“Excess Deaths” in Puerto Rico

Uncertainty over the hurricane-related death toll in Puerto Rico has been a problem from the day of Hurricane Maria’s landfall last September. Early official numbers were based on a complex arrangement requiring official verification which was not available in most parts of the Island. Only people whose death certificates listed the hurricane as the cause of death were included in the count. It is generally agreed that the official death toll gained by this process — 64 — was far too low.

Later estimates extrapolated from data collected by interviewers. Harvard researchers included deaths caused by a lack of medical care following the hurricane, including those individuals who needed treatment that required electricity and were not able to get it. They also included deaths from contaminated water, since the hurricane made drinkable water scarce in many areas. The possible total estimated by this means ranged from 793 to 8498. While the researchers set their estimate in the middle of the range, at nearly 5,000, this number is widely considered too large and too uncertain.

A new report, by Alexis Santos-Lozada of Pennsylvania State University and Jeffrey Howard at the University of Texas at San Antonio, uses “excess deaths” to estimate the hurricane’s death toll.

What are excess deaths?

Data from the past allows researchers to calculate the normal number of deaths during each month of the year. In this case, the total official death rate for the years from 2010 to 2016 was considered the baseline for the number of deaths expected.

Researchers then compared the official death records from September 2017 to February 2018. The number of deaths over the expected totals was defined as “excess deaths.” If there are 100 deaths in a typical September in Puerto Rico and 110 in September 2017, that would come to 10 excess deaths.

Excess deaths were found in the first three months after the hurricane:

  • 459 excess deaths in September
  • 564 in October
  • 116 in November

December’s official death count was higher than average, but was within the normal range. January and February did not show excess deaths. In total, 1139 excess deaths were identified.

The authors of the study acknowledge that changes in the population of Puerto Rico, which has been shrinking for some years, were not taken into account. They therefore consider their results conservative.

Future study

The government of Puerto Rico hired researchers from the Milken School of Public Health at George Washington University to come up with an official death toll. Their initial report shows that they also plan to use excess death data to calculate the death toll.

However, that initial report does not include any estimates.

The preliminary report is described as “update on the activities” of the research team. These activities are described at great length, but no data is shared.

The ‘expected’ (as if the hurricanes had not occurred) monthly mortality from September 1, 2017  – February 28, 2018 (study timeframe) will be estimated, taking into account historical heterogeneity of monthly mortality by age, sex, and region. These estimates will be evaluated relative to actual mortality in this period, allowing for estimation of excess mortality due to the hurricanes with 95% confidence intervals.”

The implication here is that the estimates will be more accurate than if the death tolls from the Centers for Disease Control and its National Center for Health Statistics.  The project also includes examination of the CDC’s death certification process as it was used following the hurricanes.

Read the full report.

One Comment

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.