Covid-19: testing times

Dear Editor,

Went through with interest the article entitled “Covid-19: testing times” by NJ Beeching et al. [1]. The article clearly outlines the need for testing to overcome the challenges posed by Covid-19 to healthcare professionals. The authors reflect on “Test, test, test” as the key strategy suggested by the World Health Organization to control the spread of SARS-CoV-2 and its clinical manifestation. [2]. They conclude that tests cannot be interpreted if they are not available—and that is the major challenge currently facing many countries, including the UK. [1]

The demand and supply gap in making test kits available to the general population does not appear to be ending soon. Furthermore, the implications of testing for low and middle income countries (LMIC) like India, with a large population base are huge. The time, resources and logistics required to test are almost improbable and benefits are limited in view of their sensitivity and specificity. Therefore probably it is time to treat healthcare facilities (those with available laboratory facilities) in LMIC like India as sentinel sites for Covid-19 and use a set of criteria to rationalize the use of testing kits.

Major Criteria:
1. Inter-state travel in last 21 days
2. Inter-district, inter-city travel to hot spot areas in last 21 days
3. Vulnerable populations like migrants, daily wagers, laborers
4. Frontline workers in Covid-19 duty (for example, policemen on check post, healthcare workers).

Minor Criteria:
1. Inter-district, inter-city travel (non hotspot) in last 21 days
2. Shortness of Breath
3. Anosmia
4. Loss of taste
5. Fever (≥37.8°C)
6. Extreme Fatigue
7. Chills and Body aches
8. New cough (with or without sputum)
9. Nasal discharge or congestion
10. Diarrhea
11. Abdominal pain
12. Loss of appetite
Minor criteria from 3-12 should have a history of more than 3 days.

Recommendations for testing

A diagnostic test can be offered to patients presenting to health facilities (sentinel sites) fulfilling the following criteria:
1. Two major + One minor (if fulfilling 1 and 2 of major then exclude 1 of minor criteria)
2. One major + Three minor (if fulfilling 1 or 2 of major then exclude 1 of minor criteria)
3. Five minor

Dr. Sunil Kumar Raina
Professor and Head
Department of Community Medicine
Dr. RP Govt. Medical College, Tanda (HP), India-176002
Email: ojasrainasunil@yahoo.co.in

References:
1. NJ Beeching et al. Covid-19: testing times. BMJ 2020;369:m1403. https://www.bmj.com/content/369/bmj.m1403
2. World Health Organization. Laboratory testing strategy recommendations for COVID-19. 22 Mar 2020. https://apps.who.int/iris/bitstream/handle/ 10665/331509/ WHO-COVID-19-lab_testing-2020.1-eng.pdf.
3. Guidelines on Clinical Management of COVID – 19 – MoHFW. Available online at: www.mohfw.gov.in › pdf › Guidelines on Clinical Management. Last accessed; April 18, 2020.

Dear Editor

Serological test misclassifications and potential misapplications

Beeching, Fletcher, and Beadsworth suggested that rapid and wide availability for near-patient testing for covid-19 antibodies using flow devices similar to those used for pregnancy testing would be a game changer [1]. I have concerns regarding the veracity of such a perception.

All current serological tests for antibodies in plasma/serum (but not in urine) have an inherently high analytical, unavoidable error rate. This is due to different classes and subclasses of antibodies produced in response to infection, compounded by the huge array and diversity of endogenously produced immunoglobulin antibodies (around 10 billion) and which by cheer chance may interact with the test's reagent antibodies. False positive or false negative results occur even when the best available methodologies are used (accuracy >95%) and irrespective of the test’s formats. Urine is free from such matrix interference because it has no antibodies, hence the very high reliability of immunological tests for hCG in pregnancy.

We used Bayes' theorem to estimate the likelihood of erroneous result in different cohorts with information such as the rate of inherent analytical inaccuracy of the test and the prevalence of disease markers [2]. As the authors suggested, the majority (95% or more) of previously symptomatic infected patients who tested positive using reverse transcriptase (RT)-PCR are expected to have covid-19 antibodies [1]. Using a test with >95% accuracy, some 30% of the negative results in this cohort would be false negatives. On the other hand, in a cohort with expected low prevalence of covid-19 antibodies at 10% -- e.g. general population and health carers -- 17% of all positive results would be false positives, giving a false sense of security on the assumption that past infection confers some levels of immunity [1].

In summary, false positive rates are more predominant in cohorts with a low prevalence of covid-19 antibodies. On the other hand, in a cohort with a high prevalence of covid-19 antibodies erroneously false negative results are more prominent; misclassification and potential misapplication is therefore unavoidable. However, wide availability of testing for antibodies as the authors suggested would be desirable, correctly identifying most (but not all) individuals with and without covid-19 antibodies. Furthermore, if in doubt, follow-up tests could be carried out by the laboratory to identify erroneously false results [3].

