Monitoring

Content

• Content
• Monitor Evolution
• Monitor Prevalence
  • Different Counts
  • Motivation to Monitor
  • Numbers to Monitor
  • How to Monitor
  • Estimating Prevalence
• Testing Schemes for different Situations
  • Testing Priorities
  • Practical Testing Methods
• From Numbers to Measures

Monitor Evolution

[in work] From time to time full genome samples should be analyzed to monitor the evolution of Sars-CoV-2 and to deduce the international spread (e.g. done by Hodcroft et al). Dense sampling even helps to reconstruct the local spread (e.g. done by Brueningk et al).

Monitor Prevalence

To adjust the measures and to communicate these, it is important to estimate and monitor different incidence rates.

Different Counts

There are several counts to distinguish [not very clear on this page yet, in work]:

• Number of people getting currently infected. This number predicts how many people may need hospital treatment starting in roughly 1 to 3 weeks and being there in 1 to 4 weeks so they may need a place in the next 1 to 7 weeks. This number can only be estimated. The numbers depends on the protection measures and the count of currently infected (mostly those not knowing).
• Number of people with viruses detectable with PCR/Antigen tests. They may not yet be symptomatic but may get tested through sampling or contact tracing. They may or may not develop symptoms.
• Number of people with symptoms. Symptoms usually come between 1 and 10 days after the infection occurred.
• Number of people admitted to hospitals. (This number can be lowered in absolute emergencies with triage.)
• End of disease. Often about 10 days after symptom onset for a mild disease. The disease lasts usually much longer in severe cases and is fatal in about 1 of 400 cases which yields the death count.

Motivation to Monitor

• Determine control which control measures are optimal and/or needed depends on incidence rates. If the incidence rate is out of some bounds the measures may fail to work or be unnecessary strict:
  • Controlling methods relying on contact tracing need low incidence rates and detecting most cases with monitoring. To rely on contact tracing detection rate should be at least 75%. Still useful if lower but one shouldn’t rely on it.
  • Many controlling methods rely on methods reducing the infection probability but not preventing it (e.g. distancing, intermediate filtration face masks), that only works if there are not to many infected.
  • Shutting down businesses (badly ventilated offices and fabrics are the fare bigger risk than schools or outdoor gastronomy) may not be adequate at an incidence rate of 0.1% of currently infected, adequate to close targeted at 1% and all possible indoor settings at 10%.
• Planning in hospitals: Knowing the current incidence rates gives an estimate for the capacity needed in a few weeks.
• Communication: Knowing the cases helps on decision for everybody such if to avoid crowded spaces or staying home when having respiratory symptoms.

Numbers to Monitor

1. How many people have mild, severe, hospitalization requiring Covid or even deadly Covid.
2. How many of the people, which have respiratory diseases have Covid-19. In many areas likely the most common respiratory disease is Covid-19 and so one should stay home with any respiratory symptoms.

   Personal guess, 16.10.20: In many regions of Europe, the most common respiratory disease is Covid-19. Moreover the steep increases in cases despite protection measures in place including strict quarantine of known infected, indicate that across Europe the majority of cases are undetected and carry the spread. (Example Calculation of Spread in the section Handling Covid-19 Cases and Quarantine)

3. The incidence of Covid-19 in different population groups (e.g. hospital workers, young people: students and school children)
4. Other respiratory diseases: influenza and common colds should be monitored too:
   • knowing what is around makes diagnosis easier, especially if it turns out there is ‘only’ Covid-19
   • long term adverse effects of some common cold viruses are likely underestimated. This is a good opportunity to diminish there prevalence and reduce the number of strains circulating.

How to Monitor

• Frequency: Ideally these are taken near daily, else at least weekly. Though this depends
• Communication: Often (as of 16.10.20) only the positive tests and the positive test rate are communicated. However more complete data would allow more accurate interpretations:
  • many of the tested often are health care professionals (who know how to protect or have an increased immunity and incidence is thus underestimated),
  • how many of the tested are healthy and just need some clean sheet to travel,
  • how many who are tested have symptoms,
  • how many are tested because of company monitoring.
• Methods: see also testing priorities
  • very low to low incidence: case counts
  • low to mid incidence: case counts people turn up for testing can be used. The contact tracing yields then additional cases which can be used to scale counts from those who asked for tests.
  • mid to high incidence: sampling

Estimating Prevalence

For practical reasons the overall prevalence has to be estimated.

Validate Estimates

The equation current_death_count = estimated_past_case_count * fatality_rate can be used to detect if the estimation methods worked in the past. How fare in the past depends how long on average the patients survive and whether takes the date of infection or the date of going for a test (Different Counts). Assuming a fatality rate of 0.25% is equal to one in 400 Covid infectees dies. So the estimated_past_case_count = 400 * current_death_count The case can be different yielding a correction factor: correction_factor = estimated_past_case_count/past_case_count_from_death_count If the estimation method didn’t change the current actual infected can be estimated with the correction_factor: new_estimated_case_count = correction_factor * estimated_case_count.

