Analysis of Adverse Events Associated With Adult Moderate Procedural Sedation Outside the Operating Room.

ORIGINAL ARTICLE

Analysis of Adverse Events Associated With Adult Moderate Procedural Sedation Outside the Operating Room Sergey Karamnov, MD,* Natalia Sarkisian, PhD,† Rebecca Grammer, DMD,* Wendy L. Gross, MD, MHCM,* and Richard D. Urman, MD, MBA*

Introduction: Moderate sedation outside the operating room is performed for a variety of medical and surgical procedures. It involves the administration of different drug combinations by nonanesthesia professionals. Few data exist on risk stratification and patient outcomes in the adult population. Current literature suggests that sedation can be associated with significant adverse outcomes. Objectives: The aims of this study were to evaluate the nature of adverse events associated with moderate sedation and to examine their relation to patient characteristics and outcomes. Methods: In this retrospective review, 52 cases with moderate sedation safety incidents were identified out of approximately 143,000 cases during an 8-year period at a tertiary care medical center.We describe types of adverse events and the severity of associated harm. We used bivariate and multivariate analyses to examine the links between event types and both patient and procedure characteristics. Results: The most common adverse event and unplanned intervention were oversedation leading to apnea (57.7% of cases) and the use of reversal agents (55.8%), respectively. Oversedation, hypoxemia, reversal agent use, and prolonged bag-mask ventilation were most common in cardiology (84.6%, 53.9%, 84.6%, and 38.5% of cases, respectively) and gastroenterology (87.5%, 75%, 87.5%, and 50%) suites. Miscommunication was reported most frequently in the emergency department (83.3%) and on the inpatient floor (69.2%). Higher body mass index was associated with increased rates of hypoxemia and intubation but lower rates of hypotension. Advanced age boosted the rates of oversedation, hypoxemia, and reversal agent use. Women were more likely than men to experience oversedation, hypotension, prolonged bag-mask ventilation, and reversal agent use. Patient harm was associated with age, body mass index, comorbidities, female sex, and procedures in the gastroenterology suite. Conclusions: Providers should take into account patient characteristics and procedure types when assessing the risks of harmful sedation-related complications. Key Words: procedural sedation, moderate sedation, sedation outcomes, sedation complications, analgesia adverse effects, nonanesthesia providers (J Patient Saf 2014;00: 00–00)

M

oderate procedural sedation is administered hundreds of thousands of times per year in the United States1 by nonanesthesia providers. It is used for a wide variety of interventional procedures in multiple settings outside the operating room (OOR). The increasing numbers and broadened scope of interventional procedures performed OOR have increased the demand

From the *Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts; and †Department of Sociology, Boston College, Chestnut Hill, Massachusetts. Correspondence: Richard D. Urman, MD, MBA, Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis St, Boston, MA 02115 (e‐mail: [email protected]). The authors disclose no conflict of interest. Copyright © 2014 by Lippincott Williams & Wilkins

J Patient Saf • Volume 00, Number 00, Month 2014

for competent and safe nonanesthesia providers. Sedation can be administered by a diverse group of health care professionals such as non–anesthesia-trained physicians, nurses, and physician assistants.2–7 Modern pharmacology, monitoring technology, and a clear understanding of potential complications allow providers without the special anesthesia-based training to safely deliver sedation. Strict regulation of medications being administered and monitoring used to assess patient stability and intended depth of sedation are necessary to ensure patient safety during these procedures. The American Society of Anesthesiologists (ASA) recognizes several levels of sedation depth based on patient responsiveness to various stimuli. Moderate sedation is defined as “drug-induced depression of consciousness during which patients respond purposefully to verbal commands, either alone or accompanied by light tactile stimulation. No interventions are required to maintain a patent airway, and spontaneous ventilation is adequate. Cardiovascular function is usually maintained.” 8,9 A large volume of multi-institutional data using outcomes databases have been reported in the pediatric population as part of the Pediatric Sedation Research Consortium, with the most commonly reported complications being related to the respiratory system. However, few studies address risk stratification and sedation outcomes during procedural sedation in an adult population.10–16 A recent study comprising 49,839 cases of propofol sedation found that the most commonly reported complications were O2 desaturation lower than 90% for more than 30 seconds as well as central apnea/airway obstruction, excessive secretions, stridor, and laryngospasm. There were no deaths reported and a very small number of aspirations or cases requiring cardiopulmonary resuscitation.15 The ASA Closed Claims database has provided additional insight into the nature of adverse events associated with anesthesia and sedation OOR. Bhananker et al17 examined all surgical anesthesia claims associated with monitored anesthesia care (MAC) and compared them with those associated with general and regional anesthesia. Close to half of these claims were classified as preventable with better monitoring such as capnography and audible alarms. The authors found that oversedation resulting in respiratory depression was an important factor in patient injuries during MAC. On the basis of the ASA Closed Claims data, Metzner et al18 found that injurious respiratory events were significantly more common in remote location claims (44% versus 20%), with inadequate oxygenation/ventilation as the most common specific adverse event (21% versus 3% in operating room claims). Events in remote location claims were more often judged as being preventable by better monitoring. Pino13 examined 63,000 adult anesthesia and sedation cases performed OOR at a tertiary care academic institution, with 41% of the cases having been performed by nonanesthesiologists. On the basis of the data collected, significant delays in the detection of apnea were notable in cases in which capnography monitoring www.journalpatientsafety.com

