Reader Comments

Post a new comment on this article

The numbers are wrong

Posted by Spiritosl on 05 Apr 2010 at 12:11 GMT

Sirs!

This article is interesting. All eight studies included are from the previous century.

I can not find that the numbers are correct.

The sum of % Weight in the last column of figure 2 is 100.20 instead of 100,00.

The sum of Events is 1072. When I divide Events in each of the eight lines with total # of events I get another figure of % Weigth in the last column, I get 124/1072= 11,56 % instead of the first line’s 13,44 %.

When I calculate 13,44% * 1072 (total # of events) I get the answer 144 Events in the first line (LA Veterans) instead of 124 Events.

Summing up all events in a similar way they end up to be 1074. That is two more than the sum of events in the first column. It also explains the discrepancy of sums of % Weight, 100.00 written in the table vs 100.20 summed in the very same column .

So all numbers in that last column are wrong.

Are there any numbers that are correct?

How come this article is peer reviewed and the peer did not find this discrepancy?

I am baffled

No competing interests declared.

RE: The numbers are wrong

npatel replied to Spiritosl on 11 Apr 2010 at 20:27 GMT

Mr. Hammarskjöld, your concerns about the accuracy of the numbers are unfounded. The weights shown in the last column are rounded, which would explain why you calculate a sum of 100.02%.

The sum of the events is 1042, as reported in the text of the paper. Furthermore, the weights are derived using inverse variances - not the number of events, which is why your calculations yield a different result.

No competing interests declared.