| 16:59 | | Questions. |
| 17:05 | | Sweden Radio journalist: I have a question for the Public Health Agency, |
| 17:11 | | regarding the model used to show the spread in Stockholm, |
| 17:17 | | which was presented yesterday and was withdrawn this morning “due to serious faults.” |
| 17:23 | | What are these serious faults? And, does that also indicate that the peak |
| 17:28 | | which was believed to have already been reached on April 15 isn’t correct? |
| 17:33 | | We are working on it, and we will release an updated version tomorrow. |
| 17:39 | | But to say it briefly without getting very technical: Such models are difficult. |
| 17:45 | | We had a wrong variable from the beginning and the number of cases weren’t updated in that model. |
| 17:54 | | That variable, in its turn, caused an increase in another variable |
| 17:59 | | to reach the benchmark, the 2.5% which was calculated. |
| 18:04 | | Therefore, the variable was very high. |
| 18:07 | | That variable was the difference between the discovered |
| 18:10 | | and the undiscovered cases. That’s why it was very high. |
| 18:13 | | Many on our side have read it and we should have had noticed it. |
| 18:17 | | We are happy that someone has detected it and, |
| 18:20 | | we were able to stop it in time before it was spread widely! |
| 18:23 | | But it doesn’t cause changes in the other parts, such as the peak; |
| 18:26 | | it will stay approximately at the same level. But we will get back to this tomorrow. |
| 18:29 | | We will read it again, and will get back tomorrow. |
| 18:33 | | The model was off for a period of time. Aren’t you worried |
| 18:36 | | that it would somehow harm your credibility? |
| 18:39 | | I hope not. With very good help from outside, |
| 18:43 | | the mistake was swiftly detected and can be corrected. |
| 18:48 | | The basic assessments of the model still stand and are valid. |
| 18:57 | | TV4 journalist: I would like to ask a similar question, |
| 19:03 | | concerning the report you published yesterday, and it contained mistakes! |
| 19:09 | | What is in that model that still stands and valid? Was the peak on April 15 or not? |
| 19:14 | | It will reach there on the day. The peak varies by a few or some days apart. Not longer. |
| 19:23 | | What about the assessment that two-thirds of Stockholm’s residents |
| 19:27 | | would be infected by the beginning of May. Is that still correct? |
| 19:31 | | Yes, it is approximately at the same level. —So what was the mistake? |
| 19:35 | | The mistake was about the compensation effect generated in the model, which proposed that: |
| 19:42 | | “Behind every case we found there are one thousand more cases in the community.” |
| 19:46 | | That was wrong and we are correcting it. |
| 24:20 | | I would like to ask about the peak reached “some day” “between” April 15th and 18th. |
| 24:29 | | Dr. Per Follin, who is an Infectious Disease Control doctor, |
| 24:34 | | has experienced that cases are increasing in Stockholm County. |
| 24:40 | | How can your view be very different? |
| 24:44 | | I don’t know if he says that it’s increasing, but rather that we should keep it consistent. |
| 24:48 | | The peak we are talking about is the spread of the disease in respect to when it is the largest. |
| 24:53 | | It takes about a week until you get sick, and several more days |
| 24:57 | | to require medical care in hospitals, if needed. |
| 25:01 | | So it takes roughly ten days for the peak to reach |
| 25:05 | | the medical care system. We haven’t even come that far yet. |
| 25:09 | | What he experiences is real in that sense. It’s currently a flattened curve. |
| 25:15 | | But we understand the concern in Stockholm, because it approaches the limit of capacity. |
| 25:21 | | We sincerely want to go through it. So that is important, |
| 25:25 | | especially in Stockholm, to stay home and limit social contacts. |
| 25:30 | | We have encountered this for a long time. I understand the concern |
| 25:35 | | when we come close to something which can be an improvement. |
| 25:40 | | Emanuel Karlsten is with us online. |
| 25:45 | | I am wondering about the graphic of the number of the COVID-19 deaths. |
| 25:54 | | It is a huge time lag! What you are showing here today is from March, |
| 25:58 | | which makes the data three weeks old. |
| 26:04 | | But you, Anders Tegnell, still stated that |
| 26:08 | | “We have reached the plateau [flattening the curve].” |
| 26:13 | | Is it reasonable to conclude that it is actually not a plateau, |
| 26:16 | | but it is more accurately a time lag? |
| 26:19 | | How are we to understand it? And, how would that affect the model? |
| 26:25 | | We are not using the number of deaths in the model. |
| 26:29 | | Because they are uncertain in many senses, such a case time lag. |
| 26:33 | | The model is built comprehensively on the number of diagnosed cases. It doesn’t affect it. |
| 26:39 | | I have said numerous times that the number of deaths is significant in many senses, |
| 26:43 | | but not in regard to the model’s composition! The lag is very large. |
| 26:48 | | It takes a quite long time between infection and decease. It is not a good tool to control. |
| 26:55 | | We have apparently run a quality match against the number of registered deaths. |
| 27:04 | | And we have found a number of cases where they passed away long after they were infected. |
| 27:11 | | For some reason, they weren’t detected in the medical care system, but we are currently |
| 27:16 | | running a quality control process, and we can detect them. |
| 27:21 | | We are doing it approximately once a week and we sometimes find cases which date back a ways. |
| 27:27 | | When we reach out now and ask about deaths because of COVID-19, |
| 27:31 | | we get unusual answers which we weren’t getting earlier. |
| 27:35 | | It could possibly be because of the Easter Holidays. So the staff had a lot to do, |
| 27:39 | | and perhaps they didn’t have time to register the death cases. |
| 27:44 | | Is it your view that you are currently on a flattened curve? How to assess it? |
| 27:52 | | It has spread out across so many days and doesn’t affect what the curve looks like. |
| 28:00 | | It’s apparent specifics in the report included mistakes, and such criticisms were raised |
| 28:06 | | against it … but not against the earlier models, because |
| 28:12 | | you aren’t transparent in regard to which models you are using to reach such conclusions. |
| 28:16 | | Are you going to be more transparent in the future in regard to |
| 28:19 | | which mathematical models you are using to conclude your results? |
| 28:22 | | We have always been transparent. |
| 28:25 | | And the model will be published and be available for everybody. |
| 28:29 | | Everybody can download the data from our website. |
| 28:32 | | So I don’t know in which part we aren’t transparent. |