140 characters can say a lot about a community's well-being.

More than income, education, obesity or even smoking, a new study out this week finds the best predictor for high rates of heart disease mortality may be the language used in Tweets.

Researchers at the University of Pennsylvania sifted through 148 million Tweets from 2009-2010, and compared language with county-by-county data from the CDC on heart disease death rates.

"In terms of psychological variables, the biggest thing we found that goes with higher heart disease is anger and hostility," said Johannes Eichstaedt, lead author of the paper out this week in the journal Psychological Science.

Curses, and the words "hate," "bored" and "drama" all linked to higher rates, while "wonderful," "strength," "community" and "weekend" were used more often in counties with lower rates.

The paper makes no claims that language causes or prevents heart disease, but Eichstaedt says there is evidence that stress levels are a risk factor.

"What we think we are tracking here is the psychological states of communities," he says.