1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 2
The numbers on the DOH website don't add up from day to day. How will we bend the curve if this keeps happening? Photo by George Calvelo, ABS-CBN
Culture

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems

When will we reconcile these mistakes consistently enough to make sound decisions? BY BENJAMIN CO MD
ANCX | May 14 2020

Infectious diseases and clinical pharmacology expert Dr. Benjamin Co has been thankfully breaking down coronavirus numbers in his personal blog since the outbreak started. The perspective he provides is informative, and comforting in those who are craving for a clear picture of how we are faring against the virus. Dr. Co will share daily updates and analysis of the Department of Health reported numbers with ANCX. 

--

Refer to the link DOH.gov.ph  or up-to-date data or to COVID19.gov.ph. (The latter is not a secure site.) The new site for the Department of Health is user friendly, provides more information where a COVID19 tracker is seen. Readers can check their official site where Data Drop for raw data can be found. 

One useful site is COVID19stats, where one can see most of the DoH data in graph format.

Transparency and sharing of vital information is important as part of good governance. After all, these data are used for prediction models and scales in order to serve as platforms for planning in the business sector. The uncertainties are quantified and qualified so that a country is more prepared for a pandemic that has affected the very core of the economy of the poorest of the poor and the SMEs. 

If there is no reliable and timely information provided by the health department, one cannot help but question where they are obtaining the data from and the quality of data being shared to the public as part of public information. Selective data announcement is never encouraged because it creates disinformation when delivering news about the real situation of the outbreak in the country. 

One cannot overemphasize that the only way we will beat this pandemic is to do more aggressive and intensive contact tracing, testing and isolation and quarantine of those found to be positive for the virus in the community. 

The DoH reported 258 new cases, 86 new recoveries, and 18 new deaths. Compared to the global confirmed cases, the Philippines contributes to 0.27 percent of the cases and 0.26 percent of the deaths from COVID-19 in the world.

Two things to remember with the daily data analytics: 

(1) These three parameters (new confirmed cases, new recoveries, new deaths) are not real-time data. The data provided by the Department of Health is the date of public announcement. In reality it’s very, very late. 

(2) Depending on where testing is done, RT-PCR test results take an average of one to two days to process. Barring any delays, all tests done should ideally be released by at least 48 hours (the earlier the better). However, the test results released from government facilities range from three to 14 days, probably due to an overwhelming number of tests being conducted when compared to private hospitals where fewer number of tests done.

The announced new cases, recoveries, and deaths are the tally of reported cases of the day.

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 3

Comparison of daily new daily cases, deaths and recoveries. There are now 3,127 closed cases

The case fatality is up at 6.65 percent (vs 6.72 percent world average vs 2.44 percent ASEAN average) and the recovery rate up at 19.7 percent (vs 37.6 percent world average vs 30.1 percent ASEAN average) for the day.

The case fatality rate of the Philippines closes in to the world average fatality rate. (Note that a few weeks ago the CFR of the world was at 7.75 percent. With more global testing and tracing, there are now less deaths and more asymptomatic cases recorded. Even within the ASEAN region, the average death rate is lower compared to ours and the average recovery rate is higher than the Philippines). Which means one of two things: we are not doing enough testing and tracing and/or our healthcare system has many gaps. No matter how authorities try to provide excuses for the data, it keeps going back to the quality of data available. 

Doubling time lets us know the number of days it takes for the confirmed cases (or death rates) to double and can be determined linearly or exponentially. A logarithmic scale is the ideal graph to use. The figure below shows that the doubling time in death rates is around 8.3 days. The growth rate of cases is now down at 1.99 percent (based on the seven-day average).

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 4

Logarithmic scale for death rates in the Philippines. Note doubling time is approximately 8.3 days on this scale (35 days on a linear scale).

 

DOH Reporting

More than half (427 cases) are posthumous results. This is the number of people who have passed away before they were declared positive for SARS-CoV-2.

 

New cases per day

Of the 258 new cases announced today, all had tagged with their residence information. One hundred eighty-eight (73 percent) were from the NCR, 16 (6 percent) from Region VII, and 54 (21 percent) from other areas. Region-level data is accurate, but cases by city are reported only for those that could be verified. There were quite a handful still for validation as of this report. 

In the NCR, the breakdown is as follows: Quezon City (32), Parañaque (23), Muntinlupa (12), Manila (10); 5 each for Caloocan and Taguig; 4 each for Pasay, Las Piñas, and Valenzuela; San Juan and Makati (3 each); Marikina (2), Navotas (1), and 80 for validation. 

Region VII reported 11 cases in Cebu City, 4 in Cebu province, 1 in Mandaue and 1 in Negros. 

Other areas with reports include: Rizal (2), Cavite (2), Batangas (8), Quezon (2), Bulacan (3), Davao City (7), Iloilo Province (1), Bacolod City (1), Negros Occidental (3), Albay (2), Zambaoanga City (12), Oriental Mindoro (1), Cagayan de Oro (1), Lanao del Norte (1), Samar (1), and Leyte (1). There are a few for validation in the provinces.