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1


Karamnov et al

was not used. Surprisingly, sedation using only opiates and benzodiazepines resulted in a higher incidence of respiratory compromise compared with sedation using propofol. This study advocated for a quality assurance system to track events associated with procedural sedation and anesthesia done OOR as a way to maintain and improve quality and safety. In light of the recent anesthesia and sedation outcomes findings, the ASA amended its Standards for Basic Anesthetic Monitoring in 2011 to require monitoring of exhaled carbon dioxide during all moderate sedation cases.19 There is existing literature examining the safety of sedation during gastrointestinal (GI) endoscopic,20–22 dental,23,24 and emergency department (ED) procedures.25 However, high-quality data on the incidence of complications associated with sedation OOR are lacking. Specifically, there are no data comparing practice outcomes across different practitioners and specialties.26,27 In this study, we analyze risks of adverse events related to moderate sedation for various procedures conducted outside the operating room. Our goal is to evaluate adverse event reports associated with moderate procedural sedation in adults during an 8-year period from 2005 to 2013 in a tertiary academic center and analyzed outcome severity, patient characteristics, and procedure characteristics most frequently associated with reported safety incidents. We examined the nature of such incidents representing a medically diverse group of patients, procedures, and specialties at a multidisciplinary institution.

METHODS With an institutional review board approval, data were collected from the quality hospital database based at the Brigham and Women’s Hospital, a 793-bed academic tertiary care hospital. The institutional procedural sedation policy at our hospital allows only for benzodiazepines and opioids to be used for moderate sedation, with the most common drug combination being midazolam and fentanyl. Fentanyl dosing cannot exceed 50 μg/min, up to a maximum amount of 500 μg/h. Propofol is permitted for use only by emergency medicine physicians and anesthesiologists for the purpose of providing deep sedation or general anesthesia. Therefore, cases involving planned use of propofol were excluded from our study. Physicians who participate in moderate sedation are required to possess current Advanced Cardiac Life Support (ACLS) certification and complete a biyearly online procedural sedation course and a multiple choice test. The nurses, although not required to have ACLS certification, are required to complete a similar online sedation course and demonstrate adequate airway skills on a yearly basis. The hospital does not have dedicated sedation teams, although most physicians and nurses tend to work in the same clinical area. During evening and overnight hours, there is a pool of sedation nurses available to be dispatched to a clinical area where moderate sedation is needed. Procedural sedation safety incident reports during an 8-year period from 2005 to 2013 were collected and analyzed from hospital quality database.Of approximately 143,000 cases of moderate procedural sedation performed during that period, 75 cases of adverse events and unplanned interventions were reported. These cases were screened; 52 cases were selected on the basis of their relevance to moderate procedural sedation (the rest were unrelated to sedation or involved deep sedation as the intended level of sedation). Records were reviewed independently by 2 study coordinators. Three sets of variables were examined: patient characteristics, procedural focus, and the nature of adverse events. Patient characteristics included sex; age; body mass index (BMI); indicators of comorbidities including chronic obstructive pulmonary disease, diabetes mellitus, heart disease, hyperlipidemia, hypertension, malignancy, cardiac arrhythmia, gastroesophageal reflux disease

2

www.journalpatientsafety.com

(GERD), peripheral vascular disease, obstructive sleep apnea, and anxiety/depression; the total count of the patient’s comorbidities (ranging from 0 to 7 and topcoded at 4); as well as the ASA class of the patient based on the ASA’s physical status classification system.28 In this population, ASA class variable ranged from 1, healthy, to 4, severe systemic disease that is a constant threat to life. Because very few individuals received scores of 1 or 4, we grouped together ASA class scores 1–2 and 3–4 for the purposes of statistical analysis. Second, procedural focus included a nominal variable indicating which medical service performed the procedure, another nominal variable indicating the location of that procedure, and a dichotomy indicating whether the procedure was performed in an emergency. We also included a variable indicating the calendar year when the procedure was done to assess trends over time. Finally, the specific nature of the adverse event or unplanned intervention included dichotomous indicators of specific adverse events or specific unplanned interventions as well as an overall measure of outcome adversity. To measure outcome adversity, the reports were categorized by severity of harm to the patient, with severity score 0, no harm, error did not reach the patient; 1, no harm, error did reach the patient; 2, temporary or minor harm; and 3, permanent or major harm. Because the vast majority of reports received scores 1 or 2, we grouped together severity scores 0–1 (no harm) and 2–3 (some harm) for the purposes of statistical analysis. Univariate descriptive statistics were calculated for all variables, using percentages for categorical variables and mean (±SD) for continuous variables. Next, bivariate tests of relationships between the features of adverse events and both patient and procedure characteristics were conducted. The Fisher’s exact test was used for most patient and procedure characteristics (those represented by all categorical variables), whereas 2-tailed t tests of differences in means were used for continuous variables (BMI and age). In addition, because the links between BMI or age and types of adverse events could potentially be nonlinear, we also conducted nonparametric bivariate analyses using locally weighted running-mean smoother (lowess). To reduce the impact of outliers in these analyses as well as in logistic regression models, age was bottom-coded at 27, and BMI was bottom-coded at 16.8 and top-coded at 38.8. The final step involved estimating logistic regression models that simultaneously evaluated multiple predictors of the likelihood of occurrence for each type of adverse event or unplanned intervention in our sample of those with at least 1 such event or intervention. Logistic regression model fit was evaluated using McFadden’s R2. Because our sample size was relatively small (N = 52), a post hoc power analysis conducted with the use of G*Power software29 revealed that, for most tests involved, even assuming large effect sizes, statistical power was much lower than the desired value of 0.8 if we set α to 0.05. Therefore, to improve power and reduce the risk for type II error, we used P values of less than 0.10 to identify statistically significant findings. All analyses (other than the power analysis) were performed using STATA 13.0.30

