Many months ago my wife and I were served our meal in a local restaurant by a woman who was our waitress on several other occasions. It was very noticeable she had acquired a severe limp since we last saw her. She explained she had an accident, went to the emergency room for treatment which was provided, but she needed corrective surgery. She explained she didn’t have health insurance as she couldn’t afford it but was hoping for the best outcome. Yesterday I saw her walking down a local street with a very severe limp. Yesterday’s experience made me wonder about people who cannot see a doctor because of cost.
The “Behavioral Risk Factor Surveillance System” provides raw data to help answer my question. One of the questions asked is, “Was there a time in the past 12 months when you needed to see a doctor but could not because of cost?” Responding with a yes or no answer represents a binary, categorical variable, which when regressed on other variables — age, gender, general health, educational attainment and income — sheds some light on predicting “could not see a doctor because of cost.”
I ran a logistic regression with a sample size of 384,196 observations providing the necessary information for the following graphs, depicting the relationship between “could not see a doctor because of cost” and age, gender, general health, educational attainment and income.
Income
The following graph depicts the predictive margins of income associated with “could not see a doctor because of cost” for people aged 20 to 80.
Only 5% of respondents at age 45 with incomes of $75,000 or more are predicted to answer yes to the question, “Was there a time in the past 12 months when you needed to see a doctor but could not because of cost?” In comparison, individuals who are 45 years old with incomes of less than $10,000 have an approximate predictive probability of 0.28 for answering yes to the same question.
Education
To demonstrate the relationship between educational attainment and “could not see a doctor because of cost” I show in the following graph the widest discrepancy between individuals who never attended high school and people who have attained 4 or more years of college education.
Gender
Females across the age distribution are more likely to say they couldn’t see a doctor because of cost.
General Health
Individuals who report they can’t afford to see a doctor are much more likely to report “poor general health.”
Our Waitress
Making the following assumptions about our female waitress (age = 24, general health is good, income is ‘less than $20,000′, and educational attainment is a high school diploma) we can predict that someone with these characteristics has a conditional probability of answering yes to the question, “Was there a time in the past 12 months when you needed to see a doctor but could not because of cost?”, of 0.532 with a 95% confidence interval of 0.522 to 0.542. The predicted probabilities were against her from the moment of her accident and the outcome appears to be adversely related to her overall health as she ages. Is this the kind of America we want?