Health Insurance and Primary Care Access for Taxi/FHV Drivers in New York City: Trends from 2015–2018

Abstract

There are hundreds of thousands of metropolitan United States taxi and for-hire vehicle (FHV) drivers who lack health insurance and have limited access to primary care. The Affordable Care Act provided increased opportunities for insurance coverage. The current study used a 1,864 person 2015–2018 NYC taxi/FHV driver dataset, which included health insurance coverage and primary care access information. The data revealed an increase in insurance coverage and primary care uptake across the four years, from 40% to 63% and 52% to 60%, respectively. Drivers’ age, region of birth, and hours driving per week predicted insurance coverage, and drivers’ age, region of birth, hours driving per week, and insurance status predicted primary care coverage. Recommendations for addressing the pervasive low rates of insurance and primary care coverage among this understudied marginalized population are presented.

There are between approximately 300,000 and 750,000 taxi, ride-hailing (e.g., Uber, Lyft), and chauffer drivers across the United States.^{1, 2} In New York City, there are 187,467 licensed taxi and FHV (for-hire vehicle) drivers.³ Taxi/FHV drivers have high rates of elevated body mass index, diabetes, musculoskeletal disorders, and low-back pain.^4–8 Longtime taxi/FHV drivers are also likely to have a high-risk profile for cardiovascular disease, and experience high rates of myocardial infarction, hypertension and multi-vessel disease, likely due, in part, to unfavorable work and lifestyle factors.^{5, 9–11} Lack of access to a primary care provider (PCP) is associated with premature mortality, poor health, and depression.¹² Health insurance coverage increases access to PCPs, which is in turn linked to receipt of routine preventive care and to improved health outcomes.^13–15

Although taxi/FHV drivers are at risk for chronic conditions,^{4, 7, 9, 10, 16} historically, they have lacked health care access due to lower rates of health insurance coverage^17–19 and to structural barriers to care such as long, irregular work hours that overlap with health care providers’ hours as well as financial pressure from unstable income.^{10, 20} Studies in metropolitan cities prior to the advent of the Affordable Care Act (ACA) found that upwards of 50% of taxi drivers were uninsured.^{10, 17, 19–21} In Los Angeles, a 2006 study found that 61% of taxi drivers lacked health insurance coverage.²⁰ A 2010 San Francisco study found that 58% of taxi drivers were uninsured,¹⁰ a 2012 Chicago study reported that approximately 70% of taxi drivers were uninsured,¹⁷ and studies in New York City from 2014–2015 found that approximately 50% or more of taxi drivers were uninsured.^{19, 21}

There has similarly been a lack of PCP uptake among taxi and FHV drivers.^{7, 10, 13} Consistent with findings in San Francisco in 2010, where 54% of taxi drivers lacked a regular health care provider or place for medical care,¹⁰ 51% NYC taxi drivers in 2011 lacked a PCP.¹⁹ In comparison, the Centers for Disease Control estimated that approximately 20% of adults lacked a usual source of care across 2009–2012, with a decrease to 17% in 2015–2016.²² Having a PCP facilitates management of chronic conditions and receipt of preventive care services, reducing health disparities,²³ ultimately leading to increased life expectancy and lower all-cause mortality.^{13, 14, 24} Possessing health insurance is linked to increased PCP usage and improved health outcomes.^13–15

Despite being underinsured¹⁹ and not qualifying for employer-sponsored health insurance due to their status as independent contractors,^{25, 26} the majority of taxi drivers express a desire to obtain health insurance and to improve their health.^{4, 7, 17} Over the past decade, the ACA provided one such opportunity to increase health insurance rates among taxi/FHV drivers through its investment in low-income populations.^27–29 It created a new subsidized health insurance exchange, often called the Marketplace, and broadened the Medicaid program.^{30, 31} This coverage facilitated the delivery of clinical preventive services to previously uninsured Americans, a population of 46.5 million in 2010³² amid increasingly expensive insurance premiums and declines in employer-sponsored insurance.^{18, 29, 33}

