Chapter 3


as of December 29, 2017


I will discuss the “experts” we go to first—doctors and their sources of information—in this chapter.  But that is not the end of it.  In principle, medicine is science—what we know about health should be the product of science research.  So I return to medical research in Chapter 14, Science.


An Operation

Shoulder: Rotator Cuff I


Shoulder: Rotator Cuff II

Medical Errors

How Common Are Medical Errors?

What Are Medical Errors?

Are Medical Errors Recorded?

So What Do We Know?

Cost of Medical Care




An Experiment

Medical Research

Reporting Medical Research

Publication Bias

Public Policy



The equation I estimated is printed below, in a truncated version of the format the computer program (Stata) provides.  I found that my heart rate did not affect either blood pressure measure. That left the time of day (hour in tenths) and whether I had taken my meds as variables that might “explain” variation in my blood pressure.


The dependent variable—that which is being “explained”—is listed above the independent, or explanatory variables.  Only 52 observations are counted here, 22 the day after taking my medications, 30 when I had not taken my meds the night before.  A real study would go on longer, and would record whether I had taken meds two nights before, for I do not know that medications last only one day.  It is difficult to get reliable knowledge.

The column marked “P>|t|” contains probabilities.  They measure the improbability of having found such a coefficient (a coefficient this far from zero) if the “true” effect is zero.  The smaller this probability is, the less likely that zero is the true relationship.  That is why these figures are better thought of as “improbabilities.”  We are not estimating the probability that the coefficient, on the left, is “correct.”  We are estimating the likelihood that a true zero relationship would have produced it.  The lower the probability, the more sure we are of the sign of the relationship.  In the example above, blood pressure medication likely reduces blood pressure.  It is improbable that it doesn’t, or didn’t in these measures.

That does not answer the question I am asking, but it is one step towards it.  My best estimates are that, the next day, the medication reduces the diastolic pressure by 6 points, and the systolic pressure by about 4 points.  By and large I go to bed at about the same time every night, and go through the same routine when I wake before taking my blood pressure.  So time of day is a rough measure of the amount of sleep I had.  The later in the morning, the more sleep, the lower the blood pressure on measurement.24A

This is not an expert study, but it does indicate that blood pressure is lower the more sleep one has had, and that the blood pressure reducing medication (Losartan) works. More importantly, it measures the extent to which the medication works. This is information I can take to my doctor. I need to know what differences in my life expectancy I can derive from these differences in blood pressure. And then I need to know the costs, in money and side-effects, and decide if the benefits are worth the cost.