Angry Doctor

Wednesday, March 07, 2007

NUMB3RS

angry doc likes the TV series NUMB3RS.

He would like to say it's because it has got all those funky mathematical models in it, but he suspects the real reason is because he find Jewish men sexy.

We all use math every day. We use maths and numbers in medicine in ways that angry doc confesses he does not understand.


An anonymous commentor drew angry doc's attention to a
study led by a mathematician, which looked at the data from a major heart study using a novel mathematical model, and concluded that the relationship between blood pressure and mortality is not linear as was commonly believed.

It does not conclude, as the commentor does, that 'there is no evidence whatsoever that lowering blood pressure has any effect on CHD.'

angry doc does not understand the complex mathematical model used in the analysis, but he did managed to find this medscape article to help him understand it.


Analysis of Recent Papers in Hypertension
from Journal of Clinical Hypertension J Clin Hypertens 2(4):290-294, 2000.
Jan Basile, MD, Walter A. Brzezinski, MD; Medical University of South Carolina; Ralph H. Johnson VA Medical Center, Charleston, SC.

Systolic Blood Pressure and Mortality

The sixth report of the Joint National Committee on the Detection, Evaluation, and Treatment of Hypertension (JNC VI), the World Health Organization (WHO), and the International Society of Hypertension (ISH) all accept and promote the concept that there is a strong, continuous, and linear relationship between increasing systolic blood pressure, cardiovascular risk, and overall mortality.

Additionally, these expert groups feel this risk becomes unacceptable at 140 mm Hg and, therefore, therapy should be initiated regardless of either age or gender at this level of pressure. This concept has now been challenged by a group of statistician investigators from southern California.

Reassessing the first 18 year Framingham data, the investigators constructed a mathematical model after arranging systolic pressure by deciles (10%-90%), age groups (45-54, 55-64, and 65-74), and gender. The new mathematical model, called a reduced horizontal logistic spline, assumes two curves, one to the left of a predetermined set point (called "the knot") and a second curve to the right. They used death as their end point, not stroke or heart attack, which was evaluated in the Framingham study. This more complicated model showed no increase in the risk for death up to at least the 70th decile of systolic blood pressure ("the knot"), an increase began between the 70th and 80th decile, and a steep increase thereafter. Furthermore, this analysis was both age and gender-specific, with elderly women tolerating higher systolic pressures than younger men. Using this model, when men were compared against women, it is suggested that hypertension need not be treated until a threshold systolic blood pressure exceeds 159 mm Hg vs. 167 mm Hg in those 65-74 years of age, 148 mm Hg vs. 158 mm Hg in those 55-64 years of age, and 141 mm Hg vs. 142 mm Hg in those 45-54 years of age, in men and women respectively.

The authors conclude that although systolic blood pressure is important, basing treatment solely on the systolic pressure being above 140 mm Hg, is not justified and subjects millions of people to unnecessary treatment. -- Port S, Demur L, Jennrich R, et al. Systolic blood pressure and mortality. Lancet. 2000;355:175-180.

Comment

We continually hear how poorly blood pressure is being controlled, as the most recent guidelines suggest we should reduce systolic blood pressure to 140 mm Hg or lower. Along comes this mathematical analysis suggesting that we are unnecessarily treating many patients.

If accepted, these suggested criteria would have a major impact on the number of elderly patients who would be treated for systolic hypertension. By their estimates, the authors state that although 22% of men and 40% of women, 65-74 years of age, would meet current criteria for treatment (>140 mm Hg), they would fall to the left of "the knot" and, therefore, would not meet the suggested threshold of systolic pressure that merits treatment. As the public health implications of these two strategies are enormous, a number of important clinical points should be raised when critically reviewing this study. The present study only used the first 18 years of the Framingham data because in the earlier study years "the study was little affected by antihypertensive drug intervention," even though there is more robust 40 year follow up data from Framingham. Previous analysis of this same data used a linear logistics regression model, which was supported by the vast mix of evidence from multiple epidemiological and clinical trials. This is the first exposure most clinicians have had to the logistic spline model, which lacks not only clinical evidence of support, but biologic plausibility as well. Finally, these statisticians have ignored stroke and heart attack in their analysis and have only evaluated mortality because it was "free of misclassification." The clinical problem we all face, however, is that our patients will benefit by preventing both stroke and cardiovascular events and not just preventing death.

If we accept the present analysis suggesting that we reduce systolic pressure to 155 mm Hg in men and 167 mm Hg in women, aged 65-74 years, we will ignore the Systolic Hypertension in the Elderly Program (SHEP), which demonstrated a marked reduction in stroke and cardiovascular event rates with systolic blood pressure reduction to 143 mm Hg as opposed to 155 mm Hg in the placebo group.

Decisions regarding usefulness of treatment should be based on clinical trials evaluating both morbidity and mortality. Therefore, we should continue to use current guidelines in the treatment of hypertension. For now, we should reduce systolic blood pressure to <140 mm Hg and watch for additional trials that address the issue raised by this study.


There is always the danger in medicine of drawing a conclusion from just one study and taking that as the truth.

The study looked at *parts* of the data from one study (albeit a big one) and looked only at death as an end point. It certainly does not allow one to conclude that 'there is no evidence whatsoever that lowering blood pressure has any effect on CHD', which does not always end in death.

Certainly we are now realising that there may be factors other than blood pressure which determine cardiovascular risks, but angry doc is not yet convinced that blood pressure has nothing at all to do with them, especially when the current evidence continue to show a relation between blood pressure and risk of cardiovascular disease.

Perhaps one day it will be proven that blood pressure does not correlate with mortality or morbidity, but until then, shouldn't we base our practices on best available evidence? After all, if you will not accept best available evidence, what will you base your practice on?

1 Comments:

Post a Comment

Subscribe to Post Comments [Atom]



Links to this post:

Create a Link

<< Home