On Estimates Out of Bounds

A correction factor greater than 2 needs a redesign of the case estimates. A correction factor larger than 5 combined with a death count greater than 2.5 per million per day I consider worrying (Sections Strategy and On Numbers). How much worrying depends whether, when and how effective the measures have been adapted since. Transmissions should be reduced or even prevented until a clear view is available. In serious situations (possible hospital overflow), a tactics uniform across movement areas and between them movement monitoring/control/reduction can be necessary.

1. Pausing infections especially everything super-spreading opportunities for a few days until testing and estimates are available. Do outdoor days and possibly some restrictions on indoor rooms:
   • closure of non essential businesses
   • for somewhat essential businesses e.g. longer distance public transport: names can be recorded and people need to hand in some argumentation in the next weeks why it was essential to enter. This causes few immediate work/tumults but likely discourages most from unnecessarily entering.
   • whitelist approach if doable e.g. negative test/no fever required
   • pause of non essential visits to hospitals and so on
2. Sample and estimate the numbers
   • should be doable within a few days
   • 5’000 tests per million inhabitants should work to estimate to infection rate with an accuracy of .5% (Uniform -random- sampling)
3. Choosing a tactics suiting the strategy given the situation

Testing Schemes for different Situations

Testing Priorities

partly subjective section

Testing some for low and high incidence. Reasons are noted or already mentioned in the section Numbers to Monitor.

Testing Priorities at High Incidence

1. Testing of risk patients. Reason: Appropriate treatment and prevention of spread (to other risk persons).
2. Health care workers and other groups with contact to risk groups. Reason: Shield risk groups.
3. How many people with respiratory disease symptoms have Covid-19. Reason: Appropriate advices for those experiencing respiratory symptoms.
4. Tests to do contact tracing. Reason: Break infection chains.
5. Sampling different population groups e.g. young people: students and school children, region based, incoming travelers. Reason: Have an overview what’s going on to adapt measures.
6. Monitoring respiratory diseases: influenza and common colds.
7. Tests for travel permissions, quarantine cancel and so on.

Testing Priorities at Low Incidence

1. Testing of risk patients. Reason: An appropriate treatment from early on.
2. Health care workers and other groups with contact to risk groups
3. Tests to do contact tracing and quarantine cancel
4. Tests to estimate the overall incidence (travel test below can give an indication). Reason: When the incidence is high, a different tactics for controlling can be appropriate.
5. Tests for travel permissions and so on.

Practical Testing Methods

Depending on the situation different testing methods are adequate. Early morning saliva has a good detection rate of about 70% (high specificity with PCR; slightly lower sensitivity and lower specificity with antigen test.) and is easy to sample. So for most situations early morning saliva is recommended. Further discussed in the chapter Diagnosis and Viral Load section Diagnosis Suggestions by Test Goal.

From Numbers to Measures

[in work]

Assumptions and Comments

• Fatality Rate: Fatality rate of .25% i.e. 1 in 400 (The death rate depends on many factors and varies considerably);
  • The fatality rate depends on many factors and varies considerably for different regions and population groups.
  • The numbers below are for a population without immunity in risk groups (mainly people without a good immune system). Increased resistance can be increased
    • via the acquired immune system by previous infections or vaccines
    • via the overall immune system by a healthy living including daily movement.
    • As it is known by summer 21, overly strict measures can diminish the immune defense against covid.
• Numbers:
  • The numbers vary across regions and hospitals capacities, population structure and how the infections are distributed across the population.
  • The current numbers need to be estimated. The death count detects infections happened about 20 days in the past. The counts from contact tracing or testing symptomatic are a few days in the past.
• Covid Cases:
  • average disease duration 10 days;
  • The case counts are the actual infected and not only the positive tested.
  • The case counts/incidence can be higher for young people without directly affecting the death counts/hospital beds, so the case counts/incidences are averages for the population groups with the possibility for severe Covid.
  • the Covid cases should be lower in dry air settings (e.g.during cold seasons), since aerosol transmission yields more severe cases and is hard to control and prevent. Immune protection can be built up in humid-air seasons (e.g. summer in moderate climate zones) but not in dry-air seasons.

Notations and Equivalences

• all counts are day counts unless noted otherwise
• case count per million inhabitants per day = 400 * death count per million inhabitants per day (assuming 1 death in 400)
• total current infected per million = 10 * case count per million per day (assuming 10 day infection duration)
• K = 1000 e.g. 10 K = 10’000

On Numbers

written for autumn/winter 2020; as of summer 2021 the goal is with appropriate preparations such as healthy living and vaccination of high risk groups and early diagnosis combined with early adequate treatments to reduce Covid deaths while keeping overall health in mind (Following the Strategy discussed on the controlling page). Most measures (including vaccines) have their costs and risks