Between all existing ACA coverage expansions, including expanded eligibility for Medicaid and the implementation of national standards for private insurance around preexisting conditions or age, 20 million previously uninsured Americans were estimated to have gained health coverage by early 2016.^{34, 35} Approximately seven million consumers selected a plan through the ACA between 2017 and 2018.^{36, 37} Low-income minorities have benefited from this success, although this has varied across states depending on each state’s Medicaid expansion program, as well as across population groups.^{38, 39} The Hispanic population has experienced the largest coverage increase compared with White non-Hispanic and non-White non-Hispanic groups in both Medicaid expansion and non-expansion states.³⁹ Between 2013 and 2017, the uninsurance rate for Blacks/African Americans dropped by 9% in expansion states and 7% in non-expansion states.⁴⁰ Between 2010 and 2014, the national uninsurance rate for Asian Indians dropped from 12.1% to 7.4%.⁴¹ Non-U.S. citizen immigrants have higher uninsurance rates than United States citizens: 17% of lawfully present immigrants and 41% of immigrants without authorized status lacked health insurance in a 2016 study, compared with 9% of citizens.⁴² A large proportion of taxi/FHV drivers in NYC are immigrants.³ South Asian drivers are the largest group of NYC taxi/FHV drivers, with Bangladesh accounting for 14.5% of drivers, Pakistan for 9%, and India for 6.5%; the Dominican Republic accounts for 13%,³ and a growing number of taxi/FHV drivers are African immigrants.²⁶

It is unknown if changes in insurance coverage rates, and the expected resultant increase in PCP uptake, have carried over to the taxi/FHV driver population. Under the ACA, independent contractors, such as taxi/FHV drivers, can enroll in health insurance by completing an application through the Marketplace and are potentially eligible for free or low-cost care through Medicaid, if their incomes qualify them.⁴³ Drivers, like all qualified populations, can also purchase subsidized plans with varying premiums and out-of-pocket costs, and cannot be denied coverage due to pre-existing conditions.^{30, 44}

This study examines changes in health insurance rates and PCP uptake during the years 2015–2018 among taxi and FHV drivers, a group with limited prior affordable paths to obtaining quality insurance as independent contractors. We utilize data collected during workplace and community health fairs for taxi/FHV drivers, and the results derived from this convenience sample are therefore limited to non-causal interpretation. Nevertheless, our findings could inform local and national outreach and policy efforts for taxi/FHV and for other at-risk, vulnerable groups, especially those working as independent contractors.

Methods

Health fairs.

This study used data from 420 health fairs conducted with taxi and FHV drivers at taxi garages, airport holding lots, app-based support centers, and community sites (e.g., mosques, temples) from August 2015 through December 2018. Health fairs were conducted with the aim of understanding drivers’ health profiles, assessing their barriers to health care, and connecting drivers to health insurance and health care. Study assistants administered an intake on participants’ demographic and health care access and behavior profiles. Staff provided health care access linkages for those seeking insurance or primary care assistance. Drivers were then introduced to, and sometimes enrolled in, a number of health interventions designed to increase drivers’ access to insurance and PCPs, and assist with healthful behavior changes, including increased physical activity. Prior work has detailed the process for these worksite health initiatives.^{19, 21} This protocol was approved as an exempt research study by the Institutional Review Board at Memorial Sloan Kettering Cancer Center.

Measures.

Age, region of birth, English proficiency, hours driving per week, and vehicle/medallion ownership were collected as part of health care access and healthful behavior change intervention eligibility screening tools. Drivers were queried about whether they currently had any kind of health insurance coverage, including prepaid plans such as HMOs, or government plans such as Medicaid or Medicare, using a question derived from the Behavioral Risk Factor Surveillance System questionnaire.⁴⁵ To determine whether drivers had a source for primary care, defined as a PCP, drivers were also asked if there was a doctor’s office, clinic, health center, or other place they usually go to if they are sick and/or need advice about their health. This item was derived from the Medical Expenditures Panel Survey (MEPS) questionnaire,⁴⁶ which determined usual source of care across a nationally representative sample.

Data analysis.

Participants were considered eligible for this analysis if they were male, a licensed taxi/FHV driver, and if they responded to the items on PCP and health insurance status. We also restricted analyses to drivers under the age of 65 because of the possibility that drivers 65 years of age or older may be more likely to have insurance given their potential eligibility for Medicare. Descriptive analyses were employed to determine the frequencies of health insurance and PCP uptake, per year of study. To examine the demographic and workplace predictors of health insurance and PCP coverage, two binary logistic regressions were modelled, including age, region of birth, driving hours per week, and vehicle ownership. We also included insurance as a predictor of PCP coverage. For region of birth, we chose Sub-Saharan African drivers as the reference category because they were the largest group. Analyses were conducted using SPSS Version 25.⁴⁷

Results

Descriptive statistics.

