Saundra at Good Intentions are Not Enough has some sad news:
"I got engaged the second week of January and two weeks later, on January 30th, my mother was diagnosed with Stage IV cancer. Before then we knew she was sick but had no idea it was cancer. She delayed seeing the doctor because she was uninsured and waiting to get onto Medicare. I too am uninsured – in the country that supposedly has the best healthcare in the world it is inaccessible to far too many people."
Saundra's post led me to think a little bit about the how much additional mortality there is due to uninsured people delaying healthcare. Given the recent political debate about universal healthcare, maybe it is not surprising that I found quite a few studies that address this question on Google Scholar, including a few by well-known economists. Many of the earlier studies (this is a more recent example) conduct a survey of insurance status in some base year, and then estimate the effect of baseline insurance status on mortality over time. What makes this difficult is that uninsured people also tend to be have worse health behavior and lower income. The trick is then to include everything relevant to mortality on the right-hand side of the estimation equation. An omitted variable may be correlated with both being uninsured and mortality, causing an overestimation of the effect of being uninsured on mortality. For what it is worth, the study cited above finds 45,000 deaths a year are associated with being uninsured--but I find the methodology a bit hard to swallow.
I often find simple specifications more convincing. Finkelstein and McKnight (2007) look at relative mortality rates between the "young-elderly" aged 65-75 and "non-elderly" aged 55-65 around the time medicare was introduced in 1965. If insurance is important, then around 1965 the young-elderly should suddenly have lower average mortality rates. However, Finkelstein and McKnight find instead that the elderly relative mortality rate begins decreasing a few years before medicare was introduced, and the trend continues into the early 1970's, at which point relative mortality of the young-elderly begins increasing again. One might be worried about the effect of expectations on pre-1965 behaviors, but it seems like it should make both the young-elderly and non-elderly delay health care. After testing a number of other estimation specifications, Finkelstein and McKnight conclude that they can not reject the null hypothesis that the introduction of medicare had no effect on mortality.
None of the studies I saw convinced me one way or the other, but a casual reading suggests that even if there is an effect of insurance coverage on mortality, it is likely small.