Sedation Risk Assessment
Reviewed by Cheri Landers, MD
Predicting outcomes of a sedation encounter based on a patient’s pre-sedation clinical state has been the work of the SPS Emergent Sedation Committee (now a sub-committee of the SPS Quality and Safety Committee) for some time. The goal is to develop a tool to predict an accurate sedation outcome to allow for more safely stratifying patients to the appropriate care provider and provide a more efficient service line through early identification of patients at risk.
Dr. Mark Buckmaster began this session by summarizing the challenges behind such a task. Poor sedation outcomes are less distinct than those measured by other prediction models in medicine such as the APACHE or PRISM scores in adult and pediatric ICUs which predict the likelihood of survival vs. death. Death or cardiac arrest are so rare during sedation as to be non-existent in most sedation services therefore they cannot be used as an outcome measure. Even the other relatively more severe outcomes from procedural sedation (e.g. emergency anesthesia consult, aspiration, increased level of care) are exceedingly rare occurrences in the PSRC database making statistical associations difficult.
Prior evaluations of the PSRC database have identified single patient characteristics that are associated with increased incidence of events (age, weight [obesity], ASA status, certain comorbid conditions and the types of procedure to be done). However, what cannot be quantified in the PSRC database are those patients that were excluded based on patient characteristics the sedation center felt necessitated the child be sedated by anesthesia rather than the sedation provider. Within these limitations, using multiple regression and the PSRC database, the goal has been to build a predictive model for adverse events/complications surrounding pediatric sedation. (Dr. Buckmaster made reference to a Star Trek episode during his segment of the session, but not being a Trekkie, I would not even come close to doing it justice; so out of respect for those that are experts in the field, I will stay silent on the matter.)
Dr. Kevin Couloures described in more detail the work to date on the specific risk score. Records from the PSRC database from 2007 to 2011 were reviewed upon which multivariate regression analysis for association with any of 15 selected adverse events was performed. Adverse events were broken down into three severity categories. Variables that were found to be associated with the adverse outcomes were: Patient age, ASA status, primary diagnosis of cardiac disease, co-existing diagnosis of respiratory cough or asthma, co-morbidity or mental retardation/cerebral palsy, co-morbidity of Down Syndrome, type of procedure to be performed and patient weight (both underweight and overweight). The RAPS (Risk Associated with Pediatric Sedation score) was developed based on this analysis. Points were assigned to each of the variables based on the weight of contribution to the adverse event outcome (e.g. age 0-6 months scored one point, comorbidity cough scored seven points, LP/BMA scored 0 points and bronchoscopy scored 18 points). This prediction model was compared against the current version of the database for validation. At a score of five, the specificity of the model was only 26% but this increased at a score of 10 to 92%.
The proposed utilization of the RAPS score is not to have a set cutoff for all institutions to refer patients to anesthesia, but for each institution to adjust their scoring based on their practice setting. A smaller hospital with limited resources may use a lower score of five or six whereas a robust sedation service within a larger children’s hospital may find score of eight or nine more appropriate. Additionally, different scores may be used based on the procedure to be performed. The next step is to use the RAPS score prospectively in a small number of centers prior to a patient being triaged to either anesthesia or sedation and track adverse event rates compared to predicted rates.
Dr. Pradip Kamat then reviewed what we have learned from research in procedural sedation. There are many limits to the data available to us in this field, most sedation studies are retrospective, single center, and low powered. Large prospective randomized blinded trials are difficult to perform, require multiple institutions, a large number of patients, consistent definitions of adverse events and are labor intensive and expensive.
The PSRC database has contributed to research in four areas: 1) sedation provider/adverse events, 2) sedation plan/medications, 3) location, procedure and monitoring and 4) patient risk factors. Major conclusions from research to date cited by Dr. Kamat included the low incidence of serious adverse events (SAEs) in sedations outside the OR for individual medications as well as when looking at sedation provider specialty.
Factors that may be associated with increased risk of SAEs are small size, both the very young and the older than 12 year age groups, prematurity (contributes to sedation risk throughout the patient’s lifespan), and sedation location. Individual medication studies utilizing the information in the PSRC database have evaluated nitrous oxide, dexmeditomidine, ketamine, and propofol. Although all have a low incidence of SAEs in the PSRC database, there have been suggestions of ways to minimize the more common side effects with ketamine. The article in Anesthesiology in 2016 demonstrating that aspiration was uncommon when children violated the ASA npo guidelines has led to interesting discussions about the necessity to adhere to same. The authors are quick to point out that the severity of the npo violation was not able to be determined, and that a part of a goldfish cracker eaten five hours prior to an LP is very different than the consumption of biscuits and gravy, bacon and eggs immediately prior to an EGD.
Despite the information we have obtained from the volumes of sedation encounters in the PSRC, it has its limitations. The balance between the desire to obtain every piece of patient data and the time to enter data requires limiting the information gathered which means data on dosing, sedation depth, hemodynamics, resource utilization, and pre/post neurocognitive development data is not collected. However, what the PSRC has shown us is the type and frequency of adverse events for the contributing centers, that pediatric subspecialists can deliver deep sedation that meets standards advocated by anesthesiologists and the importance of appropriate training in the recognition and management of airway and ventilation events.
The PSRC remains both a source of quality data for contributing centers as well as an investigational database. It is contributing to the exciting work that continues towards further validation of a pediatric sedation risk score (the RAPS) to assist both established sedation programs as well as less experienced sedation providers in determining the best sedation environment for their patients.