Health insurance does not insure health, but wealth. Exercising to prevent disease is often better than working to buy health insurance to cover treating that disease. For example, cancer, stroke and cardiovascular disease predominantly occur in old age, so insurance against these is highly substitutable with exercise.
The American Association for Critical Illness Insurance in 2011 listed the following average annual premiums by age group for a male nonsmoker based on a $40,000 benefit for treatment of cancer, stroke or heart attack. Age 40: $575 to $610; age 45: $745 to $785; age 50: $940 to $980. Similar premiums in 2019 only buy cancer insurance.
At an after-tax hourly wage of $20, paying these premiums requires 30-50 work hours per year, about 0.6-1 hour per week, or 0.6-1% of waking hours. From a baseline of zero sports, one hour per week of exercise increases lifespan by one year, or by about 1/80 of life expectancy. Whether switching an hour per week from work to exercise (and cutting health insurance to compensate for the lost hourly wage) is a good investment in terms of lifespan depends on how much treatment lengthens life and how much health insurance increases the probability or quality of treatment. Data is difficult to find on both the effect of treatment on lifespan and the effect of health insurance on treatment.
The median survival rate to hospital discharge after EMS-treated out-of-hospital cardiac arrest with any first recorded rhythm is 7.9%. So for serious heart conditions, treatment and thus health insurance does not make much difference. Lung cancer treatment is said to prolong survival by about three months, which also seems small. Even if no health insurance implies no treatment, which is not the case because emergency care is still provided, investing worktime to buy health insurance seems to have a low benefit. People with cancer survive with a probability about 2/3 of the survival probability of a comparable population without cancer, so the upper bound on the benefit of treatment is 2/3 times the probability of getting cancer times the remaining life expectancy. This upper bound is loose, because zero treatment does not reduce the 5-year survival probability to zero.