On 2017 Dec 08, Jesper M Kivelä commented:

Vivarelli and colleagues stated that late hepatic artery thrombosis (HAT) was diagnosed 0.4% (1/236) of patients with antiplatelet prophylaxis (AP) and 2.2% (13/592) of patients without AP. P-value for comparison between these groups was 0.049. The authors used Fisher’s exact test with SPSS version 10, and significance threshold for P-value was 0.05.

Based on my calculations, 2-tailed P-value is 0.130 (1-tailed 0.059) with Fisher’s exact test. I used three different statistical software (i.e. SPSS 22, Stata 12.1 and R 3.1.1).

Research colleague of mine tried to independently replicate (SPSS 22) the aforementioned result presented by Vivarelli and colleagues but without success. Our results, however, were identical.

Naturally, the proportion of late HAT cases is lower in patients with AP (0.4%) compared to patients without AP (2.2%) no matter what the P-value but the results should align with the methods used.

Of course, our calculations can also be wrong. However, we hope that Vivarelli and colleagues could re-examine their calculations.

On 2017 Dec 08, Jesper M Kivelä commented:

Vivarelli and colleagues stated that late hepatic artery thrombosis (HAT) was diagnosed 0.4% (1/236) of patients with antiplatelet prophylaxis (AP) and 2.2% (13/592) of patients without AP. P-value for comparison between these groups was 0.049. The authors used Fisher’s exact test with SPSS version 10, and significance threshold for P-value was 0.05.

Based on my calculations, 2-tailed P-value is 0.130 (1-tailed 0.059) with Fisher’s exact test. I used three different statistical software (i.e. SPSS 22, Stata 12.1 and R 3.1.1).

Research colleague of mine tried to independently replicate (SPSS 22) the aforementioned result presented by Vivarelli and colleagues but without success. Our results, however, were identical.

Naturally, the proportion of late HAT cases is lower in patients with AP (0.4%) compared to patients without AP (2.2%) no matter what the P-value but the results should align with the methods used.

Of course, our calculations can also be wrong. However, we hope that Vivarelli and colleagues could re-examine their calculations.