Evidence Based Best Care Practice for Benign Paroxysmal Positional Vertigo

The Dix-Hallpike Test (DHT) and Canalith Repositioning Maneuver (CRM) are used to diagnose and treat Benign Paroxysmal Positional Vertigo (BPPV). BPPV processes have an evidence base that is at the clinical practice guideline level. The DHT is the gold standard test for DHT and the CRM is supported by numerous randomized controlled trials and systematic reviews.

The problem is that BPPV processes are substantially underutilized.

This study looks to increase the use of BPPV processes within a community ED setting by implementing a theory-based educational intervention. The unit of randomization and target of intervention is the hospital. After an initial observation period, the six community hospitals will undergo the intervention in five waves (the smallest two hospitals will be paired). The order will be randomized. This is a partnered research project with local physicians engaging in best practice implementation.

Sample Size

The trial will start with an initial no intervention period of approximately 4 months followed by randomized staggered intervention with a new hospital entering approximately every 2 months, finalized by approximately 4 post-intervention months will result in the approximately balanced number of 867 visits occurring without intervention and 933 visits occurring under (post) intervention. This calculation assumes the average anticipated total patient visit rate of 100 patients per month. Based on our pilot studies and the literature we expect the DHT or CRM procedure to be done in 5% patients before the intervention. With the expected number of visits calculated above, we will be able to detect the increased DHT or CRM rate of 9% and above with 90% power by a two-sided test at the significance level of 5%. We expect a much bigger difference of 5% vs. 20% pre- vs. post-intervention DHT/CRM rates, respectively. Under this expected difference, we will have the power exceeding 99%. In fact, an order of magnitude smaller visit rate of 10.4 patients per month would be sufficient for 90% power under the anticipated difference. The reserves of power will be used to provide more power to fine-tune the multivariate mixed regression models and associated secondary analyses.

Data Analyses

The intervention is delivered to hospitals. Intervention is a binary variable with two levels, pre-intervention (no intervention), post-intervention (under intervention).

The primary analysis will use binary logistic regression and will include covariates for hospital, month (to handle secular trends), and intervention (see below). For a set of new patient visits, the binary random variable DHT/CRM/referral yes/no will serve as the primary response. Patient visits will be supplied with patient-, hospital- and provider-level covariates as well as the calendar time variable modeling the secular trend, and the intervention yes/no variable measuring whether the visit occurs under intervention or not. To take hospital- and provider-specific unmeasured features into account, hospital and provider categorical variables will be included in the analysis. Due to the fact that the number of hospitals and providers is much smaller than the number of patient visits, adjusting for hospital and provider effects by way of categorical variables will not lead to bias. Secondary analyses will explore alternative approaches using random effects (Gaussian) models. A two-sided model-based test for the intervention variable will be used to test the primary hypothesis at the significance level of 5%.

The secondary analysis (safety) will numerically summarize the 90-day stroke rate - cumulatively and stratified for stroke diagnosed on the index dizziness visits and for post index visit strokes (delayed diagnosis) in patients seen at EDs with and without the intervention. This is anticipated to be very rare. The intervention does not target improving stroke diagnosis. However, evaluating both the index visit stroke diagnosis rate and the delayed diagnosis rate should allow for determination of major changes. We anticipate the index visit stroke diagnosis rate to be approximately 2% and the delayed diagnosis rate approximately to be 1%.