitted a regression model (data not shown) with PAM as the outcome and language terms and demographic factors as predictors. The bilingual predictor was highly significant (P
0.001), and terms for language of interview and nativity were not significant. Hence, PAM appears to have a primary association with whether a person is bilingual.
Table shows ordered logistic regression models for self-reported quality of care with separate models for US and foreign born Latinos. After adjusting for demographics, health status, and other factors, PAM is a strongly associated with self-reported quality of care for both US born Latinos and foreign born Latinos (Table ). In the model for US born Latinos, this odds ratio was 20 (95% CI [1.6, 247]; P
0.02). For the foreign born, the odds ratio was 3.9 (95% CI [1.3, 11]; P
0.02). Both these odds ratios were the largest in their respective models, indicating that PAM has effect size greater than the other variables included in the models.
In the model for US born Latinos, being bilingual, interviewing in Spanish, and having health insurance were also associated with greater reported quality of care. In contrast, having low education (0–8 years) and getting “a lot” of health care information from television were associated with lower perceived quality of care. In the model for the foreign born, having health insurance and having excellent health status were associated with greater reported quality of care, while being young (18–29 years) and having fair health status were associated with lower quality of care. Interestingly, there was no significant effect of time in the US for the foreign born, nor were there any significant language effects associated with self-perceived quality of care among the foreign born.
Table shows that PAM was also strongly associated with doctor–patient communication among both US born Latinos and foreign born Latinos. Note that in these models, the outcome variable is scaled to have a range of 0–100. Hence, the PAM coefficient for the US born of 61 (95% CI [38, 85]; P
0.001) means that persons with the largest PAM scores have, according to this model, a mean doctor-patient communication score 61 points higher (out of a possible 100) than persons with the lowest PAM scores. For foreign born Latinos, the effect was highly significant but the effect size was smaller (coefficient 24 with 95% CI [9, 39] and P
0.002). In the model for US born Latinos, older adults (age ≥65 years) had significantly lower scores on the doctor-patient communication scale. In the model for the foreign born, only PAM scores were significantly associated with doctor–patient communication.