MANILA, Philippines – Every day, the Department of Health (DOH) announces the latest figures for total confirmed cases of the coronavirus in the Philippines, along with the number of those who have died and recovered from the disease.

These are also reflected in the DOH’s COVID-19 Tracker, a dashboard where the numbers are broken down by location, age, gender, and other statistics.

Local government units (LGU) affected by the virus also report the figures at their level through their official Facebook pages. Regional health offices report summaries from provinces, too.

However, the reports of many LGUs have not matched so far with the DOH’s.

One example is Cebu City, a major virus hotspot in the Visayas. As of May 20, the DOH COVID-19 Tracker shows that there are 1,572 confirmed cases with 15 deaths and 27 recoveries. But in a Facebook post by the Cebu City Health Department on the same day, the number of cases there has already reached 1,782 with 16 deaths and 108 recoveries.

See the extent of the discrepancies in concerned provinces and independent cities as of May 15 in the chart below. The dark orange dots represent the DOH numbers, while the yellow dots represent the LGU numbers. Other colored dots indicate other numbers reported by a few LGUs.

Click the navigation buttons on the upper left corner, to explore the data for confirmed cases, deaths, and recoveries. Hover over or click on the dots for more information.

Notice that the gaps are wider in virus hotspots like some areas in Metro Manila, Calabarzon, and Central Luzon, as well as Cebu City and Zamboanga City. On the other hand, the numbers largely matched in places with a lower number of cases.

In the case of Quezon City and Makati City, they included the number of DOH confirmed cases in their report, but also indicated how many among these cases have complete addresses and, in the case of Quezon City, which have been validated by its epidemiology and surveillance unit and district health office, or in the case of Makati, how many have been turned over to its health department.

So what’s the real score?

Data validation

On April 16, the DOH attributed the discrepancies to its process of verifying reports from the ground. In a press briefing, Dr Beverly Ho, director of the department’s health promotion and communication service, explained that their figures are based on the paper case investigation forms (CIF) that contain details about patients.

“Our process of verifying information from regional epidemiology surveillance units (RESU) is ongoing, and the numbers we report might change and align in the following days as we finish case processing and verification,” Ho said in Filipino.

She added that the DOH is continuously coordinating with various agencies to ask for additional data points to make sure that the information in the CIF of patients is accurate. The DOH also acknowledged then that it had an encoding backlog of over 10,000, which it eventually resolved over the Holy Week break.

This explanation is reiterated in technical notes attached to the DOH’s daily data releases. “Data is encoded from mostly paper [CIFs], therefore some encoding discrepancies may occur. All case data are dynamically updated daily as more accurate information is triangulated from a variety of official sources,” the DOH’s note says.

Topline values are then processed and validated by the DOH’s Epidemiology Bureau. “Regional and provincial/highly urbanized city/independent component city breakdowns of cases, recoveries, and deaths may vary from official values reported by [RESUs] as validation of reported residence in CIFs are ongoing,” the note also says.

Persisting issues

However, some weeks since then, the data discrepancies still persisted.

In a policy note published on May 12, the UP COVID-19 Pandemic Response Team pointed out “troubling anomalies” in the data of the DOH.

Among them are the discrepancies in the counts at the different levels, like what the team observed for Laguna as of May 3, when the DOH – through its tracker – reported 7 deaths and 28 recoveries, which was vastly different from the LGU’s log of 22 deaths and 65 recoveries.

The UP team also noted inconsistencies in the patients’ place of residence in successive data releases, like in 516 cases where the residence data “was reclassified to another city, if not a completely imaginary city.”

In response, the DOH included in its daily data drops a “change log” file that lists down changes and corrections in the data. (READ: U.P. experts say ‘open data’ is part of fight vs COVID-19)

Dr Rio Magpantay, director of the DOH Epidemiology Bureau, explained that the timely submission of complete and verified data “requires an adequate number of disease surveillance staff who shall find and interview cases, accomplish the CIF, and [encode and update] these data in an electronic information system.”

“We lack disease surveillance officers (DSO) who can do these tasks in hospitals and other health facilities, local health offices, and isolation and quarantine facilities,” he added.

While the national and local counts differ, Magpantay said it would be best to use the numbers in the national data released by the DOH. However, he added, “aggressive efforts must be made to verify and validate data coming from the LGUs to avoid discrepancies.”

Better system, coordination

To close the gap in the numbers, Magpantay said that in addition to ensuring there are available DSOs to carry out disease surveillance tasks, LGUs should have strong leadership “to ensure that they have a functional disease surveillance and response for case detection, case notification, and monitoring of response.”

He also recommended the use of a standard electronic information system “where case, lab, and close contact data by reporting units” are stored, as well as for the “close monitoring of local health offices and LGUs by its regional [counterparts].”

Professor Jomar Rabajante, a member of the UP COVID-19 Pandemic Response Team, stressed that having the same numbers in the national and local levels builds public trust, because even if the data carries disclaimers to explain the different figures, “people usually won’t read that.” 

He also noted that there are also reporting delays to the provinces from the towns and cities. “So if we account for delay, what we see at the national [level], how “real-time” is that? Not all LGUs have dashboards,” he said.

Accurate, matching data also helps in decision-making especially at the local level, as residents and businesses will be affected, Rabajante added.

The national and local governments should strengthen their coordination further “by having an automated system and streamlining the protocols,” said Rabajante. (Rappler Talk: U.P. Professor Jomar Rabajante on flattening the curve)

Former health secretary Manuel Dayrit likewise pushed for a system “where they iteratively validate and confirm the data to be reported.” Without such system, he said, “the delays, missed reports, and lapses in the reporting cycle will continue to result in discordant reports.”

In a department circular issued on April 24, the DOH promoted the “COVID Kaya” information system, developed with assistance from the World Health Organization, to “capture and analyze COVID-19 data and transform this data into meaningful information.” 

Magpantay reported that the DOH has migrated all of its COVID-19 surveillance data to the COVID Kaya app. “It has been piloted in Parañaque City and Davao Region,” he said. “It intends to capture case, laboratory, and close contact data.”

Users at the regional and local health offices are currently being trained to use the COVID Kaya app, which can be used on computers or mobile devices, with online and offline capabilities, added Magpantay. – Rappler.com