Data from 1,864 drivers were analyzed (Table 1). Drivers were an average of 45 years old (SD = 11.31, range = 17–64). Most drivers reported being born in sub-Saharan Africa (37.1%), South Asia (22.5%), Latin America (14.8%), North Africa and the Middle East (7.9%), East and Southeast Asia (1.9%), and other regions (10.5%). The majority of taxi/FHV drivers (74.8%) rented or leased their vehicle/medallion from a garage. Drivers worked a mean of 9.98 hours per day (SD = 1.94, range = 1 – 24), with 74% working between 9–12 hours per day. Drivers worked approximately 5.42 days per week (SD = 1.08, range = 1–7), with slightly over half of drivers reporting that they drove six to seven days a week (53.1%). The average number of hours driven per week was 54.05 hours (SD = 14.67, range = 4 – 140). The majority of drivers (92.5%) reported driving 35 hours or more per week. The average number of years driving was 8.97 years (SD = 8.64, range = less than 1 year to 43 years). Most drivers (84.3%) reported that they spoke English very well. From 2015– 2018, rates of health insurance coverage increased from 39.6% to 63% and rates of PCP uptake increased from 52.1% to 60%.

Table 1.

Descriptives (n = 1864).

	Total n	n	%	M	SD
Age	1823			44.53	11.31
Region of birth	1766
Sub Saharan Africa		692	37.1%
North Africa & Middle East		148	7.9%
South Asia		420	22.5%
Latin America		275	14.8%
East Asia/Tibet/Southeast Asia		36	1.9%
Other		195	10.5%
Years driving	1843			8.97	8.64
Vehicle ownership	1834
Owns vehicle/medallion		439	23.6%
Rents/leases vehicle/medallion		1395	74.8%
English proficiency	1864
Speaks English very well		1571	84.3%
Does not speak English very well		293	15.7%
Work hours per week	1846			54.05	14.67
Less than 35 hours per week		122	6.5%
35 hours per week or more		1724	92.5%
Insured	1864	1110	59.5%
2015	217	86	39.6%
2016	284	159	56%
2017	555	356	64.1%
2018	808	509	63%
Has PCP	1864	1076	57.7%
2015	217	113	52.1%
2016	284	144	50.7%
2017	555	334	60.2%
2018	808	485	60%

^*Note: Percentages may not always add to 100 due to missing data.

Predictors of health insurance coverage.

A binary logistic regression was conducted to investigate the demographic and year of study enrollment predictors of health insurance status (Table 2). Enrollment year, age, region of birth, English proficiency, and hours driving per week were significant predictors of health insurance coverage. The model indicated that as the study enrollment year increased, so too did the likelihood of having health insurance until 2018 (OR = 1.23, 95% CI (1.10, 1.38)), during which it remained the same as in 2017 (Table 1). Age, examined as a continuous variable, was positively associated with an increased likelihood of having health insurance (OR = 1.02, 95% CI (1.01, 1.03)).

Table 2.

Demographic and workplace predictors of health insurance coverage.

	Exp (B)	95% CI	p
Year	1.23	1.10, 1.38	.000
Age	1.02	1.01, 1.03	.001
Region of birth (Ref: Sub Saharan Africa)
North Africa & Middle East	1.87	1.25, 2.78	.002
South Asia	2.48	1.87, 3.29	.000
Latin America	.73	.54, .98	.036
East Asia/Tibet/Southeast Asia	.39	.19, .81	.012
Other	.75	.54, 1.05	.090
Driving hours per week	.98	.98, .99	.000
Taxi/medallion ownership (Ref: Non-owner)	1.23	.95, 1.61	.123

In comparison with sub-Saharan African drivers, North African and Middle Eastern drivers were 1.87 times as likely (95% CI (1.25, 2.78)) and South Asian drivers were 2.48 times as likely (95% CI (1.87, 3.29)) to have health insurance. Drivers from Latin America (OR = .73, 95% CI (.54, .98)) and East/Southeast Asia (OR = .39, 95% CI (.19, .81)) were less likely to have insurance than sub-Saharan African drivers. Driving more hours per week was associated with a 2% lower likelihood of having health insurance (95% CI (.98, .99)).

Predictors of PCP uptake.

A binary logistic regression was conducted to ascertain the demographic variables that influenced PCP uptake status. Similar to the results for health insurance coverage, significant predictors of having a PCP included year of study enrollment, age, region of birth, and hours driving per week. When health insurance coverage was included in the model, year of study was no longer significant. Age, region of birth, and number of hours driving per week remained significant. Age, examined as a continuous variable, was positively associated with the likelihood of having a usual provider (OR = 1.04, 95% CI (1.03, 1.06)).

