Prospective Description of Coughing, Hemodynamic Changes, and Oxygen Desaturation During Endoscopic Sedation
- First Online:
- Cite this article as:
- El Chafic, A.H., Eckert, G. & Rex, D.K. Dig Dis Sci (2012) 57: 1899. doi:10.1007/s10620-012-2057-z
Deep sedation is increasingly used for endoscopy. The impact of sedation level on hemodynamic status, oxygenation, and aspiration risk is incompletely described.
To describe the incidence of intraprocedural cough, hemodynamic changes, oxygen desaturation, and their relationship to clinical factors and sedation level.
Detailed prospective recordings of hemodynamic changes, oxygen desaturation, and cough during 757 nonemergent endoscopic procedures done under sedation using propofol, midazolam, and/or fentanyl.
Thirteen percent of patients had at least one cough and 3% had prolonged cough. Cough was more common in nonsmokers (P = 0.05), upper endoscopy (P < 0.0001), with propofol (P = 0.0008), longer procedures (P = 0.0001), and hiccups (P = 0.01). The association between supine positioning during colonoscopy and cough approached significance (P = 0.06). Oxygen desaturation was rare (4%) and associated only with deep sedation (P = 0.02). Mean systolic and diastolic blood pressure (BP) dropped by 7.3 and 5.6% respectively. Decreases in systolic BP were more common in whites (P = 0.03), males (P = 0.004), nonsmokers (P = 0.04), during colonoscopy (P < 0.0001), and in patients receiving midazolam and fentanyl (P = 0.01). Heart rate (HR) dropped >20% from baseline in 15% of patients and was more common during colonoscopy (P = 0.002). HR increased >20% in 20% of patients and was more common with coughing (P < 0.0001) and in younger patients (P = 0.0002). No patient required pharmacologic treatment of BP or HR.
We have described procedural predictors of cough that may help clinicians reduce the risk of aspiration during endoscopy. Hemodynamic changes during endoscopy are common but largely clinically insignificant.
KeywordsEndoscopy, digestive system/adverse effectsDeep sedation/adverse effectsPropofol/adverse effectsHemodynamicsCoughOxygen saturation
Sedation has been found to be a major contributor to complications occurring during endoscopic procedures [1–6]. However, what constitutes a clinically significant complication of sedation is arguable. For example, sedation-induced death, neurologic sequelae, and aspiration pneumonia are clear-cut complications. On the other hand, development of hypotension or bradycardia, even if they require pharmacologic treatment but are readily reversible without sequelae, could be considered clinically insignificant complications that are an expected part of routine endoscopic sedation. Similarly, microaspiration could be considered clinically insignificant if it results in coughing but not in pulmonary infection or induction of prolonged bronchospasm. The experienced endoscopist recognizes that the occurrence of changes such as hypotension, severe heart rate changes, coughing, and oxygen desaturation, while not in themselves necessarily complications, must be recognized and in some cases treated because, if allowed to persist, clinically important complications may develop.
In our unit, we began using propofol administered by trained registered nurses more than a decade ago. During this interval, we commonly induced deep sedation and even general anesthesia during routine endoscopic procedures, without the occurrence of death or permanent sequellae [7–14]. Anecdotally, however, we frequently saw substantial hemodynamic changes which seemed almost entirely to lack clinical significance. We also anecdotally observed that occasional patients developed prolonged coughing, postprocedural bronchospasm, unexplained fever, and evidence of pulmonary infection requiring treatment with antibiotics. Such anecdotal cases led us to an approach of meticulous airway protection, including avoidance of deep sedation in patients with acute gastrointestinal bleeding, achalasia, and known delayed gastric emptying [12, 13], a policy of vigorous suctioning of the hypopharynx prior to esophageal intubation during upper endoscopy, complete suctioning of the esophagus and stomach prior to examination of the upper GI tract, and careful attention to position change during colonoscopy and to maintaining the availability of oral suctioning to manage reflux during colonoscopy . A study from Italy recently described that 18 of 17,542 patients sedated by Monitored Anesthesia Care during endoscopy required antibiotic treatment for aspiration (0.1%), and that aspiration was a significant contributor to episodes of desaturation . Thus, the occurrence of aspiration is more frequent than perforation in most endoscopic practices, and factors associated with aspiration risk should be of interest to endoscopists. It is reasonable to speculate that the increased use of deep sedation with propofol endoscopy could result in a clinically significant increase in pulmonary infections associated with aspiration. In our experience, most patients who develop pulmonary infection after endoscopy had evident coughing and often had evidence of reflux of gastric contents during the procedure.