RESULTS Patient Population For the 52 cases (31 female and 21 male patients) in which procedural sedation safety incident reports were filed, the mean patient age was 59.2 years, with range of 18 to 84 years. The mean BMI of the patients was 32. Among patient-related comorbidities, the most common were hypertension (50% of patients), heart disease (29%), and diabetes (25%). The mean number of comorbidities per person was 2.2. Most patients fell into ASA categories 2 (34.6% of patients) and 3 (57.7%). © 2014 Lippincott Williams & Wilkins



Adverse Events of Adult Moderate Sedation

FIGURE 1. Cases associated with adverse events or unplanned interventions. Results are expressed in percentage from the total of 52 reported cases. A, Patient-related adverse events; B, provider-related adverse events; C, unplanned interventions. IVCS, intravenous “conscious” sedation.

Adverse Events and Unplanned Interventions Various adverse events were reported to our database (Fig. 1). We grouped them into 2 categories: those related to the provider’s error (Fig. 1B) and those that could be related to the patient’s characteristics (Fig. 1A). By far, the most common event in the patient group was oversedation of the patient that led to apnea:57.7% of the reported cases were associated with it. The other common adverse events in this group were hypoxemia (32.7%), hypotension (15.4%), and patient discomfort (15.4%). In the provider’s group, miscommunication was the most common event (34.6% of cases). Examples of events in this category were poor communication between different services involved in the procedure (10 cases), disagreement between the physician and the nurse on how to proceed with sedation (5 cases), and miscommunication between providers within the same service (2 cases). Another relatively common event in this group was administration of sedation by a provider who was not certified to deliver moderate sedation (13.5%). © 2014 Lippincott Williams & Wilkins

Adverse events in turn resulted in various unplanned interventions (Fig. 1C), among which the most common were use of reversal agent (55.8% of reported cases), necessity of prolonged bag-mask ventilation (23.1%), and unplanned intensive care unit (ICU) or hospital admission for complications related to sedation (13.5%). With regard to severity score, 13.5% of events were classified as “no harm, error did not reach patient” and another 23.1% of events were classified as “no harm, error did reach patient.” The most common outcome (57.7%) was one in which there was temporary or minor harm to the patient. Permanent or major harm was inflicted in 5.8% of cases.

Links Between Adverse Events and Patient Characteristics The results of our study suggest that several patient-related factors were associated with adverse events and unplanned www.journalpatientsafety.com


3


Karamnov et al

FIGURE 2. Relationship of patient characteristics to adverse events: A, BMI; B, age; C, sex. A and B, Results are reported as mean BMI and mean age ± SE by group defined by the specific event occurrence in the population that experienced as least 1 adverse event or unplanned intervention. Statistical significance is indicated using P values based on 2-tailed t tests. C, Results are expressed in percentage ±SE from total number of adverse events or required unplanned interventions for men and women. Statistical significance is indicated using P values based on Fisher’s exact test.

interventions (Fig. 2). For these analyses, we focused on the events experienced by at least 5% of our sample. There were statistically significant associations between BMI and several adverse events. The patients reported to have hypoxemia during procedures had a mean BMI of 32.6, whereas the patients who did not experience hypoxemia had a mean BMI of 26.1 (P = 0.0004). Interestingly, however, our data also indicated that hypotension happened more often in the patients with lower BMI. The hypotension group had a mean BMI of 24, whereas the mean BMI of the patients who did not experience hypotension during sedation (albeit experienced at least 1 adverse

4


event) was 29.3 (P = 0.04). Despite this benefit of lower hypotension rates for patients with higher BMI, overall, severity of complications was clearly linked to higher BMI. The patients who experienced adverse events in which at least some harm was done to the patient (severity scores of 2 or 3) had a mean BMI of 30, whereas the patients in the “no harm to the patient” group had a mean BMI of 25.6 (P = 0.02). The patients who had to be intubated had a mean BMI of 6.8 kg/m2 higher than those who were not intubated: 34.9 versus 28.1 (P = 0.09). Another patient characteristic linked to adverse events was advanced age: it had statistically significant associations with both © 2014 Lippincott Williams & Wilkins




FIGURE 3. Patient outcomes based on preexisting comorbidities. Results are expressed as percentage ± SE of patients who experienced an adverse event among those with or without a given comorbidity or a specific number of comorbidities. A to C, relation of adverse events to number of comorbidities. D and E, occurrence of hypotension and overall harm to the patient based on preexisting GERD or diabetes mellitus, respectively.

