On the weekend, there was news of hope that we could get rid of the current corona disaster. Clinical trial of oral drug to treat new coronavirus under development by Pfizer (No. 1) 2/3 In the interim analysis of the phase study), the risk of becoming severe 89% The result of the reduction was reported.
In this article, this drug, Paxlobido (PAXLOVID) But the risk of aggravation 89% We will tell you what kind of result was used to calculate the reduction. At the same time, of risk 89 % Decrease is an estimate obtained from a limited number of patients, so we will calculate and consider the confidence interval for the estimate.
Pfizer homepage According to the news in, a group of non-hospitalized adult patients at risk of aggravation who received random paxlobide and a placebo. ( placebo ) As a result of allocating to the group to which was administered, after registration 28 We are comparing the percentage of patients hospitalized within a day.
Paxlobide administration: 389 Philtrum 3 Person is hospitalized (no death)
Placebo administration: 385 Philtrum 27 Person is hospitalized (after that 7 Name died)
■ Comparison of hospitalization risks
To get the risk reduction figures in the title, you need to calculate the relative risk. Therefore, JMP creates the data as follows. The column "Group" corresponds to the drug administration group, the column "Hopitalized" corresponds to whether or not the patient was hospitalized as a result (Yes, No), and the column "N" corresponds to the values of "Group" and "Hospitalized". Enter the number of people to do.
After this, you'll usually find it using the "Bivariate Relationship" (Fit Y by X) platform, but for now, let's do the same with the "Categorical" platform.
Select Analysis> Consumer Survey> Categorical from the menu bar and specify the columns as follows: The point is to specify "N" as [Frequency].
After specifying, select the option Relative Risk. Then, you will be asked for the target level and the target sample level. Select "Yes" and "PAXLOVID" respectively. That is, it specifies that the hospitalization rate for the Paxlobide group is compared to the hospitalization rate for the placebo group.
The report displays a cross-tabulation table and a graph (share chart) showing the percentage of each group. As shown in the cross-table, the hospitalization rates for each treatment group are 0.77% (= 3/389) for the Paxlobide group and 7.01% (= 27/385) for the placebo group.
At the bottom of the report is a relative risk report. In this example, of the "Paxlobide administration group" ” The relative risk of hospitalization for a placebo-treated group is calculated as follows:
Relative risk = = Percentage of hospitalizations in the Paxlobido-administered group ÷ Percentage of hospitalizations in the placebo-administered group
= (3/389) ÷ (27/385)
= 0.11
That is, when Paxlobide is given, the relative risk to taking placebo is 0.11 (11%) It shows that it decreases to. Considering how much the risk of hospitalization is reduced from this 89% ( 100% - 11% ) Is the number reported in the news.
■ Confidence interval for relative risk
Relative risk value calculated in the previous section (= 0.11) Is an estimate for a limited number of patients included in the study. Therefore, it is better to consider not only the estimated value but also its confidence interval.
Let's look at the relative risk report again. The 95% confidence limit is displayed.
Relative risk from this (= 0.11) Against 95% The confidence interval is approximately 0.03 ~ 0.36 You can see that it is an unexpectedly wide section. So, to put it roughly, the reduction in the risk of hospitalization is 64% ~ 97% I will think with a range. 64% I feel that it is quite effective even if you think low.
Not surprisingly, the larger the number of samples, the more credible the estimation is and the narrower the confidence interval. Temporarily, both the number of admins and the number of hospitalized patients in this clinical trial Ten Suppose it was twice as many. That is, the Paxlobide administration group 3,890 Philtrum 30 Inpatients, placebo-administered groups 3,850 Philtrum 27 A person will be hospitalized, but the relative risk at this time is the same as before 0.11 but, 95% The confidence interval is 0.08 ~ 0.16 is. Compared to the actual example, the confidence interval is considerably narrower.
The results of this interim analysis show that it is sufficiently effective and safe, and it seems that an application for approval will be made in the future, but if it is approved without problems and can be used for actual corona patients, patients It seems that you can enjoy great benefits both on the side and on the medical institution side. By that time, you may be waiting for a life where you don't have to wear a mask in public, like before Corona.