We found no reports in the medical literature that describe the incidence of coughing during endoscopy, or factors associated with coughing. Such a study could be valuable to understanding mechanisms of aspiration during endoscopy and preventing aspiration, especially if some or most episodes of postprocedural pulmonary infection are associated with significant coughing. Therefore, we undertook a prospective documentation of the occurrence of intraprocedural cough, hemodynamic changes, and oxygen desaturation, and the relationship of these events to procedural factors and sedation level. Several previous studies have included the results of prospective measurements of hemodynamic changes throughout the course of endoscopic procedures [2, 3, 6, 9, 11–13, 17–41]. However, few studies have had the sole focus of a detailed description of the quantitative changes in hemodynamic parameters during endoscopy and their relationship to specific agents, sedation level, and clinical factors.
We prospectively recorded sedation-related events in 757 patients. The study was conducted as a quality improvement project to identify measures that could reduce our incidence of postprocedural pulmonary infections. The events were recorded by college students working in the endoscopy unit on quality-related projects in the summer of 2009. Permission to review the data for purposes of publication was granted by the Institutional Review Board at Clarian Health Partners. The students were trained to make the recordings by the senior investigator (D.R.). Sample size reflected the number of studies they could complete during their summer vacation. Consecutive patients were included to the extent that a student was available to make measurements, i.e. a patient could not be included if the students were already recording data on another patient or not otherwise available.
Patient Monitoring and Data Collection
Patient demographics (n = 757)
Age, years, mean (SD)
Gender n (%)
Patient location n (%)
Race n (%)
Respiratory risks n (%)
During the procedure, blood pressure was recorded every 5 min beginning with the onset of the initial sedative medication, or more often if indicated by changes in clinical status. Heart rate was recorded every 2 min (for monitoring purposes, the heart rate was measured continuously as was the electrocardiogram). Oxygen saturation was measured continuously and recorded for the study purposes every 1 min, and sedation level was recorded every 2 min using the Modified Observer’s Assessment of Alertness/Sedation Scale (MOAAS) scale. [The MOAAS score ranges from 0 to 6 (0 = unresponsive to deep stimuli, 1 = unresponsive to shaking, 2 = responsive to shaking only, 3 = responsive to loud verbal command, 4 = lethargic response to normal verbal command, 5 = responsive and alert, and 6 = agitated) . Deep sedation was defined as a MOAAS score ≤ 1 .] Every instance of cough and hiccup was recorded by the study assistant. The study assistant also made note of any interventions involving sedation-related issues, such as the use of atropine for bradycardia, pharmacologic treatments of blood pressure, placement of oral or nasal airways, and use of positive pressure ventilation (bag-mask ventilation). All patients were given supplemental oxygen by nasal cannula at a rate of 4 l per minute.
All the patients were sedated by trained registered nurses under the supervision of the endoscopist. No anesthesiologist or anesthetist was involved in any case. No patient received topical or pharyngeal anesthesia. Some patients received intravenous lidocaine in an attempt to prevent pain from injection of propofol. Our sedation protocols using propofol as well as opioids and benzodiazepines are based on titration of selected agents to desired levels of sedation and have been described previously [9, 11–13]. Agents were selected for each patient based on the clinical judgment of the endoscopist and the sedation nurse. Patients were more likely to receive opioids and benzodiazepines or combinations of opioids, benzodiazepines, and propofol (as opposed to propofol alone) if they had sleep apnea or higher risk airway assessments. All patients undergoing upper endoscopy received at least small doses of opioids and/or benzodiazepine.
Two sets of statistical analyses were performed. The first examined associations of changes in blood pressure, heart rate, cough, and oxygen saturation with other factors at the level of the entire procedure and the second analysis used the measurements collected longitudinally throughout the procedure. For the procedure-level analyses, each of the predictors of systolic blood pressure (SBP) and diastolic blood pressure (DBP) percent change from baseline during the procedure were tested for significance individually using correlation coefficients or ANOVA for continuous categorical predictors, respectively. Multivariable analyses were performed by including all possible predictors simultaneously in an analysis of covariance model. Analyses of the SBP and DBP data collected throughout the procedure were performed using mixed-model ANOVA with a random effect to account for the within-patient correlation. Similarly, predictors of the occurrence of any cough during the procedure, oxygen saturation <90% during the procedure, heart rate percent change from baseline >20%, were examined using ANOVA and χ2 tests for continuous and categorical predictors, respectively. The multivariable models for these outcomes were fit using logistic regression. The longitudinal analyses of the measurements throughout the procedure were analyzed using generalized estimating equation (GEE) methodology applied to logistic regression models account for the within patient correlation.