oversedation (P = 0.02) and the use of reversal agents (P = 0.03). The patients who experienced hypoxemia also tended to be older: 64.8 versus 56 (P = 0.072). Overall, the patients with higher severity scores (2–3) were 10 years older on average than those with lower severity scores (0–1): 62.9 ± 13.0 versus 52.8 (P = 0.036). Next, sex analysis revealed some differences between male and female patients’ adverse event profiles (Fig. 2C). The female patients experienced oversedation significantly more often than the male patients (67.7% versus 42.9% of cases) (P = 0.093). The female patients also required bag-mask ventilation for more than 5 minutes (32.3% versus 9.5%, P = 0.093) and reversal agent (67.7% versus 38.1% of cases, P = 0.048) significantly more frequently than the male patients. We analyzed relation of adverse events to preexisting conditions of the patients (Fig. 3). Results were expressed as percentage of patients who experienced an adverse event among those with or without a given comorbidity. We grouped patients on the basis of total number of preexisting comorbidities: 0, 1, 2, 3, and 4 or greater (hereafter referred to as groups 0, 1, 2, 3, and 4). Patient discomfort (Fig. 3A) during the procedure most commonly occurred in groups 2 and 3 (27.3% and 40%, respectively) (P = 0.032). Intubation (Fig. 3B) happened only in group 3 (30% of cases in that group) (P = 0.016). We found a linear relationship © 2014 Lippincott Williams & Wilkins

between percentage of cases with adverse events that resulted in some harm to the patient and the number of the patient’s preexisting conditions: 40% in group 0, 60% in group 1, 81.8% in group 2, and 100% in group 3. Interestingly, however, only 36.4% of cases in group 4 had adverse events with harm score of 2 or 3 (Fig. 3C). We also found that patients with preexisting GERD more often experienced hypotension during the procedure: 42.9% of cases with GERD versus 11.1% of cases without (P = 0.03) (Fig. 3D). Procedures for patients with diabetes mellitus were more frequently associated with adverse events with higher harm score: 56.4% versus 84.6% of cases (P = 0.07) (Fig. 3E).

Procedures Procedures associated with adverse events were performed in various locations by different services (Table 1). We analyzed them by incidence, which is presented here as percentage of the total number of reported cases. Our analysis of adverse events and unplanned interventions based on the hospital location group where procedures took place indicated statistically significant links (Fig. 4). We expressed the results as percentage of total number of reported cases in a specific location. Oversedation (Fig. 4A) and the necessity of reversal www.journalpatientsafety.com


5


Karamnov et al

TABLE 1. Characteristics of Procedures No. Procedures

% of Total (52)

6 6 5 4 3 3 3 2 2 2 2 14

11.5 11.5 9.6 7.7 5.8 5.8 5.8 3.8 3.8 3.8 3.8 26.9

16 8 7 5 5 11

30.8 15.4 13.5 9.6 9.6 21.2

9 6 13 8 13 3

17.3 11.5 25.0 15.4 25.0 5.8

Procedure type Chest tube placement Endoscopy IR-guided line placement Ablation Bronchoscopy Cardiac catheterization ICD placement/revision Joint dislocation reduction Dressing change Pacemaker placement Spinal biopsy Other Service Cardiology Gastroenterology Radiology Emergency Department Thoracic surgery Other Location Radiology Emergency Department Inpatient Floor GI endoscopy Cardiology Other

IR, interventional radiology; ICD, implantable cardioverter defibrillator.

agents (Fig. 4C) had the same incidence and were most commonly reported in gastroenterology (87.5%) and cardiology suites (84.6%) and much less frequently reported on the inpatient floor (30.8% and 23.1%, respectively), in the ED (33.3%), or in radiology suites (44.4%) (P = 0.011 and P = 0.004, respectively). Hypoxemia (Fig. 4B) was most frequently observed in gastrointestinal (GI) endoscopy (75%) and cardiology (53.9) suites (P = 0.011). Prolonged bag-mask ventilation (>5 min) was reported in 50% of GI endoscopy–related cases and 38.5% of cardiology-related cases but in none of ED cases and in only 8% of cases on the inpatient floor (P = 0.087) (Fig. 4D). Miscommunication (Fig. 4E) reported as an adverse event happened most commonly in the ED (83.3%) and on the inpatient floor (69.2%); it was never reported in the gastroenterology suite (P = 0.0004). Providers delivered sedation without moderate sedation certification (Fig. 4F) most frequently on the inpatient floor (38.5% of the reported floor procedures); such events never happened in the ED and gastroenterology, cardiology, or radiology suites (P = 0.022). Additional analyses that focused on the service that performed the procedure (rather than the location where it was performed) demonstrated no differences except for miscommunication (data not shown). Specifically, we found that miscommunication was most common when the service involved was the ED (80% of adverse event cases involving ED were miscommunication cases). Other services with relatively high rates of miscommunication were cardiology and thoracic: 31.3% of cardiology cases and 40% of thoracic cases in our data set were associated with miscommunication events. Miscommunication never happened when the

6


service performing the procedure was radiology and happened in only 12.5% of cases when the service was gastroenterology. Finally, although 9 of the 52 procedures involved an emergency, we did not find any significant differences by emergency status in terms of the likelihood of individual adverse events and unplanned interventions or the overall harm (data not shown). Because the number of emergencies was low, however, we cannot be certain that emergencies were truly similar to nonemergent procedures.

