Some data released by the Department of Health on COVID-19 infections differ from the ones the University of the Philippines Pandemic Response Team had calculated.
Two experts, statistician Peter Cayton and disaster scientist Mahar Lagmay, raised this concern on Twitter.
Last Saturday, Cayton shared a screenshot of the DOH’s COVID-19 Situationer for April 16 titled “COVID-19 Situationer #354.”
It showed the time-varying reproductive numbers (Rt) for the dates March 30 and 31 which are:
- March 30: 0.49 ±02
- March 31: 0.39 ±02
The “reproductive number” is a mathematical term or element used to indicate how contagious diseases are.
“The effective reproduction number, Rt, determines the potential for epidemic spread at a specific time t under the control measures in place,” wrote physician and public health associate professor Thomas Inglesby on Jama Network.
In terms of COVID-19, an Rt value that is less than one simply indicates that a positive case was unable to infect another person, thus the disease did not spread.
Cayton, however, questioned where the DOH got such data for that day, citing the varying numbers his response team calculated.
“Hi, @DOHgovph! Saan niyo napulot na Rt ay ganitong values? Ridiculous,” he said.
Hi, @DOHgovph!
Saan niyo napulot na Rt ay ganitong values? Ridiculous. pic.twitter.com/T8QA4qnPkF
— Peter Cayton, the Stats Guy (@PJACaytonPhD) April 17, 2021
Under his tweet, the statistician shared a screenshot of their own version, wherein the the time-varying reproductive numbers are:
- March 30: 1.2034212
- March 31: 1.1901428
Cayton noted his team used the estimated date of onset in their computation.
“This UP COVID-19 Pandemic Response Team’s computation of Rt using estimated date of onset, showing Rt > 1 (reproductive number is greater than 1),” he wrote.
He also shared another version using confirmed dates in counting cases. In here, the reproductive number was still greater than one.
Lagmay later noticed the discrepancy in the figures and joined Cayton in calling out DOH for it.
“Rt = 0.39?! It does not match what is happening. @DOHgovph you owe it to the public to explain this to everyone because this number you are posting might be used for decisions in response. It is dangerous,” he wrote.
Rt = 0.39 ?! It does not match what is happening. @DOHgovph you owe it to the public to explain this to everyone because this number you are posting might be used for decisions in response. It is dangerous.
— Mahar Lagmay (@nababaha) April 17, 2021
Under the same thread, Cayton attached a link to the UP response team’s report titled the “Compendium of Philippine COVID-19 Statistics as of April 11, 2021” for the public’s reference.
A compilation of the health department’s data, meanwhile, could be viewed via this Google Drive, which was also released on the agency’s Facebook page.
In DOH’s latest situation report on April 18 titled “COVID-19 Situationer #356,” the figures had changed:
- March 30: 1.003 ± 02
- March 31: 1.002 ± 02
Some differences
In its situationer reports, the DOH explained its own interpretation for measuring the reproductive number.
“The reproduction number is measured using symptom onset, we do not include in the interpretation the past 14-16 days given the time between symptoms, health seeking, testing, and reporting,” it said.
“When interpreting the reproduction number, keep in mind that it is a lagging indicator (that is, it provides a snapshot of the situation two weeks ago). Please do not expect this number to change daily,” it added.
UP’s compendium has a similar interpretation. However, delays in data processing prompted them to compute cases based on two periods, the symptom onset date and the confirmation date.
“The work shows both Rt evaluations as confirmation date Rt is the most timely yet susceptible to biases due to validation and release of results, while onset date Rt may be reliable but timely data is not available at current time,” it said.
As of writing, the DOH has yet to respond or clarify the concerns about alleged discrepancies in its time-varying reproductive numbers.