Yours Sincerely

Adel A A Ismail
Consultant in clinical biochemistry and chemical
Endocrinology (RTD)
Wakefield

References
(1) Beeching NJ, Fletcher TE, Beadsworth MBJ. Rapid near patient testing for both current and past infections is urgently required, BMJ 2020; 369 doi: https://doi.org/10.1136/bmj.m1403
(2) Ismail AAA. When laboratory tests can mislead even when they appear plausible. Clinical Medicine 2017; 17, 329–32
(3) Ismail AAA. Identifying and reducing potentially wrong immunoassay results even when plausible and “not-unreasonable”. Adv Clin Chem 2014 ; 66 : 241 – 93

Dear Editor

In the BMJ editorial “Testing-Times”, Beeching and colleagues describe aggressive programmes of testing, contact tracing, and isolation contributing to early control of infection in South Korea.(1) Similarly, the editorial by Kickbusch et al. reminds us that “Citizens and experts outside the corridors of power are holding governments to account by comparing their response to that of other countries, to the relative success of South Korea or the relative failure of Italy”.(2) Indeed, benchmarking the UK government approach, which has been different to South Korea and its European neighbours, due to limited community testing and contact tracing,(3) could help inform the evolving strategy and response in the UK and elsewhere.

However, on April 1st 2020, during the UK government daily briefing, Dr Yvonne Doyle, the director of Public Health England, provided an interpretation of the international Covid-19 death statistics that could reflect the controversy about testing and contact tracing at the heart of UK politics and public health strategy. In terms of global death comparisons, Dr Doyle stated that “it has not been as severe here as in France…and, obviously, Italy is on a different trajectory”.(4) This does not account for the timeframe, or temporal evolution, of the crisis, which is country specific and, importantly, is also dependent on testing and the population size.

Using publicly available mortality data and demographic information presented on worldometers.info,(5) it is possible to estimate cumulative mortality adjusted for timeframe of crisis and population size. To standardise the timeframe, we designated the baseline (day 1) as the day that mortality rates exceeded 1 per 5 million population. This occurred in the UK, France and Germany on the 14th, 7th and 16th of March 2020 respectively. Italy passed this mark on the 27th February. The UK government statement occurred on day 19 and, contrary to the UK government’s statement on that day, the UK had higher Covid-19 death rates (n=2,352, or 35.4 per million population) than France (n=1,331 or 20.4 per million) and similar death rates to Italy (n=2,158 or 35.7 per million) when adjusting for timeframe and population size. Two days later, three weeks from passing 1 death per 5 million, the UK cumulative mortality had increased by 53% and was higher (n=3,605 or 54.3 per million) than Italy (n=2,978 or 49.2 per million). The timeframe and population size adjusted mortality remains higher in the UK than Italy and France at the time of writing.

Several comparisons have been drawn with Germany, where testing is 5-fold greater than UK and contact tracing is intensive. Germany’s day 19 cumulative mortality rate is less than half (n=1275 or 15.2 per million) that of the UK. Finally, the current timeframe adjusted cumulative death rate per million in the UK is now more than 40 times that of South Korea, which could serve for many countries as a benchmark of timely and comprehensive political and public health coordination during the pandemic.

There are many factors which could contribute to differences in mortality rates resulting in such as median population age, population density and urbanization. Furthermore, the full impact of downside risks from current clusters and overwhelming of Intensive Care Unit occupancy as well as the upside benefits of lockdown on mortality rates in the UK and Europe will not be clear for several months. Comparative death rates amongst countries with different Covid-19 responses may help inform governments, public health professionals and citizens about evolving strategies. However, the early response of the UK government was to pivot all testing capacity to identify people in hospitals who have got symptoms.(3)

The first two high-risk Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) cases had not met testing and case definition criteria.(6) Community testing of suspect cases and contact-tracing strategies in the UK remain at odds with best-practice and clear World Health Organisation recommendations over 2 months later. (7,8) Greater hospital support of primary care and community pharmacy is needed to triage patients and help manage high-risk chronic disease patients, such as those with heart failure, in order to avoid a second wave of non-communicable disease deaths. Indeed, overall mortality rates are likely to give the strongest indication as to international differences because of reporting differences and Covid-19 death rates in the UK reflect death rates in people who are tested, which is largely restricted to those attending hospital.

We agree with Beeching and co-authors that best-practice, involving aggressive programmes of testing, contact tracing, and isolation have contributed to mortality benefits in South Korea. However, we do not believe it is wise for the UK government to infer a better trajectory with respect to Covid-19 mortality than countries like France and Italy as it contemplates the “exit-strategy” from lockdown.

Yours sincerely

Mark Ledwidge
Chris Watson
Helen Coleman
Joseph Gallagher

1. Beeching NJ, Fletcher TE, Beadsworth MBJ. Covid-19: testing times. BMJ 2020; 369:m1403
2. Kickbusch I, Leung G, Bhutta Z, Matsoso MP, Ihekweazu C, Abbasi K. Covid-19: how a virus is turning the world upside down. BMJ 2020;369:m1336
3. Mahase E, Covid-19: UK Holds Off Closing Schools and Restricts Testing to People in Hospital. BMJ 2020, 368, m1060
4. UK Government Daily Briefing, 1st April 2020, by Alok Sharma and Dr Yvonne Doyle, https://www.youtube.com/watch?v=9mnNZ6lCdcw (8 minutes and 48 seconds from beginning), retrieved 11th April 2020
5. https://www.worldometers.info/coronavirus/, retrieved 11th April 2020.
6. Moss P, Barlow G, Easom N, Lillie P, Samson A. Lessons for managing high-consequence infections from first COVID-19 cases in the UK. Lancet VOLUME 395, ISSUE 10227, PE46
7. Peto J Covid-19 mass testing facilities could end the epidemic rapidly. BMJ 2020;368:m1163
8. Ryan M, WHO COVID-19 - virtual press conference - 30 March 2020. https://www.who.int/docs/default-source/coronaviruse/transcripts/who-aud...