Case Identification Methods.
- Automated paging system (from study initiation to June 12, 2012): Programed search of ED that is used for all registered patients and to include triage information. Data entered by triage personnel using pull down lists or free-text. Search intervals every 15 minutes of the chief complaint or history of present illness sections for the following terms: dizziness, dizzy, vertigo, spinning, imbalance, ataxia, can’t walk, nystagmus, unsteady gait, abnormal gait, or ataxic gait. Misspellings were also included.
- Automated paging system (from June 13, 2012 forward). Automated paging method needed to be revised because of a new electronic medical record system. Real time initiation of page for chief complaint of “dizziness,” “gait problem,” and “cerebrovascular accident.” There was no option for “vertigo” or other previously used terms. At later time periods, the algorithm to trigger a page was updated to include an order entry for meclizine during the visit or the recording of a dizziness or vestibular diagnosis (International Classification of Disease, 9th Revision: 386.x, 388.5, 388.9, 780.2, 780.4, 994.6) in the electronic medical record.
- Study personnel reviewing the ED list of active patients and also circulating in the ED.
- Advertising the study (via emails, conference presentations, and posted signs) to ED, neurology, and otolaryngology healthcare providers, who were encouraged to contact study personnel for potentially eligible patients in the ED, hospital, or outpatient setting.
Ocular motor Examination
The eyes were observed in primary position (straight-ahead) and then also sequentially with gaze directed approximately 30-40 degrees left, right, up, and down. In each position, the vector (i.e., horizontal, vertical, or torsional) of the fast-beating component and the intensity of nystagmus were recorded. Intensity was scored as a “0” if no nystagmus was observed, a “1” if nystagmus lasted less than 5 seconds, or a “2” to a “4” if nystagmus lasted longer than 5 seconds. Intensity scores ranging from “2” to “4” were based on the velocity of the nystagmus with “2” equal to low velocity, “3” equal to moderate velocity, and “4” equal to high velocity. Bi-directional gaze-evoked nystagmus (GEN) was defined as the presence of right-beat nystagmus when looking to the right and left-beat nystagmus when looking to the left, with the intensity score of ≥2 to one side and ≥1 and to the other side. Horizontal nystagmus with an intensity of ≤1 on both right and left gaze was not considered GEN because brief nystagmus of this type is consistent with physiologic end-gaze nystagmus. Direction-fixed horizontal nystagmus was defined as horizontal nystagmus that did not change direction on gaze testing (only right beat, or only left beat) and had an intensity of ≥ 2 in at least one position.
The assessment of skew deviation was performed by the alternate cover test while the subject fixated on a straight-ahead target.
Head-Impulse Test (HIT)
The HIT was used to assess the vestibular-ocular reflex (VOR) which is expected to be abnormal with a lesion of the vestibular nerve (i.e., a peripheral disorder) and normal with a lesion of the central nervous system (i.e., a central disorder). The HIT was performed by first having the patient focus on a target about 18 inches in front of vision. Then typically 5-10 small amplitude, high velocity movements in the horizontal plane were performed in each direction. To score the HIT result for each side, the investigator observed for a re-fixation (i.e., corrective) saccade after each head movement. If no corrective saccades were observed, then the test was assigned a score of “0.” If corrective saccades were observed then the result was assigned a score of “1” if re-fixations were not consistent, a “2” if the re-fixation saccades were consistent and mildly abnormal (i.e., re-fixations were only of small amplitude), a “3” if the re-fixation saccades were consistent and moderately abnormal (i.e., medium amplitude re-fixations), and a “4” if re-fixation saccades were consistent and severely abnormal (i.e., large amplitude re-fixations).
The MRI protocol for clinical and research studies followed radiology department protocols. The clinical protocol included the following imaging sequences: T1-weighted (pre- and post-gadolinium), T2-weighted, fluid-attenuated inversion recovery (FLAIR), and diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC) maps. The slice thickness was 5 mm with a 1 mm gap for all sequences, except for the DWI/ADC sequence which was either 5mm with a 1 mm gap or 6mm with no gap. The research protocol MRIs used a slick thickness of 4 mm with a 1mm gap for DWI/ADC, and 5mm with a 1mm gap for all other sequences, which included T1-weighted, T2-weighted and T2 FLAIR sequences without contrast. An abbreviated research MRI protocol was used if the subject already had a clinical MRI negative for a cerebrovascular cause and also when subjects could not tolerate or declined the full MRI protocol. The abbreviated protocol included fluid-attenuated inversion recovery (FLAIR) images, diffusion-weighted images, and apparent diffusion coefficient (ADC) maps. Patients with claustrophobia were offered a benzodiazepine prior to the MRI. MRIs performed for clinical purposes were interpreted by a staff neuro-radiologist whereas MRIs performed for research purposes were interpreted by the study’s neuro-radiologist (EGH). The study MRI outcome was abstracted from the MRI report.
Some factors could have impacted the patients’ examination or our interpretation of it. MRI was performed prior to the clinical assessment in 15% of the subjects. Although investigators were blinded to the MRI results, it is still possible that subtle clues were conveyed. The ocular motor findings could have been affected by medication treatment that occurred prior to the study examination in 42% of subjects, though this may be similar to the circumstances of routine care. Fifteen percent of the enrolled subjects did not receive an MRI within 14 days and therefore were not included in the main analysis. The subjects not receiving MRI were similar to the subjects who did (particularly similar to the non-stroke group) and the results of the sensitivity analysis imputing the MRI result did not substantially differ from those of the main analysis, suggesting low risk for bias from this missing data.