South Asians were more likely to have a primary care provider than sub-Saharan African drivers (OR = 1.91, 95% CI (1.38, 2.56)). Each additional hour of driving per week was associated with a 1% lower likelihood of having PCP (95% CI (.98, .99)). Drivers with health insurance were over 12 times more likely to have a PCP than drivers without health insurance (OR = 12.24, 95% CI (9.56,15.68)).

Discussion

Taxi/FHV drivers are a marginalized group who historically have had reduced access to both health insurance coverage and primary care providers.^17–19 We used a large dataset (1) to determine changes in insurance and PCP uptake rates over the past four years, and (2) to ascertain demographic predictors of health insurance and PCP uptake. Overall, rates of insurance and PCP uptake increased from 2015 to 2018. Approximately 23% more drivers had insurance in 2018 compared with 2015, with the rates remaining fairly constant from 2017 to 2018. Just over half of drivers (52.1%) had a primary care provider in 2015. Only 7.9% more of drivers reported having a PCP in 2018 than in 2015, and there was similarly little change from 2017 to 2018.

This increase in insurance coverage is promising and suggests the possibility that the ACA benefits did, indeed, accrue to taxi/FHV drivers, although we were unable to directly test this hypothesis given the lack of data prior to 2015. Consistent with findings in other populations,^{14, 15, 48} drivers with health insurance were more likely to have a PCP, over 12 times more likely in this group, than those without health insurance. Uptake of PCPs, particularly in racial/ethnic populations, is associated with improved health outcomes and decreased premature morbidity.^{12, 49}

However, rates of health insurance coverage were still low (below the national average) and there was a very modest increase in PCP uptake over the study years despite the larger increase in insurance coverage. Getting health insurance alone was not enough to increase PCP uptake and the taxi/FHV driver population lags behind the general population in this area. In the general population, 24% of adults under 65 reported having no usual source of care in 2014–2015,^{46, 50} which fell to 12% in 2016–2018.⁵¹ In comparison, from 2015 to 2018, approximately 42.3% of drivers in this study did not have a regular PCP. These findings emphasize the need for studies to examine the impediments to health insurance coverage and PCP uptake in this at-risk population. Interventions should be designed to address these barriers, and then evaluated for their effectiveness. Such interventions will likely fall within the patient outreach and education, health care system, and policy spheres, and may include, for example, well-designed navigation programs to increase both insurance and PCP uptake among taxi/FHV drivers, workplace-based enrollment programs, and mobile health care teams that visit the taxi/FHV worksite and gathering places.

Increasing age significantly increased the likelihood of health insurance and PCP coverage among the drivers studied, mirroring national trends among the general public. As people get older, emerging health conditions may necessitate that they obtain insurance and visit a regular doctor. Because drivers aged 65 and over are eligible for health insurance via Medicare, a government sponsored, national primary care health insurance program for individuals aged 65 and older, we did not include them in the current analyses. A recent survey among the general population demonstrated that younger adults are less likely to have a PCP, with as many as 45% of those aged 18–29 years reporting that they did not have a PCP, 28% of those aged 30–49 years, 18% of those aged 50–64 years, and 12% of those aged 65 and older.⁵²

In comparison with our reference group of sub-Saharan drivers, North African/Middle Eastern drivers were more likely to have health insurance and South Asian drivers were more likely to have health insurance and to have a regular PCP. Drivers from Latin America were less likely than sub-Saharan African drivers to have insurance or a regular PCP when insurance was not included in the model. Drivers from East/Southeast Asia were less likely to have insurance. While national uninsurance rates decreased for all ethnic/racial groups from 2013–2017, Hispanics had the second highest uninsurance rate at 19%, after American Indians and Alaskan Natives. Asians and Whites had the lowest uninsurance rates, at 7%.⁴⁰ There are several potential explanations for the variations in insurance coverage by region of birth in this study. South Asian drivers may be more likely to have been driving for a longer period of time or to have immigrated to the U.S. before sub-Saharan African drivers (immigration from sub-Saharan Africa increased by 52% recently, in the past decade)⁵³ and as a result, may be more familiar with the insurance and health care system. We were unable to examine these additional demographic variables, nor drivers’ income, which may have contributed to insurance rates and PCP uptake.

We found that driving more hours per week was associated with a slightly lower likelihood of having insurance or a PCP, likely because those who spent more hours driving had less available time to attend to health-related tasks such as signing up for health insurance or visiting a PCP, and possibly because their incomes were lower, necessitating increased driving hours.

