UNIVERSITY of the Philippines (UP) experts have said there is still room for improving the quality of epidemiological data from the Department of Health (DOH) as they have observed “gross errors” and inconsistencies in recent data drops in the agency’s coronavirus disease (Covid-19) case tracker.
Health Secretary Francisco Duque III on Wednesday, May 13, acknowledged the errors and assured that these have been corrected.
He also assured that the errors were less than one percent of the entire data set on the Covid-19 epidemic in the country.
“Ang DOH po ay gumawa na po ng mga hakbang para po ma-correct itong mga tinatawag nilang mga data errors found in the April 24 and 25 data drops as well as other inconsistencies communicated by private citizens through the Covid tracker at doh.gov.ph. As early as April 26, we have since caught and corrected these issues and we very much appreciate the UP Resilience Institute for raising their concerns,” Duque said during an online press briefing with Presidential Spokesperson Harry Roque Wednesday.
“We assure the public that the issues raised are less than one percent - maliit pa sa isang porsyento - of the whole data set and does not prejudice the overall interpretation of data and decision-making,” he added.
The UP Covid-19 Pandemic Response Team, however, pointed out in its May 8, 2020 policy note that the errors may seem small, “but they have significant implications on the reliability of our scientific analyses on Covid-19”.
“Patient case data is the keystone for effective and insightful metrics and analysis. The integrity of the data drops is particularly important given that no less than President Rodrigo Roa Duterte himself has said many times that the government’s decision on managing COVID-19 will be based on science,” the team stated.
The team, which was formed a few days after the government imposed community quarantine, is composed of around 200 professors, researchers, alumni and students with various specializations including epidemiology, statistics and data science.
In its policy note, the team noted that they have yet to reconcile differences in data provided by the DOH and LGU sources more than three months since Patient No. 1 was confirmed.
On May 3, the team said the DOH reported fewer deaths and recoveries in Laguna compared to the provincial government’s records.
“Accuracy, however, goes beyond correctness in reporting aggregate numbers. Recent data drops by DOH revealed a number of alarming patient-level inconsistencies, if not gross errors,” the team wrote.
Comparing the April 24 and April 25 data drops in the Covid-19 tracker, the team observed that 45 patients had their sex changed from male to female or vice-versa. A patient listed as male on April 24 was listed as female on April 25.
The age also changed for 75 patients. Patient X was 70 years old on April 24 was listed as 76 years old on April 25.
The UP team further listed 516 patients whose residence data were changed. There were 18 patients who no longer had data on residence on April 25.
Two cases also showed different recovery dates while one patient who was reported to have died on April 24 was listed as alive on April 25.
The team also raised concern over DOH’s handling of the variable for the region of residence, which is not based on the Philippine Standard Geographic Code as compiled by the Philippine Statistics Authority.
The UP team further observed inconsistencies in the use of date formats.
These have made it difficult for automated systems of extracting and updating data from case information. Data analysis was difficult because of these sudden changes, the team added.
During the press conference Wednesday, Duque said inconsistencies in date formats and residence data have been rectified.
Duque said the DOH is committed to transparency and welcomes feedback from the experts.
He said they are rolling out a new digital epidemiological surveillance information system called Covid Kaya, a case and contact tracing reporting system developed together with the World Health Organization.
“This new system automates several data collection processes which is expected to minimize encoding errors,” Duque said. (MVI with JTM/SunStar Philippines)