Multivariate Analyses To further explore these links between patient and procedural characteristics and adverse events, we turn to multivariate logistic regression results presented in Table 2. This table presents odds ratios from logistic regression models for all the events in which the frequencies permitted such an analysis. Table 2 shows that, among those patients who experienced some adverse events or unplanned interventions, the women had 20 times higher odds than the men of requiring prolonged bag-mask ventilation (P < 0.05) and 99 times higher odds of unplanned admission to an ICU or a hospital (P < 0.10). Overall, the women had 24.3 times higher odds than the men of experiencing an adverse event with some harm to them (harm score, 2–3) (P < 0.10), even when controlling for a range of other patient characteristics, procedural location, and year. Next, both our logit models and nonparametric bivariate analyses (lowess plots) show that the links between age and adverse events are curvilinear. Figure 5 illustrates these analyses and shows that probabilities of experiencing oversedation and needing reversal agent first decrease until approximately age 50 and then begin to rise, being highest for individuals in their 80s. These figures also show that probability of experiencing hypotension remains stable (or potentially even slightly decreases) until one’s 70s, and after that it rapidly rises. Although our bivariate nonparametric regression suggests that unplanned admissions first decrease until a person is approximately 60 years old and then begin to rise, multivariate results primarily confirm the initial decline but do not show a pronounced increase in the probability of unplanned ICU or hospital admission. We also find that individuals’ chances of being bag-mask ventilated for prolonged periods increase through ages 30 to 40 years but begin to drop after they turn 50 years old. Interestingly, the chances of some overall harm to the patient seem to initially increase with age, but eventually, these chances begin to decline when patients are in their 70s and 80s. Next, our multivariate results confirm that patients with high BMI are more susceptible to several adverse events. In our sample of patients who experienced some adverse events, an increase of 1 kg/m2 in BMI is linked to a 34.8% increase in the odds of hypoxemia (P < 0.05), a 33% increase in the odds of needing prolonged bag-mask ventilation (P < 0.10), and 16.8% increase in the odds of experiencing increasing discomfort during the procedure. These analyses, however, also confirm that patients with higher BMI are less likely to experience hypotension: an increase of 1 kg/m2 in BMI is linked to a 37% decrease in odds of hypotension among those who experienced at least 1 adverse event. Multivariate analyses also show some significant curvilinear relationships between BMI and adverse event characteristics. The top left section of Figure 6 shows that, as BMI increases, the likelihood of miscommunication seems to decrease until BMI reaches 32, and after that, it begins to increase, albeit slightly. The panel on the bottom left confirms that the chances of a harmful event are the highest for those with BMIs at obese levels, but it also suggests (at least based on multivariate results) that these chances of harmful events are also higher for those with © 2014 Lippincott Williams & Wilkins




FIGURE 4. Incidence of adverse events based on procedural location. Results are expressed in percentage ± SE from total number of adverse events or required unplanned interventions reported in each specific location. Statistical significance is indicated using P values based on Fisher’s exact test.

BMI on the low end of the scale than for those with BMI in the middle of the normal range. Next, turning to comorbidities, multivariate analyses in Table 2 show that every additional comorbidity is linked to almost 9 times higher odds of ICU or hospital admission for sedation-related complication. Moreover, as the top right panel in Figure 6 shows, the chances of miscommunication about a moderate sedation case are highest for those with 2 and 3 comorbidities, and these chances are lower both for those with few or no comorbidities and for those with 4 or more comorbidities. The lower panel on the right shows a similar pattern for the chances of some harm (confirming the results of bivariate analyses) because these chances are the highest for those with 2 or 3 comorbidities. Our multivariate models in Table 2 confirm a number of differences across locations where the procedures were done. Notably, some events did not occur at certain locations at all, and we had to combine those location categories with another, larger category—inpatient floor. In other cases, specific events were very rare at some locations, resulting in extremely high odds ratios. Oversedation, hypoxemia, and the need for either a reversal agent or a lengthy bag-mask use are especially likely to occur in GI endoscopy and cardiology. Hypotension is especially © 2014 Lippincott Williams & Wilkins

common in radiology and extremely unlikely in the ED. Bagmask use of more than 5 minutes or increasing patient discomfort has not occurred in the ED in this sample; in contrast, unplanned ICU admission and hospitalization are much more common in the ED than in other locations. Unplanned admissions for sedation complications are also common, however, in GI endoscopy and cardiology, which is not surprising given the high rates of oversedation and hypoxemia at those locations. Miscommunication is especially common at the ED and on the inpatient floor and has not occurred at GI endoscopy locations. Ultimately, however, adverse sedation-related events with some harm to the patient are most common at the GI endoscopy locations.

DISCUSSION The variety of diagnostic and therapeutic options available for patient care expands the field of interventional procedures and demands new approaches to patient management. Our study delineates types of adverse events typically encountered during procedural sedation, highlighting both clinical events and events related to communication within the team of care providers. Clearly, an association exists between specific patient www.journalpatientsafety.com


7


Karamnov et al

TABLE 2. Odds Ratios From Logistic Regression Models for Adverse Events

Variable Female Age Age2 BMI BMI2 Comorbidities Comorbidities2 ASA class 3 or 4 Location ED Inpatient floor (+n/e) GI endoscopy Radiology Cardiology Other location Year Constant McFadden’s R2

BagReversal Mask Oversedation Hypoxemia Hypotension Discomfort Agent > 5 min

ICU or Hospital Admission 99.132† — 1.004† 0.969 — 8.868† — 0.189

Miscommunication

2.878 1.127* 1.003† 0.977 — 0.794 — 0.547

4.284 1.040 — 1.348** — 0.931 — 0.524

— 1.143 1.006* 0.633* — 0.937 — 0.520

1.536 0.978 — 1.168† — 1.246 — 1.060

6.266 20.033* 1.139* 0.880* 1.004† 0.996† 0.973 1.333* — — 0.780 1.072 — — 0.353 5.636

1.810 (Base)

0.122 0.051†

(n/e) (Base)

(n/e) (Base)

4.457 (Base)

(n/e) (Base)

1289.871* (Base)

25357.758** 17.875†

24.621* 3.295 23.215* 18.464†

(Base) 0.031† 0.146 (n/e)

10.467 422.654† 8.005 (n/e)

0.492 0.287 0.464 (n/e)

