Could COVID-19 deaths in Kenya be higher than official records?

How many people in African countries have been killed by the pandemic? Official confirmed COVID-19 data show extremely low death rates across Africa, but we know that many if not most COVID-19 deaths are underreported, particularly where testing is scarce, and most deaths occur at home. And the data needed to estimate excess mortality — the gold standard for understanding the mortality impacts of the pandemic — is only available for three of the continent’s 54 countries.

To get perspective on pandemic deaths in Kenya, we looked at an alternative data source: a popular obituary website. Our approach is similar to historical demography studies that have used church records to examine birth and death patterns over time. We scraped all the obituaries listed going back to 2017 and calculated a monthly count. The figure below shows the raw monthly counts in blue and the 2017-2019 average counts in red. The pre-pandemic counts show a regular pattern over time, including a surge during the May-July rainy season. The grey shaded areas show the differences between the 2017-2019 averages and obituaries during the pandemic. Two points jump out:

• Recorded obituaries for April-August 2020 were far below the previous years’ averages. This deficit in deaths suggests lives were saved early in the pandemic as a by-product of restrictions that were put in place to limit COVID-19 transmission. With more people limiting social contact and fewer people circulating on roads, deaths from violence, accidents, and other infectious diseases might have declined during this period. A separate analysis of data from one site in Kenya showed no net rise in deaths in 2020, which could result from the combination of averted deaths and increased deaths from COVID-19 that year.
• At three points in time — November-December 2020, March-April 2021, and August-September 2021 — there were large spikes in the number of excess deaths. These likely reflect surges in COVID-19 deaths. (There is also a surge in the most recent data for the first half of January, which is not shown on this figure aggregated by calendar month.)

In a second figure below, we plot the excess obituary death numbers in red alongside the official COVID-19 death counts for all of Kenya in blue. The excess obituary death counts are consistently lower than the official counts. This is unsurprising, as the obituaries reflect only a small portion of overall deaths in the country as a whole, probably consisting mostly of the deaths of relatively well-off Kenyans. The ups and downs of the excess obituary deaths match fairly closely those of official COVID deaths, suggesting that both are capturing (imperfectly) the underlying true mortality trends.

Another way to look at the data is to consider the percentage increase (or decrease) in excess obituary deaths, relative to past trends. This is called the “p-score” in excess mortality analysis. Across 103 countries, overall p-scores based on excess mortality data for the pandemic through mid-2021 range from -5% for New Zealand to 153% for Peru, with a median of 13%. (The p-score for South Africa is 32%.) The Kenya excess obituary counts imply a p-score of 10% over the pandemic period. The month-by-month p-scores are labeled on the figure. The overall 10% p-score reflects a combination of high excess obituary rates at a few points and negative excess obituary rates in mid-2020.

If the 10% figure is applied to the Kenya population, using the UN’s estimate of 284,000 deaths in Kenya in 2019, it would suggest that the total excess mortality since the start of the pandemic has been roughly 28,000. In contrast, the total of officially recorded COVID-19 deaths is 5,520.

It is important to recognize that this is a very rough analysis with highly imperfect data and should be taken as only suggestive. The mortality rates in the Kenyan population not represented in the obituaries could be higher or lower than what the obituary analysis implies.

There are three takeaways from this analysis.

First, in Kenya and across Africa, the COVID-19 death count may be higher than official counts, pointing to the urgent need to ramp up global vaccine access — many countries in Africa still have not obtained enough vaccines even for just health workers and seniors. Second, given data deficiencies in many low- and middle-income countries, it is worth looking at alternative data sources to get some idea of the scope and dynamics of the pandemic. Third, governments should see investments in civil registration and vital statistics as a central pillar of their development agenda as it’s only through such systems that we can have accurate information on births and deaths. It took centuries for developed countries’ vital registration systems to mature, but with technology, the journey doesn’t have to take as long for developing countries.

As countries work to enhance pandemic preparedness, we need to keep focus on strengthening vital registration as a core public health tool.