Humanitarian crises monitoring: impact of coronavirus (interim findings) Contents

---

3Availability of information

Data challenge

56.It was clear from our evidence that there were enormous challenges in making an accurate assessment of the spread of the coronavirus in many countries, both developed and developing, from the data available. Robert Mardini from ICRC told us that “ … , very often, only 50% of the health infrastructure [in countries within which it was working] is operational. That shows how difficult it is to get access to credible and systematic data in normal times, let alone in times of COVID-19. It is extremely hard to get any sensible sets of data. We get anecdotal data and very often we only get to see the tip of the iceberg.”.

57.Other witnesses concurred and explained how they tried different approaches. Gwen Hines, Save the Children, told us that: “ We have thousands of national staff on the ground and we use them to get data to add to official data. We have recently been going through an exercise of looking country by country and cross-checking that with official sources. You can also use proxies to a certain extent, in terms of what you are hearing from people involved in burials or reporting cases at the village level. We are using that to understand what is happening. … We also have a number of health experts who are involved in the research side. We are tracking that research, which is so important, be it Imperial College London or other”.46 Aleema Shivji, Executive Director, Humanity and Inclusion, praised DFID in general for leadership on disability inclusion and in particular for seeing that through into promoting the collection of data disaggregated by disability age and gender.47

58.DFID’s Chief Scientific Adviser, Dr Charlotte Watts, also told the Committee that: “It is really hard to get accurate data. We have invested in collection of data, including supporting Africa CDC to provide technical support and improve data. We have also supported the London School of Hygiene and Tropical Medicine and Imperial College, to help us do modelling to give us better projections, at least, of what might be happening. They estimate, for example, for Sub-Saharan Africa, that about 10% of cases are being reported.”48

59.Professor Azra Ghani, Chair in Infectious Disease Epidemiology, School of Public Health, Imperial College. explained that the task was mainly finding ways of supplementing thin testing data:

The main challenge is the weakness of the surveillance system, so trying to understand who is getting tested and how many tests are performed. We look at the test positivity rate as an indicator of how many tests are being performed on negative people as well as positive, and that is very helpful.

We focus particularly on looking at reported deaths, because we know that cases tend to be under-reported. Not everybody will be seeking tests or treatment, but even those are very challenging in places that do not have registration systems in place. We are starting to look for other sources of data, for example media reports of funerals or other information, particularly looking at excess deaths if that information becomes available, to try to get a better handle.

The reason for wanting to do this is that we need to understand the stage of the epidemic in different countries. In some places we seem to have had very few Covid cases and very few deaths. This may not indicate that the transmission has not happened, but rather that it has happened and has been hidden by the weaker surveillance system.49

60.Professor Ghani confirmed the DFID Chief Scientist’s reference to a 10 per cent. reporting rate in sub Saharan Africa. She told us that 10% was a good ballpark but it varied from one country to another. For example, the larger epidemic evident in South Africa may mean that infections were confined there “but it may be that the surveillance is better there.” Another potential source of data was antibody test results (‘seroprevalence surveys’).50

61.Professor Ghani pointed to a recent survey in Kenya, which indicated that 5% of people had been infected (rising to 8% or 9% in major urban areas).51 Francesco Checchi, OBE, Professor of Epidemiology and International Health, London School of Hygiene and Tropical Medicine, said that part of his study in Yemen was going to be based on satellite imagery of Aden graveyards as a source of data on Covid deaths in Yemen.52 This is the manner in which the extent of Iran’s Covid epidemic was exposed by the New York Times in March 2020. Dr Timothy Russell, Research Fellow, London School of Hygiene and Tropical Medicine, described a further way of assembling a data picture from fatality data. He told the Committee that the way it worked was fairly simple, using the ‘hidden’ information in reported death data to infer or reverse engineer an estimate of the number of Covid cases it would have taken to produce those dead bodies. This methodology has assumed an accurate reporting of deaths (but Dr Russell did concede some under-reporting of deaths53).

Estimated proportion of Covid cases being reported54

Country	W.H.O. data for 10 November		Estimated % of cases being reported
	Cases	Deaths
Pakistan	344,839	6,977	97%
Bangladesh	421,921	6,092	89%
Nigeria	64,184	1,160	97%
Democratic Republic of Congo	11,607	316	36%
Occupied Palestinian Territories	70,841	593	99%
South Sudan	2,960	59	81%
Myanmar	61,975	1,437	N/A
Uganda	14,574	133	45%
Yemen	2,070	602	7%
Syria	6,284	321	31%

Case study: Yemen

62.Mired in conflict for years, barely half of Yemen’s health facilities are fully functional. Ghassan Abou Chaar, until recently the Médecins Sans Frontières’ Head of Mission in Yemen, described the arrival of the first Covid-19 cases, in Aden in the south and soon after in Sana’a in the north, while fighting was still active. He said people arrived ‘already suffocating’:

… which is the worst cases of people seeking hospital care only when they are in a situation where they cannot get help at home from their friends or in a private health structure.55

63.Ghassan Abou Chaar said they had been surprised that the coronavirus had managed to enter the country. He told us that Yemen’s borders were already closed and that “There are five international flights a week coming in and out of the whole country, so it is already confined.”56 According to W.H.O. figures, even in October there had only around 2,000 confirmed cases but with 600 deaths. With that ratio it is perhaps not surprising that Dr Russell’s study identifies Yemen as the likely worst performer, reporting only an estimated 7.1%. of Covid cases.57

64.Yet, in July, a DFID press release announced: “UK calls for drastic action in Yemen as coronavirus infections reach one million” and went on to state:

Infections may have already reached one million, according to UK aid-funded research by the London School of Hygiene and Tropical Medicine which projects a worst-case scenario of up to 85,000 deaths.58

65.The Minister, James Cleverly MP, is quoted as saying during the (virtual) visit that: “This visit has allowed me to hear about the devastating impact coronavirus is already having in Yemen, and I was deeply concerned to hear that there have been over a million cases.”59 However, it seems the study in question was more of a scenario and not to be confused with a forecast or estimate.

66.But returning to Ghassan Abou Chaar’s evidence of surprise at the appearance of the coronavirus at all—Professor Checchi shared an interesting theory. He told us:

Yemen … is really quite an interesting case study that we are working on quite a bit. It is one of the few countries, to my knowledge, where almost no prevention of Covid transmission has taken place, unfortunately, and the anecdotal reports we are getting from inside Yemen are pretty consistent that the epidemic has “passed”. There was a peak in May and June across Yemen of cases and of hospitalisation facilities being overwhelmed, and that is no longer the case now.60

67.Professor Checchi said that study of the satellite imagery of the graveyards in Aden points to considerable excess mortality peaking in May. With no preventative measures, Professor Checchi’s suggestion for the lack of further escalation of infection was:

A very simple explanation, and one that does not require revisiting any of our current model assumptions, is that quite simply the epidemic burned out. It is possible to imagine that it was introduced into Yemen earlier than initially recognised. Remember, this is probably among all countries on earth one of those with the smallest testing capacity, particularly in the north. Let us imagine that the virus was actually introduced in February as opposed to April, when it was first recognised. You could predict that essentially the epidemic took off, ran its course and has now reached a situation where, at least temporarily, the population has accrued some kind of herd immunity.61