Web references
w1.Tsang JC, Morin JF, Tchervenkov CI, Platt RW, Sampalis J, Shum-Tim D. Single aortic clamp versus partial occluding clamp technique for cerebral protection during coronary artery bypass: a randomized prospective trial. J Card Surg 2003;18:158-63.
w2. Bishai D, Qureshi A, Cantu N, Parks C. Contracting with children and helmet distribution in the emergency department to improve bicycle helmet use. Acad Emerg Med 2003;10:1371-7.
w3. Lykke N, Lerud PJ, Stromsoe K, Thorngren KG. Fixation of fractures of the femoral neck. A prospective, randomised trial of three Ullevaal hip screws versus two Hansson hook-pins. J Bone Joint Surg Br 2003;85:426-30.
w4. Thies F, Garry JM, Yaqoob P, Rerkasem K, Williams J, Shearman CP et al. Association of n-3 polyunsaturated fatty acids with stability of atherosclerotic plaques: a randomised controlled trial. Lancet 2003;361:477-85.
w5. Sato Y, Metoki N, Iwamoto J, Satoh K. Amelioration of osteoporosis and hypovitaminosis D by sunlight exposure in stroke patients. Neurology 2003;61:338-42.
w6. Gaster AL, Slothuus Skjoldborg U, Larsen J, Korsholm L, von Birgelen C, Jensen S et al. Continued improvement of clinical outcome and cost effectiveness following intravascular ultrasound guided PCI: insights from a prospective, randomised study. Heart 2003;89:1043-9.
w7. Sheu BS, Huang JJ, Yang HB, Huang AH, Wu JJ. The selection of triple therapy for Helicobacter pylori eradication in chronic renal insufficiency. Aliment Pharmacol Ther 2003;17:1283-90.
w8. Arshad SH, Bateman B, Matthews SM. Primary prevention of asthma and atopy during childhood by allergen avoidance in infancy: a randomised controlled study. Thorax 2003;58:489-93.
w9. Hughes JR, Novy P, Hatsukami DK, Jensen J, Callas PW. Efficacy of nicotine patch in smokers with a history of alcoholism. Alcohol Clin Exp Res 2003;27:946-54.
w10. Leon C, Alvarez-Lerma F, Ruiz-Santana S, Gonzalez V, de la Torre MV, Sierra R et al. Antiseptic chamber-containing hub reduces central venous catheter-related infection: a prospective, randomized study. Crit Care Med 2003;31:1318-24.
w11. Trivedi DP, Doll R, Khaw KT. Effect of four monthly oral vitamin D3 (cholecalciferol) supplementation on fractures and mortality in men and women living in the community: randomised double blind controlled trial. BMJ 2003;326:469.
w12. George TP, Vessicchio JC, Termine A, Jatlow PI, Kosten TR, O'Malley SS. A preliminary placebo-controlled trial of selegiline hydrochloride for smoking cessation. Biol Psychiatry 2003;53:136-43.
w13. Mion LC, Palmer RM, Meldon SW, Bass DM, Singer ME, Payne SM et al. Case finding and referral model for emergency department elders: a randomized clinical trial. Ann Emerg Med 2003;41:57-68.
w14. Basnyat B, Gertsch JH, Johnson EW, Castro-Marin F, Inoue Y, Yeh C. Efficacy of low-dose acetazolamide (125 mg BID) for the prophylaxis of acute mountain sickness: a prospective, double-blind, randomized, placebo-controlled trial. High Alt Med Biol 2003;4:45-52.
w15. Neumann FJ, Kastrati A, Pogatsa-Murray G, Mehilli J, Bollwein H, Bestehorn HP et al. Evaluation of prolonged antithrombotic pretreatment ("cooling-off" strategy) before intervention in patients with unstable coronary syndromes: a randomized controlled trial. JAMA 2003;290:1593-9.
w16. Fjaertoft H, Indredavik B, Lydersen S. Stroke unit care combined with early supported discharge: long-term follow-up of a randomized controlled trial. Stroke 2003;34:2687-91.
w17. Wood E, Kerr T, Spittal PM et al. An external evaluation of a peer-run "unsanctioned" syringe exchange program. J Urban Health 2003;80:455-64.
w18. Suter NM, Malone KE, Daling JR, Doody DR, Ostrander EA. Androgen receptor (CAG)n and (GGC)n polymorphisms and breast cancer risk in a population-based case-control study of young women. Cancer Epidemiol Biomarkers Prev 2003;12:127-35.
w19. Brassard P, Bourgault C, Brophy J, Kezouh A, Rainville B, Xhignesse M et al. Antibiotics in primary prevention of myocardial infarction among elderly patients with hypertension. Am Heart J 2003;145(5):E20 (given as E20 on PubMed but as E25 in the paper).
Web Table A. Results of recalcuation of P values in the interval 0.04≤P<0.05 in abstracts of randomised clinical trials.
