Essentials of Nutrition for Sports

A meta-analysis is a study method in which the results of similar
studies of a particular problem or issue are combined and analyzed mathematically in order to give

a combined “average” of individual

studies. This analysis increases the ability to detect statistically significant differences.

Pitfalls exist with both of these methods. Authors may include

only studies that support their view. Meta-analysis assumes all the studies included are of equal quality, which is rarely true. Results of different studies often do not specifically agree, making it difficult to average the different studies. The number of subjects in some studies is not large enough to provide a critical conclusion. Studies with positive conclusions tend to be publis

hed more than ones that have

negative results: Weakly positive studies may be published and available, strong negative studies may not. Cohort Studies

Subjects presenting a certain condition or receiving a treatment

are followed forward over a period of time and compared with subjects not affected with the condition or receiving treatment.

A common pitfall: The two groups differ in other ways.

Case-Control Studies

Those with a certain condition or treatment are compared with

those without, retrospectively.

Case-control studies begin with an outcome and go back in time

to determine if the outcome is associated with any factor(s).

Biases are common. For example, biases related to either the

selection of controls or the ability for cases and controls to recall information. Case Series & Case Reports

Descriptive, these series and reports do not or cannot statistically

evaluate a relationship.

They usually communicate or describe a new situation, problem,

condition, or rare phenomena.

Other Factors to Consider A Priori: Questions Asked Ahead of Time

The best studies have a formed, clearly stated primary question

and primary outcome measure. What do we wish to test or learn?

Be skeptical of results that are not part of a study’s original

questions. When many results are analyzed, some will appear to be statistically significant by chance. These results should not be given the same weight as the results of the primary question.

“Data snooping” results may be im

portant, but they may be an

attempt to justify a study that did not work out as hoped or intended.

Such results should be tested in follow-up studies.

Researcher Agenda

Has the researcher set out to prove a theory? Many researchers

keep testing and looking for data to support their point of view.

For example, a researcher is interested in proving that a new

cycling shoe will improve performance. In the first study, time trial times for 5 miles are compared between the new shoe and a standard shoe. No difference. In the second study, heart rate is examined. No difference.... In the tenth study, lactate levels are examined. A small, but statistical difference is... published and hyped. Dropouts

Were all the subjects who entered the study accounted for at its

conclusion? Were there dropouts because of adverse effects? Sodium bicarbonate may improve time trial performance, but if half the athletes who take it drop out because of diarrhea, you do not want to look at just finishers in drawing conclusions.

Nutrition for Sports, Essentials of 138