Tuesday, May 17

No, a calculation does not keep Norway closed

Prime Minister Jonas Gahr Støre (Labor Party) and Minister of Health and Care Services Ingvild Kjerkol (Labor Party) during the press conference on 13 January.

The scenario must be evaluated based on what we knew when they were made – not how well they hit.

This is a chronicle. Opinions in the text are at the writer’s expense.

Aftenposten writes on 1 February about the National Institute of Public Health’s modeled scenario for the omicron variant. Critical review contributes to openness and to making us better.

Diverre makes Aftenposten, through both angle and content, that legitimate criticism drowns in delusions that only create confusion. When Aftenposten writes that «the models […] kept Norway closed », fortunately they were wrong.

The cases of Aftenposten have often made data and analyzes more accessible to readers than we ourselves have managed. It is therefore surprising that it is precisely Aftenposten that repeatedly interprets mathematical modeling in such an incomprehensible way.

Avisa’s journalists also mix concepts scenario and forecast. We will try to solve.

Useful contribution to the whole

Mathematical modeling is part of a broad knowledge base that we develop and deliver. The National Institute of Public Health’s advice is always given on the basis of a holistic assessment of many different sources of knowledge.

We recommend Aftenposten to read the department’s risk assessments, which we have co-published with each new omikron modeling report. The last one is 58 pages and contains a broad summary of Norwegian and international knowledge. The Norwegian Institute of Public Health’s modeling is one side.

Mathematical modeling is part of a broad knowledge base

Modeling is a useful contribution to the whole, partly because they show possible outcomes given the assumptions we make. The models are a way of summing up and condensing available, and often uncertain, knowledge.

The results are not safer than the assumptions we make. Throughout the pandemic, we have been keen to emphasize this.

Despite the uncertainty, the models are a useful tool. By modeling the scenario for future development, the models contribute to concretizing the picture and serve as a basis for professional discussion.

Wrong to claim that we “grossly miss”

The Norwegian Institute of Public Health has published a number of modeling reports and risk assessments related to the emergence of the omicron variant. In each report, we have made updates to the model and parameters in line with newer and increasingly reliable knowledge.

Aftenposten shows modeled and actual curves of new patients admitted with covid-19 and quite rightly points out that most scenario from 12 January shows higher numbers than the fasit.

The qualitative trend, with a curve that only goes down after Christmas, and then picks up again, has so far proved to be true. The actual numbers are at the lower limit of the uncertainty band in the scenario with the lowest numbers.

The scenarios should not be interpreted in such a way that the scenarios that lie in the middle of the outcome space are more probable than others. It is wrong to say that we are “grossly mistaken”.

The overall conclusions

In the report from 12 January, we overestimated the then level of spread of infection and how many elderly people were infected. Thus, the model found a wave that came faster, and with more inlays, than what we have seen in practice.

We quickly became aware of this, and subsequently report On January 26, we made adjustments. But this does not change the overall conclusions.

The model finds that if society now increases or has already increased its contact rate back to the level before the omicron, then we must expect a significant increase in infection and hospitalization rates in the weeks ahead. The top may be moving in time, but he’s coming.

A set of scenarios

Mathematical modeling of infectious diseases is associated with great uncertainties. This is largely due to the fact that it is practically impossible to predict how people will change their behavior in the future.

The only thing that is certain is that none of the scenarios will hit completely

Therefore, we do not manage to come up with forecasts or “weather forecast” for the microwave, but we have to content ourselves with presenting a set of scenarios. As with climate modeling, where scientists do not know how countries will cut or not cut their emissions and therefore model many different scenarios for the climate in the future.

In our latest set of scenarios, we try to use the model to show how different degrees of reopening of society can be reflected in infection and hospitalization rates. The only thing that is certain is that none of the scenarios will hit completely.

Meaningless to compare with the fasit

The models are complex. Dissemination of model results with all their restraints is challenging.

In the first omicron modeling, published December 13 in a situation where we knew very little about the omicron variant, we reported modeled scenarios that showed the potential for infection and admissions in an omicron wave if the rate of infection in society remained unchanged.

As is well known, measures against the spread of infection were introduced on 15 December, and the rate of infection dropped drastically. Thus, these scenarios, which presuppose a reality without measures, are no longer relevant.

One can easily criticize the premises for the scenarios, but it is pointless to compare them with the conclusion one and a half months later.

Now we know more and more that omikron is much less serious than delta and what effect the measures have had. We have continuously incorporated this knowledge into the modeling.

The scenario must be evaluated based on what we knew when they were made – not how well they hit.

More understanding and nuance

We try to be open and clear about what the models are and do. But they are complicated. It places great demands on us to make the results and preconditions understandable to more people.

Other disciplines have similar challenges that use complex mathematical tools in areas of great social significance.

We researchers have a special responsibility to communicate what we do in a clear way, so that democracy has the right to understand and discuss.

We hope that Aftenposten continues to write critically about our modeling, but with more understanding and nuance, the critique can become more useful.

The following have written under the article:

  • Jørgen Eriksson Midtbø, senior researcher, National Institute of Public Health
  • Louis Yat Hin Chan, Postdoctoral Fellow, Institute of Public Health
  • Francesco Di Ruscio, senior researcher, National Institute of Public Health
  • Gunnar Rø, senior adviser, National Institute of Public Health
  • Birgitte Freiesleben de Blasio, Department Director, National Institute of Public Health
  • Line Vold, department director and crisis manager, National Institute of Public Health
  • Camilla Stoltenberg, director, National Institute of Public Health


Leave a Reply

Your email address will not be published.