Patients and Procedures
A clinical description of the 757 patients evaluated is given in Table 1. The number of inpatients was small, and there were no differences between inpatients and outpatients in type of sedation used or any measured parameter (data not shown).
Procedures, sedation agents, sedation levels, and ancillary maneuvers
Procedure type n (%)
Procedure duration, min (SD)
Intervention n (%)
Sedation drugs n (%)
P + M + F
M + F
P + M
P + F
Intravenous lidocaine used n (%)
Maximum sedation level n (%)
Abdominal pressure during procedure n (%)
Any position change n (%)
Significant associations of coughing in multivariate analysis
Any cough during procedure (procedural level analysis)
Any cough during procedure (longitudinal level analysis)
Smoking/non smoking or asthmatic
Position change; yes/no
P/(M + F)
(P + M + F)/(M + F)
Any hiccup during procedure; yes/no
Alteration in vital signs and minor complications during the procedure
Average Δ vital signs from baseline
SBP < 90
% Δ SBP
O2 saturation < 90
O2 saturation < 85
% Δ DBP
O2 saturation drop > 4%
HR > 100
% Δ HR
HR < 60
Δ O2 sat
% Δ O2
Changes in Blood Pressure
Mean systolic blood pressure (SBP) dropped significantly during procedures by an average of 7.3% (SD 17.8), and diastolic blood pressure (DBP) by 5.6% (SD 25.4) (Table 4). The maximum drop in SBP was <10% in 28% of patients, 10–19% in 26% of patients, 20–29% in 25% of patients, 30–39% in 14% of patients, 40–49% in 6% of patients, and >50% in 1% of patients. These changes resolved spontaneously in all cases and none of the patients required pharmacologic treatment or cessation of the procedure. There was no correlation between procedure duration and blood pressure variation during the procedure, and borderline association of diastolic blood pressure drop with deeper sedation level in the longitudinal analysis model.
Significant clinical predictors of SBP decrease in multivariate analysis using two methods
% Decrease in SBP (procedural level analysis)
% Decrease in SBP (longitudinal level analysis)
Neither smokers nor asthmatics/smokers
(M + F)/P
Significant clinical predictors of DBP decrease in multivariate analysis using two methods
% Decrease in DBP (procedural level analysis)
% Decrease in DBP (longitudinal level analysis)
(P + M + F)/(M + F)
Position change during colonoscopy; no/yes
Deeper sedation level
Changes in Heart Rate
Significant clinical associations of heart rate drop >20% from baseline in multivariate analysis using two methods
HR drop > 20% during procedure (procedural level analysis)
HR drop > 20% during procedure (longitudinal level analysis)
Smoking/neither smoking nor asthmatic
Significant clinical predictors of heart rate rise >20% from baseline in multivariate analysis using two methods
HR rise > 20% during procedure (procedural level analysis)
HR rise > 20% during procedure (longitudinal level analysis)
In this report, we describe clinical factors associated with the development of cough during endoscopic procedures as well as changes in blood pressure, heart rate, and oxygen saturation. Although others have described hemodynamic changes during endoscopic sedation [2, 3, 6, 9, 11–13, 17–41], our study is among the most detailed prospective assessment of hemodynamic changes in relation to sedation levels and clinical factors thus far reported. Furthermore, to our knowledge, this is the first report of the frequency of cough during endoscopic sedation.
Although we hypothesize that development of cough might be a surrogate marker for an increased risk of aspiration-related postprocedural infection, we acknowledge that such a relationship has not been established. However, we believe it is reasonable to presume that most episodes of postprocedural pulmonary infection would be associated with intraprocedural coughing or coughing during the period of sedation. We did observe that coughing was associated with factors that might be expected to increase the risk of microaspiration, including upper endoscopy, use of propofol as a single agent, and longer procedures. Coughing was also associated with hiccuping which may occur during endoscopy during periods of gastroesophageal reflux or diaphragmatic pressure from the endoscope. In our experience, hiccuping is often associated with regurgitation of gastric contents and the onset of coughing, and we have seen several instances of apparent hiccup-induced aspiration during colonoscopy.
