LONDON (AP) - Within the next 25 years, AIDS is set to join heart disease and stroke as the top three causes of death worldwide, according to a study published online Monday.

When global mortality projections were last calculated a decade ago, researchers had assumed the number of AIDS cases would be declining. Instead, it's on the rise.

Currently ranked fourth behind heart disease, stroke, and respiratory infections, AIDS is set to become No. 3, say researchers in a new report in the Public Library of Science's Medicine journal. It accounts for about 2.8 million deaths every year. But the researchers estimate a total of nearly 120 million people could die in the next 25 years.

Overall, the researchers predict that in three decades, the causes of global mortality will be strikingly similar worldwide - apart from the prevalence of AIDS in poorer countries. Most people will be dying at older ages of noninfectious diseases like cardiovascular disease, stroke and cancer.

The paper by Dr. Colin Mathers and Dejan Loncar of the World Health Organization estimates that at a total of least 117 million people will die from AIDS from 2006 to 2030. In an optimistic future projection, if new HIV infections are curbed and access to life-prolonging antiretrovirals is increased, 89 million people will die from the disease.

"What happens in the future depends very much on what the international community does now," Mathers said.

These marked differences should spark changes in current approaches to controlling AIDS now, say some experts.

"It will be increasingly hard to sustain treatment programs unless we can turn off the tap of new HIV infections," said Dr. Richard Hays, professor of epidemiology at London's School of Hygiene and Tropical Medicine, who was not linked to the study. "These AIDS numbers point to a need to do more in prevention."

Simply focusing on treatment or politically uncontroversial prevention methods will not suffice. "You can't put all your eggs in the abstinence basket," said Hays. "We need a menu of strategies for real people," he said, adding that condom distribution as well as new methods, such as a vaccine, are needed.

Mathers and Loncar analyzed data from more than 100 countries. The authors looked at the links between mortality trends and income per capita, as well as factors including education levels and tobacco use. Their research also used U.N. estimates for projected AIDS infection rates and the World Bank's numbers for future income per capita.

Mathers and Loncar then took all of this information and plugged it into a complex modeling equation to predict the top future causes of death and disease.

"This is an important contribution that will help us determine the priorities in public health," said Dr. Majid Ezzati, an associate professor of international health at Harvard University, who was not connected to the paper.

While it may be possible to avert some of the impending damage from HIV/AIDS, Mathers says that other predictions are unlikely to vary significantly.

As populations age, he explains, they are naturally more susceptible to illnesses like cancer and heart disease than from infectious diseases - even in the developing world. Life expectancy is expected to increase worldwide, with the highest projected life expectancy in 2030 to be in Japanese women, at 88.5 years.

Mathers and Loncar speculate that by 2030, cancer deaths will jump from 7.1 million in 2002 to 11.5 million. The number of deaths from cardiovascular disease is expected to rise from 16.7 million in 2002 to 23.3 million in 2030. Overall, they expect non-communicable diseases to account for 70 percent of all deaths globally, up from 59 percent in 2002.

Though economic development may bring better health care, it also has an unfortunate side effect: more road accidents. Based on rates of increasing car ownership, the World Bank estimates that traffic fatalities will increase globally by 66 percent by 2020. This might be avoided, Mathers says, if developing countries learn from the experience of developed countries, where laws and improved safety practices have sharply cut the numbers of road-related deaths.

Knowing the likely causes of future mortality allows policymakers to attempt to improve the expected outcome. While Mathers and Loncar are unable to account for unforeseen events such as the emergence of new deadly diseases or major outbreaks like a flu pandemic, their projections may help to set the agenda of global health.

"I hope this paper inspires change," said Mathers. "And I hope our pessimistic projections turn out to be wrong."