This is a bit meatier.

http://bmj.bmjjournals.com/cgi/content/full/331/7510/187

The problem with this stuff is that the studies are all over the map.
It's almost impossible to control for all variables that might
influence heart health.    Here is this study's attempt to "control"
for the decreasing rates of smoking in Poland:

"From 1990 to 2004, the prevalence of smoking in Poland fell among
people younger than 40 but increased from 23.1% to 35.2% among women
aged 40-60. Among men, it fell from 51.1% to 46.1% at age 40-49, from
51.1% to 46.1% at 50-59, and from 34.3% to 28.2% over 60. If we assume
a relative risk of 3 for current smoking then a reduction in prevalence
of 5% (from 55% to 50%) would reduce the risk of coronary heart disease
by about 5%. This, and similar declines in rates of coronary heart
disease among men and women, indicates that changes in smoking
contributed little to falling rates of coronary heart disease."

Do they actually have any detailed data on people with heart disease
and whether or not they smoke?  Nope.  They just make the claim that
decreased rates of smoking couldn't possibly explain the decrease in
heart disease mortality.  Also, what about the under 40s?  If there was
a significant drop in smoking among under 40s starting in 1990, this
could have a significant impact on heart disease mortality in 2004 -
but for some reason this is ignored.

Note also that this study addresses only CHD mortality, not prevalence.
Nothing is mentionned about the overall prevalence of heart disease.
Is it increasing?  Decreasing?  Perhaps the cause of lower mortality is
better health care.  This "study" doesn't even try to address the
quality of health care in Poland.  I'd imagine it's much better today
than it was in 1990.  With access to modern medical interventions, I'd
expect a coronary event to be much more survivable than it was in the
Soviet era.  In fact with better care, the overall incidence of heart
disease could be *increasing*, while mortality decreases.

So what do we have here?  A "study" with nothing more than bulk
demographic disease prevalence data combined with agricultural import
statistics that could not possibly control for confounding factors.
Weak.