Friday, January 25, 2008

Night sweats

As if postmenopausal women needed more to keep them awake at night.

Last week a study published in the British Medical Journal grabbed headlines across the nation. For years, the Osteoporosis Society of Canada has told women 50 or older that they should get 1500mg of calcium a day from all sources. Now this study comes out and says that they should no longer strive for this target because doing so can increase the risk of heart attack, stroke, and...dunh, dunhh, dunhhhhh....DEATH!

What is a night sweating middle-aged woman to do? All this time she thought that she was following the rules and taking extra calcium to reduce the risk of osteoporotic fractures. No one likes a broken wrist or fractured hip. But what's this? The pay off for a fracture-free life is DEATH! Say it ain't so?

Pharmadaddy to the rescue. Given that I have advanced education in a health profession and have taken extra courses in analyzing medical literature, I feel it is my responsibility to go to the source and find out whether this study considered newsworthy is YOUworthy. Were there any problems with the study that might bring its conclusions into question? What are the experts saying and has this study changed their recommendations? Most importantly, should YOU change what you do based on this new information?

First of all, this study was a well done study in terms of design. However, it seems that the conclusions drawn from the results and the ensuing conclusions drawn in the lay press are not supported by the data, or at least not until further analysis is done.

When you compare a group that took calcium supplements against a group that took a sugar pill, and you see that those who took the calcium had higher rates of heart attack, stroke, and sudden death you could assume that the cause of those higher rates is the calcium. This assumption would be based on the notion that if the members of these two groups were selected at random, than other characteristics that might increase risk of heart attack, stroke, and death would be randomly distributed amongst the two groups evenly, and therefore not be a factor.

Typically what is done with such studies is to look at possible confounding factors, characteristics that may exist in each group that might affect the rates of cardiovascular disease besides taking calcium or not. For example, having previously had a heart attack would increase ones risk of having another, so if the group taking calcium was shown to have a higher proportion of individuals with a history of heart attacks than those taking placebo, you could argue that the increased risk of the nasties might have been due to that difference in baseline characteristics. Or at the least, because of the existence of this potential confounder, you can state that the conclusion that calcium increases heart attack risk is not supported by the data.

As a way of showing readers that confounding variables have been analyzed, study authors show the proportions of baseline characteristics in each group and show the results of a statistical test to determine if the proportions differed significantly between the two groups. Strike one for this study. While they showed the different levels of characteristics between the two groups, they never showed whether these differences were significant or merely based on chance alone. Big no no.

Once you do that first statistical test, if you find any characteristics that differ significantly between the two groups, you can then do what is called a regression analysis. This is a fancy statistical calculation that basically determines whether, in the case of this study, the increased risk of heart attacks would have changed just as much if both groups had the same baseline characteristic. It allows you to see how removing the confounding factor from the equation affects the result. So if the result is significant at first glance, but the regression analysis shows that if you account for differences in history of heart attack and the result disappears, you know that the conclusion was confounded by differences in that baseline characteristic. Strike two for this study: while they did perform a regression analysis on some factors they did not on all possible confounders, which we don't at first know, because they didn't report on them.

What were some potential confounders? The following characteristics differed significantly between the two groups:
1. Total cholesterol level, the calcium group having higher average levels
2. % current smokers, the calcium group having a higher %
3. % former smokers, calcium group having more
4. % with previous high blood pressure
5. % with previous heart disease
6. % with previous high cholesterol
7. % with previous stroke, all of the last four being higher in the calcium group

Therefore, the calcium group already had more potential risk factors than the placebo group, meaning that they probably already had a higher baseline risk of heart attack, stroke, and death. But you don't know if the differences were large enough to erase the effects of the calcium unless you perform a regression analysis.

This study only did so on a few characteristics. They report that previous ischemic heart disease was found to be an independent risk factor for rates of heart attack, stroke, and death. But they neglect to mention that the calcium group had a significantly higher proportion of individuals with previous ischemic heart disease. Hmmm. Also, they never performed regression analysis on % previous stroke, % current smokers or % former smokers, all of which could potentially be independent risk factors for the cardio triad mentioned above.

Finally, when all was said and done, the weakest aspect of this study was that the only significant result found that matters a sniff is the event rate per 1000 person years (this adjusts for people dying off or leaving the study at different times) of CONFIRMED events of heart attack, stroke, or sudden death. Many rates reported in the lay press were from the section about reported events that were never confirmed, only reported by the study subjects. So maybe Mrs. Smith thought she had a heart attack, but it was really a bad panic attack, etc. And even this one significant result is just barely so.

For those who didn't have to suffer through STATS 101, a confidence interval is the range within which the true value of the result is likely to exist 95% of the time. It is a way of quantifying the uncertainty inherent in any scientific result. In medical studies, if the confidence interval for a hazard ratio (ie. a hazard ratio of 2 would roughly mean that those in the treatment group are twice as likely to suffer a nasty event as those in the placebo group) crosses 1, than the result is insignificant, because 1 means both groups have the same risk. Well, the 95% CI for this result was 1.01-2.04. That means that those taking calcium could be anywhere from 1.01X as likely to suffer a heart attack, stroke, or death than those not taking it to 2.04X as likely. Let's just say the bottom end of that CI approaches 1 a little too closely for my comfort.

One last thing to consider is the results of other studies that looked at this question. Many have been inconclusive, but one of the biggest showed no change in cardiovascular event rates due to calcium supplementation. Although this may be due to the different formulations used in each study, it is questionable.

All in all, I was highly disappointed with this study and the media attention that followed. Postmenopausal women have enough to worry about ever since the Women's Health Initiative study was released and the pathetic health reporting that ensued. And as long as Suzanne Somers is alive and spouting the unsubstantiated trash about bioidentical hormone replacement therapy, they will always have plenty more to worry about. So why pile this on top it of it all? Especially when there is no basis for it.

All I can say is that I will not be changing the recommendations I make to my patients regarding calcium supplementation. Until I see this study impact clinical practice guidelines published by the Osteoporosis Society or the Society of Obstetricians and Gynecologists of Canada, I will stand firm, and so should you. Before you ever change any medications or supplements you are taking or not taking, always consult your health care provider.

No comments: