Portal de Periódicos da CAPES

A nomogram to predict the probability of mortality after first-ever acute manifestations of cerebral small vessel disease

2017; Elsevier BV; Volume: 385; Linguagem: Inglês

10.1016/j.jns.2017.12.020

1878-5883

Manuel Cappellari, Cecilia Zivelonghi, Gianni Turcato, Stefano Forlivesi, Nicola Micheletti, Giampaolo Tomelleri, Paolo Bovi, Bruno Bonetti,

Neurosurgical Procedures and Complications

Background and purpose Symptomatic lacunar stroke (LS) and deep intracerebral hemorrhage (dICH) represent the acute manifestations of type 1 cerebral small vessel disease (cSVD). Recently, two studies showed that the risk factor profile of dICH differs from that associated with LS in subjects with biologically plausible cSVD; however, the prognostic predictors after acute manifestations are currently lacking. We aimed to develop a nomogram for individualized prediction of the mortality probability in a cohort of patients with a first-ever acute manifestation of biologically plausible cSVD. Methods We conducted a retrospective analysis of data collected from consecutive patients with acute symptomatic non-embolic LS or primary dICH. The outcome measure was 3-month mortality. Based on multivariate logistic model, the nomogram was generated. Results Of the 288 patients who entered into the study for biologically plausible cSVD, 131 (45%) experienced a LS and 157 (55%) a dICH. After multivariate logistic regression, 5 variables remained predictors of mortality to compose the nomogram: dICH (OR:11.36; p = 0.001), severe presentation (OR:8.08; p < 0.001), age (OR:1.08; p = 0.001), glucose (OR:1.23; p = 0.003) and creatinine (OR:1.01; p = 0.024) at admission were predictors of mortality. The discriminative performance of nomogram assessed by using the area under the receiver operating characteristic curve (AUC-ROC) was 0.898. The model was internally validated by using bootstrap (1000 samples) with AUC-ROC of 0.895 and cross-validation (deleted-d method repeated 1000 times) with AUC-ROC of 0.895. Conclusions We developed the first nomogram for prediction of the mortality probability in a cohort of patients with a first-ever acute manifestation of biologically plausible cSVD.