Being able to reliably predict the long-term outcome for a stroke patient during the early post-stroke period would allow for the planning and tailoring of the person’s care needs, help avoid unrealistic expectations of recovery by the patient and family and build confidence in the rehabilitation process.
To establish what is currently known about prognosis after stroke, a group of researchers in the Netherlands (van Almenkerk et al., 2013) carried out a systematic review of studies where stroke patients had been assessed within one month of stroke onset and followed up at least three months later. From the data gathered they then assessed whether there was strong evidence for predictors of outcome.
Unlike previous studies which have evaluated predictors of good outcome they reviewed the evidence for predictors of poor outcome, which they defined as the person being in a long-term care setting or as having a severe disability as measured on standard scales.
The group conducted a thorough search in five languages, and having applied strict inclusion criteria to the 3,971 titles they found, they identified a total of 33 studies that described 27 cohorts to include in their review.
These studies all included multivariable regression analyses which were used to identify independent prognostic factors, with effects given by point estimates and confidence intervals.
A real strength of this review was the inclusion of a measure of the consistency of findings which the authors evaluated by calculating the ratio of the number of studies that showed a factor to be predictive to the number of studies where it was non-predictive.
Having assessed twenty different factors what emerged was some obvious predictors, some intriguing ones, and a lot of potential predictors which have not been adequately researched.
- In the obvious category were greater age, a more severe stroke, a larger stroke volume and the presence of urinary incontinence in the first month. These were all strong predictors of a poor outcome three or more months later.
- The data for impaired cognitive abilities predicting outcome were not consistent enough to make a reliable assessment of the strength of prediction, and the same was true for the data on the degree of dependency in basic activities of daily living. In relation to these factors the authors noted that the variation across studies could be attributed to the use of different cognitive tests and the timing of assessments for activities of daily living.
- It was worrying to note that only two studies had included measures of depression, and there were no studies where early communication difficulties were assessed as predictors of longer-term outcome.
- There were sufficient data to assess three of the factors known to influence stroke incidence though (a history of stroke, heart disease and hypertension), and interestingly these did not predict poor outcome.
The researchers concluded that at present our ability to predict poor outcomes for stroke survivors is still minimal; in fact they suggested that the physician’s intuition about their patient’s prognosis is likely to be as good as using the predictors they identified.
They pointed to number of issues that still need to be addressed in this area, but the bottom line is that the hope of an evidence based tool for reliably predicting poor outcome after stroke is still a long way off.
The Stroke Elf view
This was a well conducted review that ultimately showed us that we know little about what factors will predict poor outcome after stroke.
The authors made some good suggestions for future work, but as they listed the difficulties that would be encountered in developing a reliable clinical prediction tool, it became clear that extensive work is still needed.
Whether there will be sufficient research funding in the future to achieve this is debatable.
van Almenkerk S, Smalbrugge M, Depla MF, Eefsting JA, Hertogh CM. What predicts a poor outcome in older stroke survivors? A systematic review of the literature. Disabil Rehabil. 2013 Oct;35(21):1774-82. doi: 10.3109/09638288.2012.756941. Epub 2013 Jan 25. [PubMed abstract]