35.982* 75.795* 5.360 5.823 52.657* 65.077* 32.595* (n/e)

389.266† (n/e) 80.529† (n/e)

(n/e) 0.777 (Base) (n/e)

1.485* 0.120 0.415

0.749† 2.503 0.426

0.885 0.001** 0.475

0.864 0.172† 0.114

1.399† 0.044* 0.460

1.068 0.001** 0.409

0.660 0.000** 0.522

0.752 0.960 — 0.835† 1.025† 1.661 0.220* 4.173

0.582* 0.077 0.488

Harm Done 24.294† 1.015 0.994† 1.745* 1.041* 1.230 0.410* 0.706 8.828 30.798 1691.744† (Base) 13.943 73.085† 1.031 0.523 0.580

Statistically significant odds ratios are indicated as follows: †P < 0.1, *P < 0.05, **P < 0.01. Continuous variables were mean-centered for these analyses. Quadratic terms were included only where statistically significant. Base = base or reference category that is used as the comparison group when determining odds ratios for specific location; base categories were selected to best highlight significant differences among locations. Two modifications were made to deal with perfect prediction problems: (1) For a few outcomes, some location categories were combined with the “floor” category because no events of that type occurred at these locations (such instances are marked “n/e”); (2) Sex variable was omitted from the hypotension model because all cases of hypotension were women.

characteristics such as age and comorbidities and an increased risk for an adverse event. These data can help stratify patients into lowand high-risk categories, allowing for a thorough preprocedure risk assessment as well as appropriate periprocedure monitoring and clinical staffing. A thorough preprocedure patient evaluation is critical. For example, age is a widely recognized risk factor of sedation-related complications, including delirium,31,32 postsedation cognitive dysfunction,33 aspiration,34 and cardiopulmonary events.35 Dose reduction of medications used for procedural sedation is one of the strategies to reduce complications in the elderly population.36 However, patient discomfort related to inadequate sedation is also a serious adverse event that might occur. Careful adjustment of sedative medications can be challenging, especially for providers with limited training. Our results show a significant increase in the incidence of hypotension, oversedation, and apnea as well as the subsequent necessity for the use of reversal agents for older patients. One may argue that it is related to the increasing overall patient fragility with age. Hypoxemia was also reported more commonly in patients with advanced age (P = 0.072), although this finding disappeared once we controlled for other patient and procedure characteristics. Interestingly, however, whereas the chances of either temporary or permanent harm seem to rise with age, these chances begin to decrease for those in their 70s and 80s. It is possible that these patients were on average healthier, and age is a better predictor of an outcome when considered with coexisting comorbidities. It is also possible that providers approached patients in such age groups with extra care, thereby avoiding some complications.

8


Overweight and obese patients were the other high-risk subgroups. Higher BMI alters airway anatomy, cardiovascular and respiratory physiology, as well as pharmacodynamics. It carries a significant additional risk to the patient undergoing procedural sedation. The BMI is reported to be associated with higher incidence of sedation-related complications.35 Our results showed higher incidence of adverse events that resulted in at least some harm to the patients who had a higher BMI. We also found that patients with a higher BMI were more prone to hypoxemia. Decreased functional residual capacity, increased oxygen consumption, and airway prone to obstruction may play a role here. Intubation and prolonged bag-mask ventilation also had a tendency to be more prevalent in high BMI group (P = 0.09 and P = 0.14, respectively). Interestingly, the patients with lower BMI experienced hypotension more often than those with higher BMI. Controlling for preexisting blood pressure abnormalities, this may be related to the more permissive approach to sedation in patients with lower BMI who are considered to be a lower-risk population. Lower reported incidence of oversedation, apnea, and use of reversal medications in these patients could be explained by a lower rate of comorbidities that can predispose to these complications, such as obstructive sleep apnea. Patient gender seemed to play a role in patient outcome and incidence of adverse events. The female patients significantly more often experienced hypotension and oversedation and subsequently required prolonged bag-mask ventilation and reversal agents. The reasons for such differences are not clear and should be investigated further. © 2014 Lippincott Williams & Wilkins




FIGURE 5. Patient outcomes based on patient age. Results are expressed as percentage of patients of a given age who experienced an adverse event or required a specific intervention. Dashed line is based on nonparametric bivariate analyses using locally weighted running-mean smoother (lowess); solid line is based on predicted probabilities from multivariate logit models controlling for sex, age, BMI, number of comorbidities, ASA class, location, and year (all controls other than age held at their means).

Next, we found that the overall number of patients’ comorbidities may affect procedural sedation and may be associated with adverse outcomes. Our results showed a similar curvilinear pattern (Fig. 6, right plots, dashed line) in number of comorbidities and percentage of cases associated with harm to the patient. The peak of this trend was in the group of patients with 2 preexisting © 2014 Lippincott Williams & Wilkins

medical conditions. Interestingly, percentage of cases with reported incidents dramatically drops from the group with 3 comorbidities to the one with 4 or more. One may argue that, similar to the extra care they took with their oldest patients, providers were more careful and vigilant during sedation of patients with 4 or more comorbidities, who appeared to be more fragile. www.journalpatientsafety.com


9


Karamnov et al

FIGURE 6. Patient outcomes based on patient BMI (left) and number of comorbidities (right). Results are expressed as percentage of patients at a given level of BMI (left) or a given number of comorbidities (right) who experienced an adverse event. Dashed line is based on nonparametric bivariate analyses using locally weighted running-mean smoother (lowess); solid line is based on predicted probabilities from multivariate logit models controlling for sex, age, BMI, number of comorbidities, ASA class, location, and year (all controls other than BMI for the left panels and number of comorbidities for the right panels are held at their means).

