Hospitalized with covid or by covid? Here’s the question…
Since the beginning of the pandemic we have tried to base decision-making on data. Different data that, although with room for varied interpretations, guided us on when and with how much intensity to implement or withdraw control measures.
The health administrations of practically all the countries have been informing us of the transmission, hospitalizations, admissions to the ICU, deaths and the pace of vaccination. These data have generally been offered by age groups and, at times, by vaccination status. Very relevant information for decision-making, given that the risk of developing severe covid increases in older people and in unvaccinated people.
Transmission is no longer relevant
In the first waves, with the population not yet vaccinated or only partially vaccinated and with especially aggressive variants, the transmission indicators were especially useful. They announced what was coming in the hospitals a few days in advance and with some precision. In addition, they made it possible to adopt measures with minimal anticipation regarding hospital data.
After the extraordinary vaccination campaign carried out in most European countries, and the no less extraordinary omicron wave, the situation changed. Now the population enjoys significant immunological protection, most cases are mild and the relationship between transmission and hospitalization has changed significantly.
These changes have led many countries to lift restrictions, despite maintaining very high transmission figures. And also to modify its indicators. Those of transmission have lost interest and decision-making is based more and more –and it is very reasonable– on those of hospitalization.
The consequences of basing decisions on hospitalization data
Hospital admission approximates the incidence of severe or moderate disease. Therefore, it is a much more important data for decision-making than the incidence of some cases that are mostly already asymptomatic or barely symptomatic. And it is earlier than ICU admissions or deaths.
The main problem with basing decision-making on hospitalization data is the definition of “hospitalization for covid-19” itself. A definition that does not refer to the cause or severity of admission but to the presence of a positive result in a diagnostic test for covid-19 (PCR or antigens). Regardless of the patient’s condition, and even when the cause that motivates hospitalization is not coronavirus infection.
For this reason, health services have begun to use the concepts of covid primary to refer to those cases in which the primary cause of hospitalization is covid, and incidental covid to refer to those admissions with a positive result in PCR or antigens, but in which covid is not the cause of hospitalization.
Plainly, primary covid means they enter for covid. Incidental covid means they enter with covid, but not for covid.
As most hospitals sensibly routinely test virtually all patients who are hospitalized and transmission is still high, hospitals are seeing a growing number of covid cases incidental that go on to swell the government statistics of hospitalizations for covid and make it difficult (magnify) the interpretation of the epidemiological situation.
Measure primary covid and incidental covid
Beyond anecdotal reports (from a hospital, in a news item in the media, etc.), there are few studies on the proportion of incident covid in hospitals. In addition, few administrations have implemented this differentiation, except for a few specific cases in the United States, Wales and Canada.
We know from some unreviewed article (preprint) that the proportion of primary hospitalizations (or, its inverse, incidental) varies ostensibly over time and in different places. For example, the proportion of hospitalizations for primary covid in 4 US hospitals in different states during 2020-21 ranged from 100% in the spring of 2020 to only 35% in the early fall of 2021. And it also differed markedly between hospitals.
The current figures occasionally reported by some countries in which the omicron variant is predominant are also variable. They depend on the transmission situation in each region (in highly vaccinated populations, a higher proportion of incidentals the more transmission), on the operational definition of incidental adopted and the sources of information.
At the end of April 2022, the Massachusetts Department of Public Health in the United States reported 31% primary covid among patients hospitalized for covid, leaving two-thirds of hospitalizations as incidental, which is a lot. Their very simple but probably useful definition of primary covid was having received dexamethasone treatment, which can be considered a surrogate indicator of moderate or severe covid.
Around the same time, the Welsh Department of Health estimated that only 19% of those hospitalized with covid were being “actively treated”, an equivalent of primary covid. In this case, the operational definition of “actively treated” is more confused, and also heterogeneous. In Canada, in February 2022, British Columbia reported 44% of incidental hospitalizations over total hospitalizations with covid.
Other preprint very recently, based on a single hospital in the Netherlands and only with patients infected by omicron, it estimated 66% of those hospitalized with PCR+ in which covid was the cause – primary or a contributing cause – of admission.
Is it useful to differentiate primary from incidental covid?
At this time of the pandemic, yes. It is not only useful, but essential for decision making. Transmission is growing and, with transmission, hospitalizations are growing with and also, for covid. But the health response is very different depending on the proportion of each one.
For example, if primary hospitalizations of the elderly or nursing homes are increasing, one option to consider is administering a 4th dose (second booster) of the vaccine. If the growth is primarily due to incidental hospitalizations, we can wait and try to give the second booster along with the flu shot. Perhaps with new or modified vaccines if they become available in the fall.
Other similar examples are the decisions on the return of isolation, masks indoors, sick leave of cases and contacts, etc. These are decisions that do not depend on the incidence of cases but on the incidence of serious cases. That is, of primary hospitalizations.
However, the relevant data is confounded by incidental hospitalizations (remember patients with covid), which prevents an adequate interpretation of the seriousness of the situation. Imperfect information leads to speculative debates and decisions with a more political than scientific background. An example would be the positions that are being adopted in the current debate on the urgency of administering a 4th dose in the elderly and residences.
Incidental covid should not be trivialized. It is clear that it delays scheduled admissions and forces a lot of hospital activity to be rescheduled, something that is not always possible in such short periods of time. There are efficiency losses and increased waiting. In cases where hospitalization cannot be delayed, it requires isolated rooms, use of PPE, surveillance and other measures to prevent contagion from spreading to professionals and other patients. And, sometimes, an incidental case on admission is complicated, or complicates the primary disease that led to admission.
Update, inform and decide
By the way, the transmission should not be considered insubstantial either. Many mild cases are a lot of work for primary care, which still has a lot to catch up on. And the increase in cases in general always leaves a sediment of serious cases. Some actions must be taken, even if they are only social communication, to improve adherence to the Ministry’s own recommendations.
Information systems must be useful for decision making. And they must be modified when the situation has changed and the decisions to be made require new data or more precise data. Differentiating hospitalizations for primary or incidental covid has been important since the middle of the sixth wave. And, to the extent that hospitalizations grow that we cannot interpret without much uncertainty, it is also urgent.