POSTED AS SUPPLIED BY AUTHOR
Reference |
Data in the abstract |
Authors’ method |
My comments |
Tsang (w1) |
P<0.05, RR=2.0 |
Chi-square. |
Wrong. 2/135 vs 0/133. The main text notes RR=2.0 (1.77-2.26) (which gives P<0.0001). With Fisher, I got P=0.50, and with Review Manager, that adds 0.5 to the cells when any cell is empty, RR=4.93 (0.24-102). No reply from author. |
Thies (w4) |
OR=0.52 (0.24-0.89) |
Chi-square. |
Wrong. Rates in per cent for completers; these correspond to 8/53 vs 17/57. Authors report P=0.0273. With chi-square I got P=0.065 without and P=0.11 with continuity correction. OR also wrong, and CI too narrow. I got OR=0.42 (0.16-1.07) or 0.44 (0.18-1.06), depending on method, and even if all patients had been included, it would be wrong. No reply from author. |
Sato (w5) |
P=0421 [sic], OR=6.1 |
Fisher. |
Wrong. 1/109 vs 6/108. The main text notes P=0.0421, I got P=0.065 with Fisher. Author’s comment: acknowledges the error. |
Gaster (w6) |
P=0.04, OR=2.5 |
Fisher. |
Wrong. Rates in per cent; these correspond to 42/54 vs 32/54. I got P=0.061 with Fisher. Author’s comment: confirmed rates but forgot to mention under Methods that the P-value resulted from a logistic regression analysis. |
Sheu (w7) |
P<0.05, RR=0.128 (0.016-0.979) |
Chi-square, with Pearson’s correction, two-sided. |
Doubtful. 2/44 vs 8/44 gives P=0.093 with adjusted chi-square. The authors excluded one of the two cases which gives P<0.05, but this post hoc exclusion seems unwarranted. No reply from author. |
Arshad (w8) |
OR=0.26 (0.07-0.96) |
Chi-square or Fisher; also logistic regression. |
Doubtful. 8/58 vs 17/62, authors reported P=0.08. They then adjusted for 10 factors and got P=0.04. Not clear why these 10 factors were chosen, e.g. one of them occurred for 13 vs 15 patients, although imbalance was given as the reason for the adjustments. Author’s comment: confounders also included. |
Hughes (w9) |
P=0.04, OR=3.2 |
Statistical methods not described in Methods, noted as "regression analyses" in Results. |
Doubtful. 61 vs 54 patients. There was no difference in the unadjusted analysis the authors reported, OR 1.4 (0.6-3.5), P=0.45. Number of factors used for adjustment described as both 4 and 6. Two factors were chosen because of significant baseline difference, and 2 others because they "appeared to differ". Author’s comment: the correct number is 6; controlling for additonal factors did not change the OR much. |
León (w10) |
P<0.049, RR=4.14 (0.8-19) |
Chi-square or Fisher. |
Doubtful. 2/116 vs 8/114, I got P=0.058 with Fisher. The authors used one-sided alpha in the sample size calculation, and I got P=0.048 with one-sided Fisher. No reply from author. |
Trivedi (w11) |
P=0.04, RR=0.78 (0.61-0.99) |
"crude rates and then, after adjustment for age, with the Cox regression method". |
Doubtful. As there were 2686 patients, and the mean age differed by only 0.1 years, there was no need to adjust for age. Without adjustment, I got P=0.058. Author’s comment: adjusted because age is a strong risk factor. My comment: although this may have been perfectly reasonable in this particular case, the more general issue is that it is not likely that authors are equally inclined to adjust for a very small age difference in a large sample if the adjustment changes the P value in the opposite direction, from 0.04 to 0.058. |
George (w12) |
P<0.05, OR=4.64 (1.02-21.00) |
Pearson chi-square. |
Could be discussed. 9/20 vs 3/20. The authors did not correct for small numbers and found a chi-square of 4.29 (which gives P=0.038). I got P=0.085 with continuity correction and P=0.082 with Fisher. Author’s comment: agrees that they should have used an adjusted chi-square or Fisher. |
Mion (w13) |
OR=0.21 (0.05-0.99) |
Chi-square likelihood ratio tests, two-sided. |
Could be discussed. 2/326 vs 9/324. The authors reported in a table that they used chi-square or Fisher as appropriate. I got P=0.037 with Fisher, but an odds ratio that overlapped 1: OR=0.22 (0.05-1.01). Author’s comment: by mistake, we did not use intention-to-treat for this analysis which explains the slight difference in odds ratios, but Fisher was significant in both cases. |
Basnyat (w14) |
20/81 vs 9/74, "statistically significant reduction" of 50.6% in RR |
Fisher or likelihood ratio tests. |
Could be discussed. The authors reported P=0.043, I got P=0.063 with Fisher, but also RR=2.03 (1.01-4.08) in support of the authors’ abstract. No reply from author. |
Bishai (w2) |
P<0.05, OR=2.66 |
Logistic regression adjusted for one factor. |
Could be discussed. 25/38 vs 13/31. The main text notes OR=2.66 (0.90-7.95) (which gives P=0.078). Author’s comment: used two methods; the CI for OR should have been 1.000544 to 7.086235 (computer print-out with correct event rates). |
RR: relative risk; 95% confidence interval (CI) in brackets; Fisher: Fisher’s exact test; Author’s comment: personal communication.