Health care provided by the English NHS is predominantly free at the point of use. Therefore, the way in which health services are configured is a more salient determinant of access than affordability. One potential barrier to access is the timing of health care appointments, which are predominantly offered during “normal” working hours, and many workers are unable to take time away from work to attend. Using data from the General Practice Patient Survey (2012 to 2017; N=1,227,329), we estimate the effect of an individual’s ability to take time away from work to attend health care appointments on health (measured by EQ-5D) in a person-centred treatment effects model. The model allows for treatment effects to be heterogeneous across observables and unobservables. We find those more likely to be constrained by their job would have worse health outcomes, with the average treatment effect on the treated being -0.161 points (95% CI -0.016 to -0.306). These effects are heterogeneous across socio-economic status, with individuals living in more deprived areas having lower returns. This dimension of access damages health. Improving access to healthcare through employment requires a targeted approach to avoid widening existing inequalities.