With Hurricane Michael's assault on Florida, officials have begun tracking storm deaths. So far, the toll stands at 16 people. But a look at how deaths were tallied after Hurricane Florence last month showed that what qualified as a storm-related death varied not only county to county but even based on who was filling out a death certificate.

Robert Edge, the coroner of Horry County in South Carolina, received a call from a florist who had provided flowers for three funerals in short succession for patients of a hospital that was evacuated before Hurricane Florence. All the family members said the deaths were unexpected.

A funeral director then called Edge to report several more deaths of people evacuated from hospitals and nursing homes. But none of the eight deaths, which occurred outside Edge's jurisdiction and in some cases in Georgia, has been attributed to the storm.

Florence's official toll does not include them, even though the federal Centers for Disease Control and Prevention last year asked those who fill out death certificates to note when they think a disaster played a role in the timing of a death — including through the stress of displacement — even if it was not a direct cause.

Inconsistencies in classifying deaths have made it difficult to compare the magnitude of different disasters, to understand why people die, and to help prevent deaths in future storms.

The problem with discrepancies in how storm-related death tolls are calculated gained attention after Hurricane Maria last year. Puerto Rico's long-held official toll of 64 was dramatically disproved by public health statistics indicating that about 3,000 more people than expected had died in the months after the storm. The vast majority did not die as a direct result of drowning in floodwaters or being struck by falling trees, but instead they likely lost their lives because of chronic illnesses exacerbated by poor conditions such as prolonged power losses. A version of the study was published in the journal Lancet Planetary Health on Thursday.

Early indications are that the tallies from this year's hurricanes may not prove more accurate.

The CDC's 14-page guide, developed with a range of experts, calls for certifiers to record both directly related disaster deaths, such as when a building collapses and crushes someone, as well as indirectly related deaths in which unsafe or unhealthy conditions stemming from a disaster likely contributed. The president of South Carolina's coroners association, Sabrina Gast, said that she hoped the CDC's guide would improve accuracy in identifying disaster-related deaths. "I think we may have missed some in the past," she said.

However, interviews with 15 coroners, medical examiners, pathologists and law enforcement officials investigating deaths in areas of the Carolinas affected by Hurricane Florence last month revealed little knowledge of the guidance or showed little consistency in applying it.

Five of six coroners interviewed in South Carolina counties either struck by Hurricane Florence or that received medical evacuees said they had never heard of the CDC's guidance. One asked a reporter to send it to him. The sixth, Kenneth Johnson of Georgetown County, when asked if he knew of it, said, "We're still reviewing it."

Dr. John Almeida, a pathologist with a regional medical examiner's facility in hard-hit Onslow County in North Carolina, said the process of reporting and classifying storm deaths had changed. "My impression is we are a little more lenient now classifying things as storm-related than they have been in the past," he said.

Still, he and others pointed out that the deaths of people whose health conditions might have been exacerbated by the storm would not necessarily be brought to the attention of medical examiners, who investigate unusual deaths or those caused by injury or violence. Certifying a natural death, by contrast, most often falls to the patient's physician.

Last week, Congress directed the Federal Emergency Management Agency to contract the National Academy of Medicine to study and provide policy recommendations on how deaths should be counted in major disasters. President Donald Trump signed the Count Act into law as part of the reauthorization of the Federal Aviation Administration.

"How death tolls are tallied shapes the public opinions of how big or the severity of the disaster, and as such it will shape the kind of federal response that the government will execute," said Rep. Nydia M. Velázquez, D-N.Y., who introduced the bill in the House. "We want a nonpartisan review."

But it is unclear whether new recommendations might be more effectively translated into policy and practice than the CDC's guidance. Without educational campaigns or a measure of enforcement, the recommendations may simply be ignored.

The majority of natural deaths across the United States are certified by regular doctors who may not have special expertise and are even less likely to have heard of the CDC's guidance. It is possible that the true toll of a disaster may only be appreciated in retrospect, by studying rates and patterns of deaths in affected populations.

Edge said confusion over how to certify disaster deaths reflected a larger lack of standardization in how death certificates are completed. "It would help for the state to make a little more clarification," Edge said, "so everyone does the same thing and it's not one way over here and one way over there."