The incidence of COVID-19 worldwide is not only going up, but the rate of growth is showing a steep rise as well. The broad range of clinical phenotypes has made it clear that better clinical screening based on explicit knowledge of COVID-associated symptoms is required, both to diagnose the disease and to monitor cases.
A recent study published on the preprint server medRxiv* in October 2020 reports the utility of alterations in taste and smell to monitor the prevalence of COVID-19 among contacts and in a community that lacks unlimited testing resources.
Early Clinical Diagnosis
At the beginning of the pandemic, the variety of clinical signs and symptoms was mostly unknown, except for the presentation with fever, non-productive cough, and breathlessness. Most descriptions of the clinical features were derived from hospitalized patients, since at that time, the scarcity of testing kits, as well as the large number of patients who required medical care, dictated that more severely ill patients be prioritized for testing. Most of these cases did have respiratory symptoms and fever.
As testing became more widespread and population-based, data from outpatients presenting with mild to moderate illness has become available. However, there is a lack of studies dealing with the issue of whether earlier definitions of COVID-19 symptoms still meet the need for case surveillance when it comes to contacts who have mild to moderate symptoms after coming into contact with patients who have tested positive for the virus.
Existing Definitions Lack Discriminatory Power Without Testing
The researchers from the Centers for Disease Control and Prevention, Utah Department of Health, and City of Milwaukee Health Department found nine papers altogether, five of which assessed self-reported symptoms. Three of these were conducted among healthcare workers. These concluded that criteria using abnormalities in taste and smell were of value in screening contacts for COVID-19.
Two were in the general population, one of which used self-reported symptoms and the other both symptoms and laboratory reports collected by smartphone. These arrived at a similar result.
Other researchers have constructed a model to convey the enormous impact of repeatedly revising COVID-19 case criteria on the number of reported cases in China. And finally, one paper shows that in the case of influenza, even case definitions already in place for surveillance fail to perform uniformly in children and adults alike when it comes to differentiating cases from uninfected individuals.
The authors of the current study point out that this may be the case with COVID-19 as well, with the discriminatory power of the surveillance criteria being age-dependent.
Some researchers have turned to the use of predictive algorithms to diagnose mild to moderate cases as well as to predict the risk of hospitalization. These studies explore only one or a few clinical features, however, or include the use of blood tests or imaging, thus ruling out their applicability to remote screening using a broad spectrum of symptoms.
Sensitive Case Definitions Lack PPV
The researchers identified the high sensitivity of the COVID-19 case definitions in everyday use at present, at over 85%, but with a low positive predictive value (PPV), since only a third to a half of positive screens actually have the disease.
Overall, therefore, current definitions of COVID-19 cases do reflect disease symptoms accurately, but throw up many false positives and do not help estimate the community prevalence. The most sensitive existing criteria overestimate the prevalence, while the most specific definitions underestimate it by up to 80%, particularly when testing resources are limited.
Either more timely and widely available testing, or more reliable definitions, are necessary for community surveillance to be more valuable to public health efforts to contain the pandemic.
The latter may be arrived at by using combinations of unusual and specific symptoms such as anosmia and ageusia as a high-priority item. The worst performance appeared to be in children, probably because such nonspecific symptoms are associated with many common pediatric illnesses. Better definitions need to be devised for children who often show different combinations of symptoms.
Study Objectives: Prevalence and Screening
The current study was aimed at assessing existing case definitions and new symptom combinations for their ability to screen outpatients for the disease and help estimate the community prevalence accurately.
The researchers used daily symptom diaries and PCR tests on respiratory specimens taken daily over two weeks, from 185 individuals who had been exposed to household contacts. They then searched through several hundred thousand combinations of 15 symptoms associated with the disease.
They used combinations of 1-15 symptoms to identify those with the highest discriminatory power. These were applied to all contacts, whether positive for COVID-19 or not. They aimed to find novel case definitions with the highest sensitivity-to-PPV balance. They narrowed down on four combinations, which had the highest score in this regard.
They found that the most valuable symptoms turned out to be alterations of taste or smell, which also had the advantage of improving the PPV to 84% concerning the sensitivity (63%). The best symptom combination required either loss or dysfunction of taste or smell to be present, or two of the following: shortness of breath, wheezing, discomfort in breathing, fever/chills.
With this combination, the PPV was 79%, sensitivity 78%, and the sensitivity-to-PPV balance 78%. It predicted the community prevalence almost perfectly as well. Such criteria are invaluable when resources are limited and false positives are unwelcome. When resources for testing and dealing with the latter are freely available, of course, existing criteria with much higher sensitivity are preferable.
The following conclusions may be drawn from these findings. Firstly, existing surveillance definitions are sensitive for COVID-19 detection, especially among household contacts of known cases, but are poorly predictive of the prevalence and have a high false-positive rate. The presence of taste/smell alterations could help build a potent and rapid case definition, to distinguish potential COVID-19 positives among contacts through case surveillance and thus interrupt viral spread.
Such definitions are also useful for large clinical trials to reduce the number of required serial SARS-CoV-2 tests needed in any given individual who reports a symptom. Also, case definitions to screen children for COVID-19 accurately are not the same as for adults, and this is an area requiring more work.
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.