We were limited to the data in this convenience sample. Because drivers are often hesitant to participate in research studies given the loss of income associated with the time commitment, we kept our questionnaire short. In our effort to capture data from as many drivers as possible, important demographic information, such as income, years in the U.S., and education level, was not included in the current analyses but should be considered to the extent possible in future research. Our inability to account for these socioeconomic indicators may have biased our results, and we emphasize the importance of interpreting these data with this caveat. For example, we found differences in insurance and PCP coverage by region of birth. Rather than these differences being attributable to drivers’ country of origin, other missing variables, such as income, could explain these differences. Our analyses included vehicle/medallion ownership, which may be indicative of income, however, this did not predict either outcome of interest.

Our sample may also be biased due to the manner in which data were collected: drivers who participated in the health fair were available and willing to respond to the survey items, may have been less likely to have access to health care, or may have been more interested in health care coverage than the average NYC taxi/FHV driver. Drivers also frequently responded to only some portions of the survey before they had to leave for various reasons, leading to missing data. Health fairs were conducted in a large variety of locations and we did not capture exact settings for each participant; therefore, there may be significant differences in health insurance and PCP coverage based on the setting in which the data were collected (e.g., borough, taxi garages, community organizations, restaurants). We are also unable to provide complete data on the type of vehicle driven, although the majority of participants were likely yellow taxi drivers based on the recruitment sites utilized. Given the enormous shifts in the ride-sharing industry, driver profiles may change dramatically in the coming years. Despite these limitations, this study provides important data on this at-risk, marginalized population.

In these analyses we did not include initial data from over 100 drivers in early 2019. However, an examination of these data show that through March 2019, there has been a health insurance rate of 70.6% and a PCP uptake rate of 68.6%. This upward trend is promising, especially in light of the end of the individual mandate in 2019.⁵⁴ As health insurance options have widened, access to free or affordable health care has expanded,⁵⁵ allowing this population to seek care. Although we cannot conclude from our findings that increases to insurance and PCP coverage among NYC taxi/FHV drivers are directly a result of the ACA, we strongly believe that changes to ACA coverage would affect the observed gains.⁵⁶

Further research is needed, including qualitative work, to examine drivers’ reasons for their uninsurance and how these might be addressed. As part of an initiative to assist drivers in enrolling in health plans, in our prior work, we examined drivers’ awareness and perceptions of the ACA and found that 78% of the sample had little understanding of the ACA and that 77% wanted more information.⁷ As a result, an effort to increase insurance knowledge and enrollment rates and PCP uptake was developed and implemented, with trained health care navigators assisting drivers in health care plan enrollment during community and workplace-based health screenings.^{7, 19}

Taxi drivers in a previous study indicated that lacking health insurance is a barrier to health care, and that despite being ineligible for Medicaid, they struggle to afford primary care visits.⁵⁷ Increasing health insurance coverage is therefore critical; awareness and perception of the ACA, along with availability of navigator assistance, are critical factors in ACA enrollment among low-income individuals.³⁸ Navigator assistance and higher levels of education have been associated with a higher insurance application completion rate, while reasons for not applying include high perceived cost and lack of knowledge about coverage options.⁵⁸ Recent proposals in New York City have included bolstering the City’s public health insurance option, including mental health care access for all residents.⁵⁹ These policy initiatives are positive steps in delivering health care to the most marginalized groups, including taxi/FHV drivers. Such measures should be considered to provide access to care for this population across the country, and for other at-risk groups that lack employer-sponsored health insurance.

Table 3.

Demographic, workplace, and insurance coverage as predictors of primary care provider access.

	Exp (B)	95% CI	p	Exp (B)	95% CI	p
Year	1.17	1.05, 1.31	.006	1.07	.93, 1.22	.343
Age	1.04	1.03, 1.05	.000	1.04	1.03, 1.06	.000
Region of birth (Ref: Sub Saharan Africa)
North Africa & Middle East	1.17	.80, 1.71	.416	.80	.51, 1.24	.319
South Asia	2.60	1.97, 3.44	.000	1.91	1.38, 2.65	.000
Latin America	.69	.51, .93	.015	.76	.53, 1.10	.142
East Asia/Tibet/Southeast Asia	.64	.31, 1.31	.220	1.11	.47, 2.63	.813
Other	.64	.45, .90	.010	.67	.45, 1.01	.054
Driving hours per week	.98	.97, .99	.000	.99	.98, .99	.001
Taxi/medallion ownership (Ref: Non-owner)	1.12	.86, 1.46	.384	1.00	.74, 1.37	.985
Insurance				12.24	9.56, 15.68	.000