Data Collection
Data collection occurred from May 2019 through August 2021. Following approval from the institutional review board, a three-prong recruitment strategy was implemented. In May 2019, in-person data collection via tablets occurred across Detroit's seven districts at neighborhood block parties, fairs, parks, churches, libraries, senior centers, and local businesses. This strategy was undertaken to obtain a representative sample of Detroit residents (i.e., neighborhood residence, race/ethnicity, socioeconomic status, and age) while addressing internet, phone, and transportation barriers.
In-person data collection was paused in March 2020 due to Michigan's stay-at-home order during the COVID-19 pandemic. Data collection resumed in December 2020 through January 2021 via phone calls and email recruitment to neighborhood book clubs, businesses, local associations, community centers, and churches.
The study's flyer was also posted on social media accounts (i.e., Facebook, Reddit, and Craig's List) to enhance recruitment efforts. However, the three strategies (phone, email, and social media) drew limited participants (n = 6). Therefore, an incentive was incorporated from February 2021 to August 2021. An electronic $50 Visa gift card was given to the first person to complete the survey for the respective week. The incentive increased participation, resulting in a representative sample size of 394. Additional details on the study's design and methodology have been previously published [26].
Measures
The primary outcome of interest was the participants' chronic condition status (yes/no). This variable was developed by combining participants' reporting of hypertension or diabetes diagnosis, and determination of obesity status. Using measures from the Behavioral Risk Factor Surveillance System Questionnaire [BRFSS] [27], participants reported ever being told by a provider they have hypertension or diabetes (yes, no, yes but female told during pregnancy, or don't know).
The study had three main independent variables; health beliefs, health behavior tracking, and top social determinants of health. Per beliefs, participants reported strongly agree, agree, neither agree/disagree, disagree, or strongly disagree on behaviors (eating a balanced meal, exercising regularly, regular checkups when not sick, not smoking, not drinking alcohol, following medication regimens, adhering to food restrictions, and maintain a healthy weight) believed should occur to stay healthy [28]. Limited responses were identified; therefore, 'strongly agree' and 'agree' were combined to 'agree'. A similar modification was conducted for 'strongly disagree' and 'disagree' to 'disagree'.
Participants also reported tracking, not tracking, interested in tracking, or not applicable to monitoring their physical activity, fruit/vegetable intake, fish consumption, sugary beverage consumed, whole grain eaten, weight status, number of cigarettes smoked, and measuring blood pressure and sugar levels. Due to lower responses, 'do not track' and 'do not track/monitor but interested' responses were combined to create the response 'do not track/monitor'. The 'does not apply' responses were coded as missing because tracking these behaviors was not applicable.
Per the third primary independent variable, participants selected their top five social determinants of health they would like assistance addressing from a list of 21 determinants (housing, food, healthcare, stress, transportation, language, substance use, mental health, employment, education, personal safety, finances, family and community support, race/ethnicity, health literacy, physical activity, self-confidence, engagement in health/activation, sexual orientation, veteran status, or other). Participants were also able to select none of the determinants as an option. 'Other' and 'none' of the determinant responses were recoded as missing due to limited-to-no responses. For the regression analysis, the social determinants of health variables were recoded to intermediary or structural as defined by the World Health Organization's framework for social determinants of health [23].
Covariates included sociodemographic characteristics adapted from the BRFSS questionnaire. These variables were age in years (≤ 19, 20–39,30–39, 50–59, 60–69, and 70+) and gender (male/female). The race/ethnicity variable (White, Black, American Indian or Alaska Native, Asian or Indian, Middle Eastern or North African Descent, Other, Mixed race, and Hispanic) was recategorized to White, Black, Hispanic, or Other to address limited responses. Similarly, income ($0 to $9,999, $10,000 to $14,999, $15,000 to $19,999, $20,000 to $34,999, $35,000 to $49,999, $50,000 to $74,999, $75,000 to $99,999, and $100,000 or higher) was recategorized to $19,999 or less, $20,000 to $34,999, $35,000 to $49,999, $50,000 to $64,999, $65,000 to $74,999, and $75,000 or higher to address limited responses. Additionally, education (less than 8 years, 8 years to 11 years, high school graduate/GED, some college, college graduate, or postgraduate) was recategorized to high school graduate/GED or less, some college, college graduate, and postgraduate to address low responses across categories.
The survey was reviewed by public health researchers external to the research team to ensure its clarity and relevance. The survey was also piloted among 55 Detroit residents before collection occurred. The piloted survey contained open-ended questions to collect participants' opinions on the overall length of the survey and the clarity of the questions. Individuals participating in the pilot were also given the contact information of the researchers involved in the study to answer any questions or concerns regarding the survey.