The upper number of blood pressure, also called systolic, is very important in older adults as it tends to rise with age, which was another risk factor from this study that was correlated with stroke risk. In my experience with helping older adults it is common to find isolated systolic hypertension, occurring in about 30% over the age of 80. Isolated systolic hypertension is where the upper number is elevated but the lower (diastolic) number is normal. Treatment of this type of high blood pressure is very effective at preventing strokes in which only 13-18 people need to be treated to prevent one stroke. What is also important to note is that “silent strokes”, in which you have very small strokes that go undetectable but contribute to cognitive decline, can be reduced by better management of your high blood pressure.

My advice, pay attention to your blood pressure and if that upper number is higher than 140, see your doctor and start a conversation. Also keep in mind that self-monitoring of your blood pressure and sharing those recordings with your doctor is more likely to catch high blood pressure, and is also known to be more accurate than the occasional blood pressure check in the doctor’s office.

My quest for a solid explanation started some time ago when reported findings from a couple studies made no sense at all, and I read the statements of “..after adjusting for confounding variables or risk factors..”. I first verified my suspicion in Dr. James DeMuth’s book on pharmaceutical statistics where he states we can not adjust for confounding variables and assume the conclusion is a direct cause and effect. All we can state is a correlation exists, but some authors seem to go much farther in drawing conclusions. In this article, by Dr. John Concato, et al, published in Annals of Internal Medicine, 1993;118:201-210, the author reviews common problems with the use of multivariable analyses, the first being the over-fitting of data, in which too few sample outcomes are applied to a model, and then correlations drawn from those too few outcomes. Other problems involve non-conformity to a linear gradient. For example, the impact of left ventricular ejection fraction on negative outcomes not being linear and is dependent upon where the initial ejection fraction was measured at baseline, e.g. a reduction from 40% to 30% has much lower risk as compared to a reduction from 25% to 15%.

Another problem is what the author says is a “violation of proportional hazards”, in which the risk or hazard of an independent variable is assumed to be constantly proportional. This false assumption is probably where I became suspicious, that a simple “adjusting for confounding variables” assumes all variables to be constant and exist in a simple relationship of direct correlation. The main point I make is that many statistical analyses have limitations and become more limited in their usefulness when they are not applied correctly thereby leading to incorrect conclusions.

Isn’t it true, that on any given day, we can read the summary of a research article which states that a treatment is associated with a negative outcome, after having adjusted for confounding factors, etc. And then one month later we read the opposite results from another study. It begs the question, which article is most accurate in their conclusions, if either? The use of statistics in medical research has accelerated in the last twenty years but not without concerns over the quality of its applications. I, like most human beings, tend to take things at face value. But as more findings hit the news I find myself having to employ discipline and not form any opinion until a detailed analysis of the study can be made to determine if, in fact, the purported conclusions have any meaning at all. The other application of this principle is for the lay person, no matter how educated, do not react to any information in the news.

A good example of how poor data can lead to disastrous results was after the RALES trial was published, which showed that use of a potassium-sparing diuretic (spironolactone) in younger-adult CHF patients reduced CHF exacerbations, hospitalizations and mortality. It appears as if the entire medical community took the results from that trial and transposed them onto the older adult, or geriatric, population. The result was an alarming increase in the rate of hospitalization and mortality from hyperkalemia in older adults with CHF. (Hyperkalemia is a potentially life-threatening elevation of serum potassium.) This was verified by the research of Jurlink et al in studying drug interactions that led to emergency room visits by older adults.

One geriatrician wrote about the transposition of knowledge from one population to another and referred to it as “induction”, a dangerous application of “evidence” by applying it to the wrong population. The reason the adverse outcomes occurred is because the kidneys of older adults work less efficiently than younger adults and tend to be at risk for adverse effects to medications. In this case it would be spironolactone conserving too much potassium, whereas in younger adults this would be far less an issue.

Summary: We must be careful when applying evidence from studies to the older adult population if there is not a substantial number of participants who are truly elderly or old-old, i.e. over 84 years old. Otherwise we may be causing more harm than good. When talking with your doctor, who’s about to write a prescription for you, ask the question: “Is there evidence that this drug is safe and effective in people my age”. That may be the question that keeps you out of the hospital.

What people are looking for is for their physician to listen to them and believe that their complaints are legitimate, that their ideas may have some merit. They are looking to partner with their physician in order to solve a problem that troubles them. People want to be listened to and believed. Referring to a couple of previous posts based on “the evidence”, people actually have a pretty good idea of what’s going on, it’s just that the “experts” think they know it all and shut down the conversation. That’s probably why adverse drug events are the 5th leading cause of death by disease. That’s why 20% of all hospital admissions in older adults are medication-related- because someone’s not listening to the complaints AND doesn’t have the know-how to process that information.  I have learned, when someone comes in and complains about a particular symptom and they believe it’s their medication causing it, I listen to what they have to say! Why? Because it’s proven that people are fairly reliable at detecting when a medication is causing a problem. One study showed that the “patient”, or shall I say the person taking the medication, is right about 90% of the time.

I’ll shorten this up by selling the anecdote: My wife observed the dishwasher leaking and tried to problem solve. She wasn’t certain what was going on and she hypothesized a few different things, but after two “experts” came in, failed to listen, and changed a few parts (thank goodness it wasn’t a hip and a heart valve!) it still leaked. They pushed their theories, yet couldn’t solve the problem. My wife examined it a bit more in depth and Eureka! She was certain it was a clogged drain cap that covers the drain hole. It became calcified so all she did was dissolve the residue with vinegar and scrub it clean. Voila! Problem solved. Now why couldn’t the “experts” fix it? Why did they throw expensive procedures and replacement parts at the problem and hope it “fixed it”? Because they’re truly not experts. Experts are smart enough to listen and gather all the information and then apply deductive reasoning in partnership with their clients.  Here’s a tip: When the health care practitioner you are visiting won’t listen, and throws procedure after procedure at you, yet can’t solve the problem- ask them to listen, and if they don’t, fire them! Find someone who will listen and work in partnership to solve the problem.

Here’s  the NY Times link

http://www.nytimes.com/2011/04/26/health/views/26essay.html?_r=1&emc=eta1