Accurate data being fundamental to any serious and effective response to the COVID-19 crisis, it is troubling to hear that the University of the Philippines’ Resilience Institute has flagged some “alarming errors” and “inconsistencies” in the Department of Health’s (DOH) data on COVID-19 cases in the country.
The UP team, composed of around 200 professors, researchers, alumni, and students from various fields such as epidemiology, emergency medicine, public health, and computer science, pointed out in a Facebook post summarizing their findings what they described as “gross errors” in the DOH’s report, specifically relating to the age, sex, civil status, addresses, and health status of some COVID-19 patients.
Comparing data releases from April 24 and 25, the UP experts said at least 45 of the virus cases went from male to female or vice versa, 75 patients became younger or older by as much as decades overnight, while 516 cases were reclassified “either to another city or a completely imaginary city.”
More mystifying still, a patient reported by the DOH as dead on April 24 was no longer among the fatalities listed the following day.
Even in stating the date, the DOH report adopted three different date formats in each case, complicating what could have been a straightforward statement of fact, said the UP experts.
Aside from errors contained in the data itself, there was a mismatch between information from the DOH and the local government units. For instance, the Health agency reported 22 deaths and 65 recoveries fewer than the provincial tally of Laguna cases on May 3.
Health Secretary Francisco Duque III’s response to the UP Resilience Institute’s report was disappointing, to say the least: Instead of taking an urgent look at the problems raised and assuring the public a transparent review process, he dismissed them as “just less than 1 percent of the whole dataset” which, he said, “does not prejudice the overall interpretation of data and decision-making.”
The errors, he said, were from the DOH’s Epidemiology Bureau and stemmed from some incomplete data on case investigation forms. The DOH has corrected such issues and was ready to move on, he said breezily, adding that the agency was rolling out a new epidemiological data system in collaboration with the World Health Organization that would automate several data collection processes to ensure minimal coding errors.
Duque’s throwaway attitude on the matter did not sit well with several senators.
It’s a simple case of “garbage in, garbage out,” said Sen. Sherwin Gatchalian. “If DOH is feeding garbage data to decision-makers, expect garbage decisions from government. Accurate data and information is fundamental in governance,” he added.
Sen. Franklin Drilon noted that mistakes, whether inadvertent or intentional, “erodes the confidence” of the people in government’s ability to contain the pandemic.
The UP team said it best when it explained how data affect policy and protocol: “These lapses may seem small relative to the total size of data contained in the daily updates, but they have significant implications on the reliability of our scientific analyses on COVID-19.” In other words, “accurate, relevant, and timely data is a basic requirement in managing the COVID-19 pandemic. Decisions depend on data, and any analysis is only as good as the data at hand.”
Because “scientific output benefits from cross-validation from peers,” a recommendation of the UP team is that all data sources should be made accessible—“while also being mindful of the same data privacy protocols that DOH is following (so as to) further empower stakeholders in our collective fight against COVID-19.”
Finally, “data must be used to generate knowledge,” this group of independent scientists and scholars, composed of some of the brightest minds in the country, reminded the government and the public. “Keeping information in silos limits our understanding of the problem, thus limiting our chances to maintain and preserve public health and security”—especially in a time as grave and unprecedented as the present.