You are factually incorrect
You have some great points. I found the article interesting. However, I have some problems with the article you cited to support your belief that being uninsured kills people.
1. There were 30% of the people excluded from the analysis because of incomplete data. Who decided on the degree of incomplete data that would disqualify a person from participation? The four liberal researchers attempting to become famous by proving we need socialized healthcare?
2. The study appropriately collected data about education, employment, tobacco use, alcohol use and leisure exercise. However, what about a host of other variables about a patient's health status or likelihood of dying? What about the rate of illegal drug use, diet, environmental safety (pollutants in air and house, lead paint, etc.), safety of cars they drive, safety belt use, gang activity, propensity to get in fights, intelligence and health literacy. The problem needs more analysis than simply asking about the four social attributes they measured.
I'll grant them this, they did admit the following-
Unmeasured characteristics (i.e., that individuals who place less value on health eschew both health insurance and healthy behaviors) might offer an alternative explanation for our findings
3. They adjusted for smoking, defining those who smoke or who have smoked by randomly deciding that smoking over 200 cigarrettes in your life counts as "smoking".
4. There is some subjectivity injected into the study by throwing in the physician assessment of health status as excellent, very good, good, fair, and poor. The subjectivity is further compounded by allowing patient self-reported health status to be included in the analysis. There is a the potential for a great deal of bias, intentional and inadvertent that could have been introduced by adding these variables.
5. This study states,
NHANES III oversampled several groups, including Black persons, Mexican Americans, the very young, and those aged older than 65 years. To account for this and other design variables we used the SUDAAN PROC SURVEYFREQ to perform all analyses. We employed unweighted survival analyses and controlled for the variables used in determining the sampling weights (age, gender, and race/ethnicity) because of the inefficiency of weighted regression analyses.
Later on, they state that they carried out an analysis, of insurance to mortality after forcing all covariates in the model.
In this Cox proportional hazards analysis, we controlled for gender, age, race/ethnicity (four categories), income (poverty income ratio), education, current unemployment, smoking status (three categories), regular alcohol use, self-rated health (four categories), and BMI (4 categories). We tested for significant ineractions between these variables and health insurance status. We tied failure times by using the Efron method… We developed a propensity score model and controlled for the variables in our previous models as well as marital status, household size, census region, number of overnight visits in hospital in past 12 months, number of visits to a physician in past 12 months; limitations in work or activities;job or housework changes or job cessation because of a disability or health problem; and a number of self-reported chronic diseases, including emphysema, prior nonskin malignancy, stroke, congestive heart failure, hypertension, diabetes, or hypercholesterolemia.
My first reaction was, "You can do that?" My next reaction was "BS." I don't believe you can control all 15 of those variables without introducing your bias as a researcher. One variable, "OK", two variables, "sure" but 15! A tweak here, a tweak there, more weight given to this variable, less given to that, and voile, I could find whatever association I wanted! The truth is that they are much like the wizard of Oz, screaming, "Pay no attention to the man behind the curtain! Just look at the screen and believe our results!"
6. I found it fascinating to see through the study that intermittent insurance patients died even more frequently than consistently insured people. If lack of insurance is "killing" people, then uninsured should have the worst health, intermittent insurance slightly better, and insured the best. The fact that intermittently insured people have WORSE mortality would suggest to me that there are unmeasured variables that cause people to have both poor job performance (causing intermittent employment) and poor health. Perhaps poor judgement that causes intermittent employment in able-bodied people is the culprit?
7. They note the following, that there is
increased likelihood of uninsurance among Mexican Americans who were nonetheless no more likely to die than non-Hispanic Whites…
Isn't that interesting? There is not increased mortality in Hispanics without insurance. It seems that the ball is in their court (and yours) to explain this.
8.
NHANES III assessed health insurance at a single point in time and did not validate self-reported insurance status. Earlier population-based surveys that did validate insurance status found that between 7% and 11% of those initially recorded as being uninsured were misclassified.
Could it be a stretch to assume that somebody too stupid or dishonest to report themselves as insured might have increased mortality?
9. The first giant elephant in the room. They do a study on mortality, attempting to show that lack of medical care kills. However, they hide the statistics that give a breakdown about specific causes of death. Did they die of murder, accidental trauma, overdose, cancer, suicide, preventable medical illness, infectious disease, etc.? If the difference could be explained by violent or accidental trauma, then the "evidence" used in this study is being misinterpreted.
10. And the second giant elephant in the room. Even if statistically valid, which I doubt due to my above concerns, this study would only prove a CORRELATION, not a CAUSATION. It is completely dishonest to cite this article and claim that lack of insurance is KILLING people. This exact dishonesty using this exact paper is pointed out in this article-
http://www.statlit.org/pdf/2010SchieldICOTS.pdf