Our study is unique in that it examined both patient- and provider-related adverse events. Poor communication and suboptimal team dynamics comprised a surprisingly high percentage of provider-related events, including poor communication between different services involved in the procedure, disagreement between the physician and the nurse on how to proceed with sedation, as well as miscommunication between providers within the same service. Another relatively common event in this group was administration of sedation by a provider who was not certified to deliver moderate sedation. This last finding points to the need for an institutional policy that outlines provider credentialing and education processes. Moreover, in addition to the importance of patient profiles, the study suggests that certain types of procedures have a higher rate of adverse events. Gastroenterology suite seems to be a location where oversedation and hypoxemia are common and many unplanned interventions take place. Providers should take extra care when planning for procedural sedation at this location, as well as in cardiology. The ED is unique in terms of the variety of patient pathology, high patient flow, and the number of consultant services involved

10


in patient care. This environment may compromise information transit, especially in overwhelming and stressful situations of crisis. In turn, this may subject patients to increased risk. Checklists and close-loop communication might help to prevent this group of adverse events from happening.35,36 The most obvious limitation of the current study is the very low number of reported events. The assumption that this small number accurately reflects the true number of untoward events may be erroneous; however, the safety reporting system is available to all providers administering sedation, and the assumption that the sample accurately reflects the types of events occurring is not unreasonable. Another potential limitation of our study is the absence of data on exhaled carbon dioxide (CO2) monitoring during reported cases. Our facility introduced a policy in 2013 mandating the use of capnography during sedation. In addition, our institutional policy allows only benzodiazepines and opioid administration for moderate sedation; the use of propofol is not permitted for this level of sedation, although it was reported to have been improperly used in 3 of the cases in our database. Delineation of provider experience/training is not currently part of our database; hence, our analysis is missing outcome © 2014 Lippincott Williams & Wilkins



comparisons between providers with different levels of training. This represents a potential area for future research. In addition, future studies should examine all cases of sedation OOR, not only those with adverse event reports, to better establish the relationships between patient and procedural characteristics and the occurrence of adverse events and unplanned interventions in the broader population of patients undergoing moderate sedation.

CONCLUSIONS Careful patient evaluation and risk assessment are necessary before delivering moderate procedural sedation. Multiple factors related to the patient, the procedural location, and the procedure itself can be predictive of sedation-related complications and should be taken into consideration by providers. REFERENCES 1. Karan SB, Bailey PL. Update and review of moderate and deep sedation. Gastrointest Endosc Clin N Am. 2004;14:289–312. 2. Otley CC, Nguyen TH. Safe and effective conscious sedation administered by dermatologic surgeons. Arch Dermatol. 2000;136:1333–1335. 3. Rodgers SF. Safety of intravenous sedation administered by the operating oral surgeon: the first 7 years of office practice. J Oral Maxillofac Surg. 2005;63:1478–1483. 4. Rodgers SF, Rodgers MS. Safety of intravenous sedation administered by the operating oral surgeon: the second 7 years of office practice. J Oral Maxillofac Surg. 2011;69:2525–2529. 5. Natale A, Kearney MM, Brandon MJ, et al. Safety of nurse-administered deep sedation for defibrillator implantation in the electrophysiology laboratory. J Cardiovasc Electrophysiol. 1996;7:301–306. 6. Jensen JT, Banning AM, Clementsen P, et al. Nurse administered propofol sedation for pulmonary endoscopies requires a specific protocol. Dan Med J. 2012;59:A4467. 7. Hick JL, Mahoney BD, Lappe M, et al. Prehospital sedation with intramuscular droperidol: a one-year pilot. Prehosp Emerg Care. 2001;5: 391–394. 8. American Society of Anesthesiologists. Practice guidelines for sedation and analgesia by non-anesthesiologists. Anesthesiology. 2002;96: 1004–1017. 9. American Society of Anesthesiologists. Continuum of depth of sedation: definition of general anesthesia and levels of sedation/analgesia. 2009. Available at: http://www.asahq.org/For-Members/Standards-Guidelinesand-Statements.aspx. Accessed October 20, 2013. 10. Mazanikov M, Udd M, Kylanpaa L, et al. A randomized comparison of target-controlled propofol infusion and patient-controlled sedation during ERCP. Endoscopy. 2013;45:915–919. 11. Goudra BG, Singh PM, Sinha AC. Outpatient endoscopic retrograde cholangiopancreatography: safety and efficacy of anesthetic management with a natural airway in 653 consecutive procedures. Saudi J Anaesth. 2013;7:259–265. 12. Newstead B, Bradburn S, Appelboam A, et al. Propofol for adult procedural sedation in a UK emergency department: safety profile in 1008 cases. Br J Anaesth. 2013;111:651–655. 13. Pino RM. The nature of anesthesia and procedural sedation outside of the operating room. Curr Opin Anaesthesiol. 2007;20:347–351. 14. Cravero JP. Risk and safety of pediatric sedation/anesthesia for procedures outside the operating room. Curr Opin Anaesthesiol. 2009;22:509–513. 15. Cravero JP, Beach ML, Blike GT, et al. The incidence and nature of adverse events during pediatric sedation/anesthesia with propofol for procedures outside the operating room: a report from the Pediatric Sedation Research Consortium. Anesth Analg. 2009;108:795–804.

