Straying out of one’s area of expertise can be dangerous, but Danger is my middle name [Not to be confused with Carlos Danger].
A paper, published online in the journal Circulation, says that every home foreclosure within 100 meters of someone’s house is associated with a statistically significant 1.71 mmHg rise in systolic blood pressure in that individual, p = 0.03, 95% CI 0.18 – 3.24.
Circulation has a journal impact factor of 15.2, making it the leading journal in cardiovascular disease. Who am I to question what it publishes?
However after reading the article, I thought of some questions anyway.
Blood pressures were taken by physicians with patients in a sitting position using a mercury column sphygmomanometer.
Is it valid to take measurements of mmHg from a sphygmomanometer out to 2 decimal places? No. “The number of decimal places and/or significant figures must reflect the degree of precision of the original measurement.” [via Rice University Experimental Biosciences]
How reliable are mercury column sphygmomanometers for taking blood pressures? Not very.
A paper from the British Medical Journal says that “terminal digit preference” can be a huge problem. Doctors may have a 12-fold bias in favor of the terminal digit zero; this has grave implications for decisions on diagnosis and treatment, although its greatest effect is in epidemiological and research studies in which it can distort the frequency distribution curve and reduce the power of statistical tests. [Emphasis mine]
A paper in the Journal of Human Hypertension reported that blood pressure can fluctuate markedly with respiration, movement, and even talking. Systolic blood pressure decreased by 4.4 mmHg with deep breathing, increased by 3.7 mmHg with opposite arm movement, and increased by 5.3 mmHg with talking (all p < 0.001).
Another paper looked at sphygmomanometers. Of the mercury sphygmomanometers tested 21 % were found to be inaccurate … due to excessive bouncing, illegibility of the gauge, blockage of the filter, and lack of mercury in the reservoir. Rubber aging occurred in 34% and 25%, leaks/holes in 19% and 18%, and leaks in the pump bulb in 16% and 30% of hospital devices and private practice devices, respectively.
What about confounding variables? The authors controlled for age, race, sex, income, years of education, BMI, and self-reported alcohol intake. From the paper: When body mass index and alcohol consumption were included simultaneously, the association between proximate foreclosures and systolic blood pressure decreased in magnitude and became statistically insignificant.
Can you think of any other confounding variables that might have occurred over the 25-year span of this study?
Here is the key question: “What is the clinical importance of a rise in systolic blood pressure of 1.71 or even 3.42 mmHg?”
As I have said before [here], statistical significance may not necessarily mean clinical significance.
Skeptical Scalpel is a retired surgeon and was a surgical department chairman and residency program director for many years. He is board-certified in general surgery and critical care and has re-certified in both several times. He blogs at SkepticalScalpel.blogspot.com and tweets as @SkepticScalpel. His blog averages over 1400 page views per day, and he has over 9300 followers on Twitter.