TO THE EDITOR: Ikawa et al.1evaluated the rupture risk in patients with small unruptured cerebral aneurysms (Ikawa F, Morita A, Tominari S, et al. Rupture risk of small unruptured cerebral aneurysms.J Neurosurg.2020;132[1]:69–78). Treatment and rupture numbers of small unruptured cerebral aneurysms (UCAs) were 1132 and 23, respectively. The adjusted hazard ratios (95% confidence intervals [CIs]) of a screening brain checkup, a history of subarachnoid hemorrhage (SAH), uncontrolled hypertension, and location on the anterior communicating artery (ACoA) for rupture were 4.1 (1.2–14.4), 10.8 (2.3–51.1), 5.2 (1.8–15.3), and 5.0 (1.6–15.5), respectively. I have two comments on their study.
First, the authors concluded that the possible selection criteria for treating small UCAs included a history of SAH, uncontrolled hypertension, location on the ACoA, and young patients, which was inconsistent with multivariate analysis. Although the log-rank test showed no significant association of rupture with screening brain checkup, there is no statistical evidence of selecting young patients as a risk factor for rupture.
Second, they handled a small number of events for the multivariate analysis. Peduzzi et al. evaluated the effect of events per independent variable (EPV) in proportional hazards regression analysis.2,3One conclusion of this evaluation was that EPV values < 10 have some problems in maintaining validity of the statistical model. Although there is an opinion that EPV values < 10 are also acceptable for investigating the association,4a prediction model requires a stable number of events. Regarding this point, results presented by Ikawa et al. need further statistical considerations. The number of events was 23, and 2 independent variables can be used for the analysis. They presented relatively wide 95% CI ranges, which might be derived from unstable estimates. In any case, I recommend that EPV should be increased by using a longer follow-up duration or involving much larger sample sizes for risk estimation.
References
-
1 ↑
IkawaF,MoritaA,TominariS,et al.Rupture risk of small unruptured cerebral aneurysms.J Neurosurg.2020;132(1):69–78.
-
2 ↑
ConcatoJ,PeduzziP,HolfordTR,FeinsteinAR.Importance of events per independent variable in proportional hazards analysis. I. Background, goals, and general strategy.J Clin Epidemiol.1995;48(12):1495–1501.
-
3 ↑
PeduzziP,ConcatoJ,FeinsteinAR,HolfordTR.Importance of events per independent variable in proportional hazards regression analysis.II. Accuracy and precision of regression estimates. J Clin Epidemiol.1995;48(12):1503–1510.
-
4 ↑
VittinghoffE,McCullochCE.Relaxing the rule of ten events per variable in logistic and Cox regression.Am J Epidemiol.2007;165(6):710–718.