© 2014 Lippincott Williams & Wilkins


16. Cravero JP, Blike GT, Beach M, et al. Incidence and nature of adverse events during pediatric sedation/anesthesia for procedures outside the operating room: report from the Pediatric Sedation Research Consortium. Pediatrics. 2006;118:1087–1096. 17. Bhananker SM, Posner KL, Cheney FW, et al. Injury and liability associated with monitored anesthesia care: a closed claims analysis. Anesthesiology. 2006;104:228–234. 18. Metzner J, Posner KL, Domino KB. The risk and safety of anesthesia at remote locations: the US closed claims analysis. Curr Opin Anaesthesiol. 2009;22:502–508. 19. American Society of Anesthesiologists Standards for Basic Anesthetic Monitoring. Amended 2011. Available at: http://www.asahq.org/ForMembers/Standards-Guidelines-and-Statements.aspx. Accessed October 20, 2013. 20. Oh JE, Lee HJ, Lee YH. Propofol versus midazolam for sedation during esophagogastroduodenoscopy in children. Clin Endosc. 2013;46:368–372. 21. Garewal D, Powell S, Milan SJ, et al. Sedative techniques for endoscopic retrograde cholangiopancreatography. Cochrane Database Systemat Rev. 2012;6:CD007274. 22. McQuaid KR, Laine L. A systematic review and meta-analysis of randomized, controlled trials of moderate sedation for routine endoscopic procedures. Gastrointest Endosc. 2008;67:910–923. 23. Collado V, Faulks D, Nicolas E, et al. Conscious sedation procedures using intravenous midazolam for dental care in patients with different cognitive profiles: a prospective study of effectiveness and safety. PLoS One. 2013;8:e71240. 24. Rothman DL. Sedation of the pediatric patient. J Calif Dent Assoc. 2013; 41:603–611. 25. McCoy S, Wakai A, Blackburn C, et al. Structured sedation programs in the emergency department, hospital and other acute settings: protocol for systematic review of effects and events. Syst Rev. 2013;2:89. 26. Metzner J, Domino KB. Risks of anesthesia or sedation outside the operating room: the role of the anesthesia care provider. Curr Opin Anaesthesiol. 2010;23:523–531. 27. Metzner J, Domino KB. Outcomes, controversies and future trends. In: Urman RD, Kaye AD ed. Moderate and Deep Sedation in Clinical Practice. New York, NY: Cambridge University Press; 2012. 28. American Society of Anesthesiologists. New classification of physical status. Anesthesiology. 1963;24:111. 29. Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3. Düsseldorf, Germany: Heinrich Heine University Düsseldorf; 2013. 30. StataCorp. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP; 2013. 31. Inouye SK, Westendorp RG, Saczynski JS. Delirium in elderly people. Lancet. 2013;383:911–922. 32. Brown EN, Purdon PL. The aging brain and anesthesia. Curr Opin Anaesthesiol. 2013;26:414–419. 33. Padmanabhan U, Leslie K, Eer AS, Maruff P, Silbert BS. Early cognitive impairment after sedation for colonoscopy: the effect of adding midazolam and/or fentanyl to propofol. Anesth Analg. 2009;109:1448–1455. 34. Cooper GS, Kou TD, Rex DK. Complications following colonoscopy with anesthesia assistance: a population-based analysis. JAMA Intern Medi. 2013;173:551–556. 35. Guo Y, Zhang H, Feng X, et al. A retrospective study of risk factors for cardiopulmonary events during propofol-mediated gastrointestinal endoscopy in patients aged over 70 years. Middle East J Anesthesiol. 2012; 21:505–511. 36. Arriaga AF, Bader AM, Wong JM, et al. Simulation-based trial of surgical-crisis checklists. N Engl J Med. 2013;368:246–253.



11

Anesthesia and sedation outside the operating room.

Risk factors leading to failed procedural sedation in children outside the operating room.

Anesthesia and sedation outside of the operating room.

Adverse events associated with procedural sedation in pediatric patients in the emergency department.

Role of novel drugs in sedation outside the operating room: dexmedetomidine, ketamine and remifentanil.

Anesthesiologist's interactions with other medical departments outside the operating room.

Emergency management of the airway outside the operating room.

Management of anesthetic emergencies and complications outside the operating room.

Incidence of adverse events in paediatric procedural sedation in the emergency department: a systematic review and meta-analysis.

Incidence of Adverse Events in Adults Undergoing Procedural Sedation in the Emergency Department: A Systematic Review and Meta-analysis.

Monitored anesthesia care in and outside the operating room.

Adverse reactions to 6% hydroxyethyl starch in the operating room.

Sedative medications outside the operating room and the pharmacology of sedatives.

A prospective evaluation of the incidence of adverse events in nurse-administered moderate sedation guided by sedation scores or Bispectral Index.

Finding a way through the sedation labyrinth: is it conscious, moderate, deep, or procedural sedation? Emerging evidence for CNS practice.

Adverse events associated with percutaneous enteral access.

Oral Sedation Postdischarge Adverse Events in Pediatric Dental Patients.

Adverse Events by Sedation Type in Gastrointestinal Endoscopy.

Back to basics: procedural sedation.

Moderate sedation: Introducing the "modified sedation continuum" and the "moderate sedation ladder".

Metamizole-associated adverse events: a systematic review and meta-analysis.

Advanced Monitoring Is Associated with Fewer Alarm Events During Planned Moderate Procedure-Related Sedation: A 2-Part Pilot Trial.

Alarms in the operating room.

Music in the operating-room.