Our results suggest that clinicians utilizing deep sedation and general anesthesia, particularly with propofol, in nonintubated patients for endoscopic procedures, should have constant vigilance for aspiration, particularly with deeper levels of sedation, longer procedures, during hiccupping, and when colonoscopy patients are rotated to the supine position. The presence of regurgitated fluid in the mouth plus coughing suggests that microaspiration is possible or may have already occurred. Conversely, smokers had significantly less coughing than nonsmokers suggesting blunting of their cough reflex . Thus, heightened vigilance to protect against microaspiration may be warranted in smokers. We acknowledge that coughing during upper endoscopy is likely related primarily to laryngeal stimulation from the upper endoscope and pharyngeal secretions. A separate study to evaluate factors associated with coughing only during colonoscopy would be informative.
There were few patients in the study who developed oxygen saturations <90%, limiting our ability to identify clinical predictors of oxygen desaturation. Multivariable analysis only identified greater depth of sedation as a risk factor for oxygen desaturation. Although most of the patients in the study reached deep sedation or general anesthesia at some point during the procedure, all episodes of desaturation were managed successfully by the sedating nurse and endoscopist, consistent with evidence that patients can be safely targeted to deep sedation by nonanesthesiologists . In fact, we are aware of no scientific evidence to support the American Society of Anesthesiologist’s contention  or the rules of the Center for Medicare and Medicaid Services  that deep sedation should only be administered by anesthesia specialists.
With regard to blood pressure changes, the principal observation of our study was that variable degrees of hypotension occurred, which in some instances were >40% from baseline. Despite this, patients tolerated these changes without development of arrhythmia or cardiac complications, and the changes did not require pharmacologic treatment or cessation of procedures. These changes suggest that blood pressure alterations during endoscopic sedation are overwhelmingly clinically insignificant. However, the number of patients with blood pressure drops >40% was only 7%, and no patient had sustained drops. Certainly, a larger study might have identified patients with sustained hypotension for which pharmacologic treatment would be needed.
Blood pressure changes were greater in whites than nonwhites for uncertain reasons. Blood pressure drops were greater during colonoscopy, probably because of relative dehydration prior to the initiation of sedation. Since colonoscopy was associated with both hypotension and bradycardia, vagal responses likely also contributed to hypotension during colonoscopy. Blood pressure drops were less pronounced with propofol than with opioids and benzodiazepines, supporting the safety of propofol with regard to hemodynamic changes during endoscopy. We advocate that patients with significant decreases in blood pressure be placed temporarily in Trendelenburg and have an increase in their intravenous fluid infusion rate, but, in our study, pharmacologic treatment was never used.
Our study identified several clinical factors associated with changes in heart rate of more than 20% from baseline. Specifically, colonoscopy was associated with a greater risk of bradycardia, and coughing and upper endoscopic procedures with a greater risk of tachycardia, and longer duration of procedures was associated with both tachycardia and bradycardia. Although no patient in this study was given pharmacologic treatment for heart rate changes, we certainly treat significant degrees of bradycardia during colonoscopy with atropine or glycopyrrholate, and our clinical experience suggests that these changes are sometimes related to scope passage and vagal reactions to loop formation, rather than to the sedation itself. The observation that tachycardia is more common during cough and upper endoscopic procedures may reflect that some patients were not sufficiently sedated to initiate their upper endoscopy, precipitating a stress-induced sinus tachycardia.
Our study has limitations. Since it was not randomized, we cannot exclude the possibility that patient factors affected the choice of sedation agents, and that these patient factors underlie to some degree observed differences in cough and hemodynamic changes associated with sedation agents. Also, we again acknowledge that cough is only a surrogate for aspiration risk, and that much of the coughing that occurs during endoscopy is clinically insignificant.
In conclusion, we performed a detailed prospective analysis of the incidence of cough, hemodynamic changes, and oxygen desaturation in patients receiving a variety of sedation regimens for endoscopy and correlated the occurrence of these events with clinical factors, procedure type and duration, and sedation agents. We identified clinical predictors of cough during endoscopy, which may suggest situations for which increased vigilance is required to prevent microaspiration and potentially prevent postprocedural pulmonary infectious complications. These situations include deeper levels of sedation, propofol use, longer procedures, patients who are hiccupping, and patients who are rotated to the supine position during colonoscopy. Given the recent demonstration that postprocedural pulmonary infection is a significant problem for endoscopy , data of this type may be useful in developing effective measures for preventing procedure-related aspiration.
Conflict of interest