INTRODUCTION

Taxi and for-hire vehicle (FHV) driving is a growing occupation, with over 185,000 licensed taxi/FHV drivers in New York City (NYC) alone and well over 300,000 in the USA.1, 2 Most licensed NYC drivers have racial/ethnic minority backgrounds and are authorized immigrants.1 Because of their sedentary occupation, poor diet, pollution exposure, discrimination, stressors, and ethnic backgrounds, drivers are at an elevated risk for cardiovascular disease (CVD), cancer, hypertension, obesity, and diabetes.3

Most NYC drivers are independent contractors without workplace-sponsored health insurance.1, 3 Hence, many are uninsured (34% in a recent study) and lack a primary care provider (PCP).3 According to Andersen’s widely acknowledged behavioral model of health services use, social context predisposes individuals to use services or not.4 In low socioeconomic status and immigrant populations, social networks are important conduits for health information.5 Subgroups of medically underserved ethnic minority individuals share healthcare utilization habits,6 and it is likely that the heterogeneous taxi/FHV driver population can be classified according to at least three healthcare utilization profiles.6 This is the first study to test whether social networks relate to healthcare utilization profiles in taxi/FHV drivers.

METHODS

An Institutional Review Board-exempted, cross-sectional needs assessment was conducted with NYC taxi/FHV drivers from March 2013 to April 2014, assessing demographics; healthcare access and utilization (having health insurance, a PCP, a past-year routine checkup, and dental visit); and drivers’ social networks, by asking with whom (“alters”) the driver (“ego”) had discussed health-related advice over the past year.

Latent class analysis grouped drivers by their healthcare access and utilization patterns. Descriptive egocentric social network analysis tested whether driver healthcare access and utilization patterns related to their social network characteristics. Bivariate associations between social network variables or demographic covariates with latent class membership were tested, and social network variables and covariates were included in the multivariable model, based on statistically significant associations with latent class membership.

RESULTS

Table 1 describes the 211 participants and shows each latent class’s healthcare access and utilization item response probabilities. Model-fit diagnostics identified three latent classes: “uninsured past-year non-utilizers” (lowest item response probabilities for all items), “insured past-year utilizers” (high item response probabilities for having health insurance and past-year PCP visit), and “uninsured past-year utilizers” (lower item response probabilities for health insurance and PCP, high probabilities for past-year PCP and dental visits).

Table 1 Sample Descriptives (N = 211)
Full size table

Medical and dental past-year checkups were more probable among uninsured past-year utilizers (99.0% and 83.5%, respectively) than insured past-year utilizers (91.3% and 57.6%) and past-year non-utilizers (22.5% and 20.5%).

Drivers with a large proportion of their network providing health advice were more likely to be uninsured past-year utilizers than uninsured past-year non-utilizers (Table 2). No other variables related to class membership in multivariable regression.

Table 2 Multinomial Logistic Regression Modeling Membership Probability of Being in the “Uninsured Past-Year Utilizers” or in the “Insured Past-Year Utilizers” Class in Reference to the “Uninsured Past-Year Non-Utilizers” Class
Full size table

DISCUSSION

The results support prior studies showing high- and low-use primary care access and utilization classes in immigrants.6 In addition to low PCP uptake rates, low dental visit rates were concerning, given the connection between poor dental health and CVD risk, coupled with drivers’ other risk factors.3 However, a latent class emerged (uninsured past-year utilizers) that sought routine PCP and dental checkups despite lacking regular PCPs or health insurance. Drivers were more likely to be in this class than in the uninsured past-year non-utilizers class if they reported a larger proportion of friends providing health advice.

Demographic characteristics typically associated with healthcare access and utilization (e.g., years in the USA, marital and socioeconomic statuses) did not relate to latent classes in the multivariable model, supporting a heightened need for tailored targeted interventions, such as drivers trained to provide health advice, to promote healthcare access among taxi/FHV drivers. These interventions could enhance connections to insurance, primary care, and annual medical and dental appointments for drivers lacking friends providing health advice.

Future studies could inform interventions by investigating non-insurance factors; motivations for dental and PCP visits among uninsured past-year utilizers; the nature of alters’ health advice; drivers’ usage of other health advice sources, e.g., online; and healthcare utilization processes among the growing numbers of app-based FHV drivers.1, 2

This study identified a latent class of uninsured taxi/FHV drivers with a high proportion of friends giving health advice and probability of dental and PCP usage, suggesting that social networks could be leveraged to facilitate healthcare utilization in this vulnerable population.