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 5

New daily cases, death and recoveries

The DoH website update shows the bar graph (above) for daily cases, active cases, daily deaths and daily recoveries. The current trend shows that based on the average number of patients in the last seven days, we should be testing the threshold of 250 cases a day, for the next seven days, to see if we’re bending the curve. 

However, this will depend on the testing capacity of the country. Compared to the global data our daily deaths (as publicly reported) are less than 50/day. Remember that much of the data are latent. 

 

Testing

Based on the DoH website, a quick glance shows that the number of tests done yesterday versus today is a total of 744 tests overnight. But that’s if you’re not busy enough to go through the individual facility outputs one by one.

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 6
1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 7

Testing Capacity in terms of number of tests done May 13 versus May 12

The two tables above were taken from May 12 and May 13. The difference in total tests conducted between the dates is 744 tests. Nevertheless, one cannot help but point out the discrepancies (again). 

Based on their data, there were 883,860 remaining tests available last May 12. And 188,725 total tests have been conducted among 172,238 unique individuals with 15,779 (9.2 percent) positives and the rest being negative (90.5 percent), equivocal (0.1 percent) or invalid (0.2 percent).

Based on data dated May 13, there were 893,158 remaining tests available (an increase of almost 10,000 kits), with 189,469 tests conducted (which meant that only 744 tests were done between May 12 and May 13). Yet if you look at the data, these 189,469 tests conducted were done in 184,857 unique individuals with 16,393 positives (8.9 percent) positives and the rest being negative (91 percent), equivocal (0.1 percent), or invalid (0.1 percent). 

The question now—how can you have a difference of 12,619 positive unique individuals (184,857 last May 13 vs 172,238 last May 12) if there was just 744 done between May 12 and May 13?

But wait. 

If you tabulate the total number of tests done today, they amount to 9,465 (which is good news). But the total number of tests done to date, May 13, is unforgivably wrong. If you totaled the tests done May 12 to May 13, that should be 198,190 tests done so far (and not 189,469).

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 8

Daily Tests Conducted

The discrepancies in the data, even in the inventory of the amount of kits available in the government facilities based on website information is confusing. Based on the total tested and the available kits left, there was an inventory of 1,072,585 kits on May 12 and an inventory of 1,082,626 (which should actually be 1,091,348)? 

The biggest question is how large the discrepancy in data actually is. After all, the figure below during the T3 Task force meeting (T3 is the DoH meeting with business and industry including the ADB(Asian Development Bank) showed the remaining available tests. If you compare the DoH website for testing capacity for RITM alone, the number of tests available according to the T3 report is 1,500 while in the official website it is 591,848. That’s a lot of missing kits…or maybe another typographical error in data entry?

1,500 or 591,848? Conflicting reports in the number of testing kits underscore data problems 9

Report provided to T3 Task force May 14, 2020. 

Even if someone from an outsourced third party is trying to save face for all these inconsistent data claims that these are acceptable (1 percent error?!) because there is no such thing as 100 percent accuracy in data, I beg to disagree. First of all, these are not 1 percent error margins. 

Never mind if it’s a grammatical faux pas. But something that artificial intelligence is already doing, should be a no brainer. Because some of the numbers just really don’t add up. And in data analytics, it is a golden rule that you really need to validate first, before you press the return key. 

To get in touch with the Department of Health, the COVID hotline is (02)894-COVID loc 1555.

 

Global statistics

Update as of 9PM 14 May 2020 (Thursday)

TOTAL CONFIRMED CASES: 4,442,466

TOTAL DEATHS: 298,322 (case fatality rate: 6.72 percent)

TOTAL RECOVERED: 1,668,251 (case recovery rate: 37.6 percent)

Total cases worldwide

Note that every reference has its own cut-off time for reporting. For the global data, WorldOMeters is used as its reference.

The total confirmed cases has passed the four million mark as previously predicted. The average trend in the past week has averaged 100,000 new confirmed cases daily with more testing being done worldwide. At the current growth rate at least one million new confirmed cases may be registered every 10 days. This means there is a high probability that the five million mark will be reached on or before May 19, 2020. 

Total number of recoveries is nearing the 1.7M mark with significantly greater recoveries than deaths. 

The United States of America continue to lead globally in the number of total confirmed cases at 1,430,348 with a case fatality rate (CFR) of 5.96 percent with 85,197 total deaths recorded. The recovery rate for the US is up at 21.7 percent. Among the states, New York leads almost 350,848 total confirmed cases and 27,290 total deaths with a CFR stable at 7.8 percent. The growth rate of new cases and deaths has slowed down significantly in New York. More than 10M tests have been done as of yesterday (30,027/M population). 

The median average of case fatality rates worldwide is lower at 6.72 percent. From the current data for the past two to three months, around 80 to 90 percent of patients are either asymptomatic or recover unremarkably. Depending on the healthcare system of the country and the risk age group (older or younger populations), the fatality rate is anywhere between five to 15 percent on the average. Recoveries far outnumber the deaths with a ratio of approximately 6:1.