Why Population Studies Are “Associated” With Limitations

Sigh…another week, another population study “linking” a food with a health consequence. This just-published study involved 70,000+ participants in the Japan Public Health Center–based Prospective Cohort looked at the “association” between plant and animal protein intake and mortality from heart disease, cancer, and “all-cause” mortality (a.k.a. “death from any reason”). 

Here’s what it concluded:

  • Higher intakes of plant protein were “associated” with lower cardiovascular-related and all-cause mortality.
  • Animal protein, per se, was NOT associated with any of these mortalities.
  • If red and processed meat was replaced with plant protein, it was “associated” with lower mortality from cancer, cardiovascular disease, and all-cause mortality — according to their statistical models, which did include some assumptions.

These population or “epidemiological” studies typically recruit thousands of people, use a questionnaire to get their dietary pattern – usually only at the study’s start, then check their health status 20+ years down the road.  The researchers then conclude that what participants ate when they filled out that food questionnaire was “associated” with their health outcome two decades later. 

Essentially, it gets a lot of statistics and draws a lot of conclusions based on what these people ate on a Tuesday 20+ years ago. 

When you see it laid out like that it’s almost comical.  That’s fine, but too often it’s given more credit than it deserves.

Plant protein is great. But it’s not the only way to ensure good health.
(Photo: pixabay.com)

Some Fine Print

This study took a single dietary assessment at the start and never again. Would it be fair to base your current health status on what you ate on a Tuesday 20 years ago? Eating habits change, as do activity levels. These factors may actually skew the results toward less dramatic differences, but again, it’s unknown. To be fair, the authors acknowledge that the possibility for “residual confounding in the association between plant protein and mortality remains.”

Some other potentially confounding variables:

• The high plant protein eaters also ate at least 300 FEWER calories than those eating more animal protein.
• High consumers of plant protein ate more fruits and vegetables in general, compared with the high consumers of animal protein.

Mortality from various causes has so many variables.  It’s hazardous to attribute it all to diet.  In commenting on yet another population study, Frank Mitloehner, professor of Animal Science at the University of California, Davis, who specializes in agricultural productivity and environmental sustainability, felt, “The correction for confounders is near impossible. It is difficult to know what the relative risk of one such variable is on morbidity and mortality.”  He actually gives “little credence” to these nutrition population studies.

When Statistics Aren’t Reality

These studies produce tons of numbers that are “statistically significant”.  The problem, too often, is that they may not be “clinically” significant. 

Example: take two empty Olympic-sized swimming pools.  Put a drop of water into one, and three drops into the other.  The second pool has THREE TIMES the water that the first one has.  Easy to show this statistically.  The problem: the statistics are meaningless because no one should dive into either pool.

Population Studies CAN’T Show Cause-and-Effect

They’re only designed to generate a hypothesis that should then be tested by clinical research.  This is an important distinction, because many times the hypotheses don’t pan out.  We’ve seen this happen many times in nutritional epidemiology studies over the years.  Recall two of the big mishaps:

  • Eggs being demonized because of their cholesterol content. 
  • All fat was considered bad.  Fat-free was better. 

Nutrition epidemiology studies produced those conclusions but such studies simply cannot get granular enough to produce real insight. Clinical research proved them to be inaccurate. 

Clinical research has already shown that lean red meat can not only be healthful but useful.  The BOLD study (Beef in an Optimal Lean Diet) showed that lean beef – about 4 oz daily, you don’t need any more than that – found that LDL-cholesterol was lowered as much as by the DASH (Dietary Approaches to Stop Hypertension) diet that offered only an ounce of beef daily. 

A more recent study showed a greater reduction in LDL-cholesterol on a Med diet that contained MORE lean beef than the one with less lean beef. 

Proteins from plants AND animals have a lot of nutritional value. Regardless of the source, it’s best to keep ALL portions reasonable.  Three or 4 ounces of animal protein (or the plant-equivalent) per meal is all that’s needed. Then fill out with fruits, veggies, and whole grains on your plate.

Cut-to-the-Chase-Nutrition Take-Away

You’ll never stop seeing these population studies.  They’re easy to do, they generate huge databases and they give researchers the ability to publish papers for years.  Just see them with some healthy skepticism and know they aren’t capable of producing ANY cause-and-effect conclusions – no matter what the headline might imply.  Solid, clinical studies usually give a better picture that’s far more likely to be meaningful in the real world.  IMHO.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *