The researchers looked into the relationship between bloodshed and HIV incidence in sub-Saharan Africa.
"It implies that there is something going on in social, political, and health care environments in those years that are conducive to HIV spread," said study lead author Brady Bennett from Brown University in Rhode Island, US.
The findings suggest that waiting to intervene until conflict is already underway may mean missing a major opportunity to prevent new infections.
The new study tracked HIV incidence statistics in 36 sub-Saharan countries from 1990 to 2012 and correlated them with periods of conflict and peace in each country.
Compared to times of peace, the analysis showed, HIV incidence increased by 2.1 infections per 1,000 people a year in the five years before a conflict where at least 25 people died as a result of fighting.
The researchers also found that during conflict the incidence rate declined by 0.07 infections per 1,000 people, compared to times of peace.
Researchers found that as conflicts became more bloody, HIV incidence tended to drift down a bit more.
Each country had a different HIV trajectory through war and peace (some had no conflict at all), but countries such as Burundi, Eritrea, Nigeria, and Uganda all endured their highest HIV incidence rates in periods before conflict and lower rates during conflict.
Bennett said the finding that HIV incidence might decline during intense conflict is likely not because conflict somehow makes anything better.
Instead, he said, fighting likely undermines accurate data collection, leaving many new infections undetected.
The study published in the journal PLOS ONE.