Validation and Recalibration of Two Multivariable Prognostic Models for Survival and Independence in Acute Stroke

Abstract

IntroductionVarious prognostic models have been developed for acute stroke, including one based on age and five binary variables (‘six simple variables’ model; SSVMod) and one based on age plus scores on the National Institutes of Health Stroke Scale (NIHSSMod). The aims of this study were to externally validate and recalibrate these models, and to compare their predictive ability in relation to both survival and independence.MethodsData from a large clinical trial of oxygen therapy (n = 8003) were used to determine the discrimination and calibration of the models, using C-statistics, calibration plots, and Hosmer-Lemeshow statistics. Methods of recalibration in the large and logistic recalibration were used to update the models.ResultsFor discrimination, both models functioned better for survival (C-statistics between .802 and .837) than for independence (C-statistics between .725 and .735). Both models showed slight shortcomings with regard to calibration, over-predicting survival and under-predicting independence; the NIHSSMod performed slightly better than the SSVMod. For the most part, there were only minor differences between ischaemic and haemorrhagic strokes. Logistic recalibration successfully updated the models for a clinical trial population.ConclusionsBoth prognostic models performed well overall in a clinical trial population. The choice between them is probably better based on clinical and practical considerations than on statistical considerations.

Risk‐adjusted hospital mortality rates for stroke: evidence from the Australian Stroke Clinical Registry (AuSCR)

“…The ability to collect NIHSS data was introduced in AuSCR in 2015, but the level of missing data currently undermines its usefulness, whereas “ability to walk on admission” information was available for 90% of episodes. In recent validation work, 9 a model based on simple variables (including ability to walk) performed as well as one employing NIHSS and age data; the choice of measure should therefore be based on practical considerations 24 . Because there were very few episodes of intracerebral haemorrhage, we included stroke type as a covariate rather than stratifying the dataset, as has been previously recommended by other authors 25…”

“…In recent validation work, 9 a model based on simple variables (including ability to walk) performed as well as one employing NIHSS and age data; the choice of measure should therefore be based on practical considerations. 24 Because there were very few episodes of intracerebral haemorrhage, we included stroke type as a covariate rather than stratifying the dataset, as has been previously recommended by other authors. 25 In conclusion, we highlight the importance of using appropriate risk adjustment variables and methods for comparing hospital outcomes for stroke, with particular emphasis on the need to account for stroke severity.…”

Quality of Acute Care and Long-Term Quality of Life and Survival

“…However, we are reassured by recent validation work that simple variables such as ability to walk perform similarly well to prediction models that use National Institutes of Health Stroke Scale and age. 26 We were also unable to account for the types of postdischarge care (eg, community rehabilitation) which may contribute to HRQoL outcomes, as well as long-term adherence to secondary prevention medication. Models include only the first admission registered in the Australian Stroke Clinical Registry adjusted for age, sex, socioeconomic position, country of birth, type of stroke, history of previous stroke, ability to walk on admission, and in-hospital stroke.…”

Stroke Recovery Genetics

“…7,8 But predictive models based on clinical factors remain limited by imprecision and difficulty with translation to the individual case. 9 This may improve when mechanisms such as brain plasticity 10 and brain stunning 11 and factors that influence these concepts become better defined.…”