Disparities in influenza vaccination: Arab Americans in California

Data were analyzed from the CHIS, the largest state health survey conducted in the US. The CHIS utilizes a dual-frame landline and cell phone random-digit-dialing sample of California’s non-institutionalized population. From 2001 to 2009 the CHIS survey was conducted every 2 years, and from 2009 to 2016 the survey was conducted annually. For this study, we analyzed data from the 2004–2016 CHIS survey cycles. More information on the CHIS can be found in previous work [15, 16].

As previously described, Arab Americans were identified using their responses to three questions on the CHIS Adult Questionnaire related to place of birth, language spoken at home, and parents’ place of birth [15, 16]. If one of 22 Arab League countries were indicated for participant’s or parent’s place of birth, then respondents were coded as Arab American. If the respondent indicated that Arabic language was spoken at home, then they were coded as Arab American.

NHW respondents were isolated by their responses to the following questions: “Are you Latino or Hispanic?” and “Please tell me which one or more of the following you would use to describe yourself?” If the respondent indicated “No” for Hispanic or Latino origin, selected “White” for their racial identification, and was determined to not be of Arab descent, then they were coded as NHW.

The outcome variable evaluated in this study was influenza vaccination within the past year. Participants were asked if they had received a flu shot within the past 12 months (yes/no). Influenza vaccination was assessed for each survey cycle between 2004 and 2016.

Demographic variables, assessed for all survey years, include the following: sex (female vs. male), age (18–44 vs. 45+), marital status (married vs. never married), and length of time living in the US (born in the US vs. 0–14 years vs. greater than 15 years). Socioeconomic variables examined for all survey years include unemployment status (employed vs. unemployed), educational attainment (high school or less vs. some college vs. graduate degree), living below the federal poverty level (< 200% vs. ≥200%), and health insurance status (uninsured vs. insured). Health variables included smoking status (current vs. not current smoker) and obesity (overweight/obese vs. not overweight/obese). Covariates were chosen based on their availability in the dataset and their potential influence on influenza vaccination.

Data were analyzed using SAS version 9.4 (SAS Institute, Cary, NC, 2013). Jackknife survey weights, provided by CHIS and implemented using Proc Surveylogistic, were used to account for non-response, compute standard errors and obtain weighted prevalence estimates based on the total population in California. Univariate comparisons between Arab Americans and NHW were made using chi-squared statistics. Vaccination prevalence was calculated for the 2015–2016 CHIS survey years in sensitivity analyses to compare to CDC data on influenza vaccination among minoritized groups during that time period. Survey-weighted logistic regression models were run with influenza vaccination as the main outcome. In the main model (Model 1), Arab ethnicity was used as the primary exposure. A series of sequential models were run. Model 2 was adjusted for age, sex, and marital status. Model 3 was additionally adjusted for education, employment, and federal poverty level. Model 4 was additionally adjusted for smoking and overweight/obesity. And Model 5 was additionally adjusted for number of years lived in the US. This sequential adjustment allowed us to examine the impact of sociodemographic, health, and acculturation related variables. A survey-weighted logistic regression model was run in the restricted sample of only Arab Americans to assess the predictors for influenza vaccination.

Our sample consisted of 1163 Arab Americans (weighted frequency: 121,294) and 166,955 NHW (weighted frequency: 9,854,763). Arab Americans in this sample were younger and had more males represented than in NHW (Table 1). There were more Arab American immigrants (born outside the US) than NHW immigrants in the sample.

Across all years, 30.3% of Arab Americans and 40.5% of NHW reported receiving an influenza vaccine. When restricting to the 2015–2016 CHIS survey year, prevalence of influenza vaccination among Arab Americans was 34.4%.

In unadjusted models, Arab Americans had reduced odds of reporting an influenza vaccination compared to NHW (OR: 0.64, 95% CI: 0.49, 0.83) (Table 2). After sequential adjustment for sociodemographic factors, health behaviors, and years living in the US, the odds of influenza vaccination did not differ for Arab Americans and NHW (Table 2).

Among Arab Americans, females had lower odds of influenza vaccination than males (OR: 0.57, 95% CI: 0.35, 0.94) (Table 3). A significant predictor of influenza vaccination among Arab Americans was being unemployed (OR: 1.86, 95% CI: 1.06, 3.28), even after adjustment for age (Table 3).

There are some limitations to our study. First, a representative sample of Arab Americans may not have been captured by random digit dialing for recruitment. Arab Americans were notably younger than NHW in this sample. It is possible that restricting the survey to English speaking individuals created an age bias among Arab Americans. Second, all outcomes in the survey were self-reported by the participants and were not verified with medical records. Third, the group of Arab Americans isolated by the CHIS is heterogeneous, as individuals from a wide breadth of diverse Arabic speaking countries are included, which may mask differential vaccination patterns for subgroups of this population. Despite these limitations, there are a number of strengths of this work. Most notable is the utilization of a large, population-representative sample of first, second, and third generation Arab Americans in the state with the largest Arab American population in the US.