Cardiovascular disease’s “obesity paradox”—the idea that being dangerously overweight can improve end-of-life survival in heart patients—was recently debunked by a team of researchers in New York and Michigan, finding the claim to be untrue for those with incident heart disease.

The body of existing research on the paradox is “controversial,” according to a release from New York University, and suggests that obesity, which is a known risk factor for cardiovascular disease (CVD), is actually associated with higher survival rates in those with CVD. This makes sense, NYU professor and lead author Virginia Chang, MD, PhD, said in the release—in a way.

“Given that many diseases result in wasting at the end of life, the notion that extra catabolic reserve can prolong survival makes intuitive sense,” she said. “However, despite the plausibility of these hypotheses, we did not find evidence of an obesity paradox when using methods that are less prone to bias.”

These biases include weight status itself, she said, which can waver based on a patient’s health. If a normal-weight individual sheds pounds because of a severe disease, researchers will underestimate the adverse effects of obesity to healthy weight. In addition, patients who are severely overweight tend to live longer than obese individuals who suffer from severe disease, skewing the eventual patient sample.

To avoid this bias, the Chang and her colleagues pulled data from the Health and Retirement Study, which has been collecting medical data on American patients 50 and older since 1992. They focused on specific cardiovascular diagnoses, including heart attack, stroke, chronic heart failure and ischemic heart disease, and evaluated information for both prevalent heart disease patients those who were newly diagnosed.

When evaluating just those with long-term heart disease, the researchers observed the same oddity as the scientists who had previously discovered the “obesity paradox.” Risk of death was 18 to 36 percent lower for patients with a BMI of 30-34.9, they found, compared to individuals at a healthy weight.

However, in patients with matched cardiovascular conditions but newer, incident disease, there was no evidence of any survival benefit.

“The loss of an obesity paradox when switching from prevalent to incident cases and pre-diagnosis weight in the same dataset suggests that prevalent models are likely biased by factors such as disease-related weight loss and selective survival,” Chang said.