Interobserver reliability in the interpretation of computed tomographic scans of stroke patients.

Interobserver reliability in interpretation of computed tomographic images was studied by six senior neurologists who independently evaluated on a standardized Stroke Data Bank form the brain lesions of 17 patients. The results analyzed with kappa statistics yielded moderate to substantial agreement on most items of interest including the stroke pathology and anatomy. In general, the levels of agreement were as high as previously reported for the diagnosis of the mechanism of the stroke, and much higher than on many stroke history items and items of neurologic examination. Excellent agreement was obtained for the detection of infarcts and intracerebral hemorrhage, and substantial agreement was obtained on whether the computed tomographic images were normal or indicative of small deep infarcts, superficial and deep infarcts, and aneurysms. The level of agreement on anatomy of the lesions was best for the frontal, parietal, and temporal lobes, putamen, cerebellum, and subarachnoid space. Implications for clinical research and diagnosis are discussed.

Interobserver variability in the assessment of neurologic history and examination in the Stroke Data Bank.

Interobserver reliability in obtaining neurologic histories and examinations was investigated among neurologists collaborating in the Stroke Data Bank (SDB). Seventeen in-hospital stroke patients were examined by six neurologists experienced in stroke over the course of three days. Patients were examined twice a day for two successive days, with each patient seen by four different neurologists. Data were recorded on SDB forms, according to definitions and procedures established for the SDB. Percent agreement and kappa coefficients were calculated to assess the levels of agreement for each item. Important differences in levels of agreement were found among items on both neurologic history and examination. Agreement among neurologists was higher for neurologic examination than for history. Patterns of agreement for items with low prevalence or with numerous unknown ratings are discussed. Improvement in interobserver agreement due to data editing for intra-observer consistency was shown.

Interobserver agreement in the diagnosis of stroke type.

Interobserver agreement is essential to the reliability of clinical data from cooperative studies and provides the foundation for applying research results to clinical practice. In the Stroke Data Bank, a large cooperative study of stroke, we sought to establish the reliability of a key aspect of stroke diagnosis: the mechanism of stroke. Seventeen patients were evaluated by six neurologists. Interobserver agreement was measured when diagnosis was based on patient history and neurologic examination only, as well as when it was based on results of a completed workup, including a computed tomographic scan. Initial clinical impressions, based solely on history and one neurologic examination, were fairly reliable in establishing the mechanism of stroke (ie, distinguishing among infarcts, subarachnoid hemorrhages, and parenchymatous hemorrhages). Classification into one of nine stroke subtypes was substantially reliable when diagnoses were based on a completed workup. Compared with previous findings for the same physicians and patients, the diagnosis of stroke type was generally more reliable than individual signs and symptoms. These results suggest that multicentered studies can rely on the independent diagnostic choices of several physicians when common definitions are employed and data from a completed workup are available. Furthermore, reliability may be less for individual measurements such as signs or symptoms than for more-complex judgments such as diagnoses.

Reliability of the activities of daily living scale and its use in telephone interview.

Activities of daily living (ADL), a popular outcome assessment tool for studies of disabled populations, was evaluated for use in multicentered studies. The interobserver reliability of a modified Barthel Index of the ADL was evaluated on 18 stroke patients. Reliabilities between administrators of the ADL and between observers of the ADL administrations were r = .99 for total scores and r greater than .90 for most of the individual component items. The validity of ADL evaluation based on telephone interviews was demonstrated relative to ADL scores obtained on 72 patients in a direct test of their performance capabilities. The correlations between the performance-based ADL and the interview-based ADL were r greater than .97 for the total score and r greater than .85 for most of the individual items. The modified Barthel Index of the ADL is a reliable measure for assessing stable stroke patients, supporting its use in multicenter studies. In addition, the high validity observed in the telephone interview supports its use in longitudinal studies and large surveys where direct performance evaluation is not feasible or too costly.