Is Turkey underreporting COVID-19 fatalities?
The Turkish government has not fared well in terms of transparency during the coronavirus pandemic, as most recently demonstrated by Friday’s “curfew mess-up”.
Interior Minister Süleyman Soylu announced a weekend curfew on Friday, just two hours before it came into effect, prompting hundreds of thousands to descend on supermarkets and bakeries, potentially spreading the virus even further.
Many countries have been revealing detailed data of infections and deaths by city. Some in Asia have gone further, notifying their citizens on the number of cases near their homes. Turkey, on the other hand, has disclosed only limited figures, not much more than the daily and total number of cases, tests and deaths.
On April 2, Health Minister Fahrettin Koca announced infections by province for the first time, stating that as of that day, the virus had spread to all of Turkey’s 81 regions. He said the government had not revealed such data earlier because it was concerned that people would flee to unaffected areas and potentially carry the virus with them.
However, city-dwellers had already been fleeing to holiday destinations for some time. Your friendly neighbourhood economist, who moonlights as a hotelier in the southern resort of Marmaris, had been hearing that many people from large cities had already arrived at their summer homes. Others were looking for furnished apartments.
There were traffic jams on the main road from the provincial centre to the coast just a day or so before Koca’s announcement. When Koca revealed Turkey’s “corona heat map” on April 7, I was not surprised to see that the infection had spread to Akyaka, a holiday area popular with Turks that reverts to a sleepy town of 4,000 during the winter.
Not only are the statistics not detailed enough in Turkey, many people have started to question their accuracy. University of Miami professor, Cengiz Zopluoğlu, using data from the government’s e-state portal, found that the weekly number of deaths of people over 64 years of age in Istanbul, more or less constant for the last five years, suddenly started to rise after mid-March. He calculated that the number of people aged over 64 who died from March 1 to March 27 was at least 234 higher than in previous years. In the same period, the official corona death toll in all of Turkey was 92.
Columbia University academic Abdullah Aydoğan, following in Zopluoğlu’s footsteps, reached similar conclusions. Koca responded in a tweet in the early hours of March 30 defending the official figures, but he provided no evidence to back them up. The ages of the deceased were removed from the e-state portal soon afterwards.
Luckily, the government has only deleted the age-related data and details of how people died, so I was able to compare the number of total dead in Istanbul this year with previous years, and analyse how they compared to official corona death figures. For the sake of simplicity, I used only 2019 in the comparisons, as including the results of the last five years does not alter the results.
There were actually fewer deaths in eight of the first 10 days of March compared with the same period a year earlier. But then there was an uptick. This suggests that the first death from the coronavirus occurred a few days before the official announcement on March 17.
My analysis suggests that a lot of deaths were not attributed to the virus in the early days of the pandemic. The actual daily death toll could have been as much as four to five times higher in the second half of March, assuming that Istanbul accounted for half of the total deceased. This result is also more or less in line with Zopluoğlu’s conclusion, based on data until March 27. The accuracy of official figures appears to have improved to some extent since then - the death toll seems to be around two to three times higher than official data in the first 12 days of April.
Unlike opposition politicians, I do not think that the government is deliberately lying about the death toll; rather it is conveniently choosing to measure it incorrectly. For one thing, official statistics only include those who have tested positive for the virus. This means the government does not include people in the data who have died with COVID-19 symptoms, but remained untested, whereas, according to the Turkish Medical Association (TTB), they should have been categorised separately under World Health Organisation (WHO) guidelines.
Similarly, the head of the TTB’s branch in the province of Aydın told the Birgün newspaper that the official number of cases for the province only included those recorded at Aydın State Hospital. Meanwhile, the Istanbul branch of the TTB stated last week that the number of health professionals infected in the city was well above 1,000. On April 1, Koca said the total number of health professionals in all of Turkey with COVID-19 was 601.
Underreporting is not specific to Turkey. Even in countries with a good track record of collecting statistics and that stick to WHO classification procedures, such as Britain and the United States, figures for COVID19 deaths do not represent the actual death toll, according to some observers.
But rather than hiding statistics or investigating doctors who report cases of the virus in their respective cities – Turkey did exactly that to the head of the Sanlıurfa branch of the TTB - such countries have come clean on the limitations of their data. For example, Public Health England, an executive agency of the Department of Health and Social Care, has published a report on excess deaths. It stated that deaths were high from a historical perspective in the week ending March 27 due to the number of people aged 65 and over who had died.
Cloaking the reporting of coronavirus data in opacity and secrecy is also bound to undermine confidence in the statistics and the officials who report them.
For example, people began sharing a table of statistics on social media late on Monday showing that Turkey’s overall case fatality rate (CFR, COVID-19 deaths/cases) has remained constant at 2.1 percent for the past 10 days. There are examples of many countries with CFRs hovering around a specific figure, at least for a while. But I could not convince anyone who shared this widely circulated table that the data did not amount to fraud, even though I pointed out a similar trend in Italy.
If nothing else, under-reporting the number of deaths could also instil a false sense of security in the public that the spread of the virus is under control. It may consequently prompt some of them to take fewer precautions.
After Soylu’s curfew blunder, we cannot afford another.