Objectives Sedation has been established for GI endoscopic procedures in most countries, but it is also associated with an added risk of complications. Reported complication rates are variable due to different study methodologies and often limited sample size.
Designs Acute sedation-associated complications were prospectively recorded in an electronic endoscopy documentation in 39 study centres between December 2011 and August 2014 (median inclusion period 24 months). The sedation regimen was decided by each study centre.
Results A total of 368 206 endoscopies was recorded; 11% without sedation. Propofol was the dominant drug used (62% only, 22.5% in combination with midazolam). Of the sedated patients, 38 (0.01%) suffered a major complication, and overall mortality was 0.005% (n=15); minor complications occurred in 0.3%. Multivariate analysis showed the following independent risk factors for all complications: American Society of Anesthesiologists class >2 (OR 2.29) and type and duration of endoscopy. Of the sedation regimens, propofol monosedation had the lowest rate (OR 0.75) compared with midazolam (reference) and combinations (OR 1.0–1.5). Compared with primary care hospitals, tertiary referral centres had higher complication rates (OR 1.61). Notably, compared with sedation by a two-person endoscopy team (endoscopist/assistant; 53.5% of all procedures), adding another person for sedation (nurse, physician) was associated with higher complication rates (ORs 1.40–4.46), probably due to higher complexity of procedures not evident in the multivariate analysis.
Conclusions This large multicentre registry study confirmed that severe acute sedation-related complications are rare during GI endoscopy with a very low mortality. The data are useful for planning risk factor-adapted sedation management to further prevent sedation-associated complications in selected patients.
Trial registration number DRKS00007768; Pre-results.
- cardiovascular complications
- gastrointesinal endoscopy
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Sedation in GI endoscopy is associated with an additional risk of complications which has been shown to be very low.
What are the new findings?
While confirming the low rate of acute sedation-related complications, the lowest risk was found with propofol monosedation.
Higher complication rates seen with patients examined at tertiary referral centres and in procedures supported by an additional person responsible for sedation confirm the necessity of intensified monitoring of at-risk patients under these circumstances.
How might it impact on clinical practice in the foreseeable future?
Risk-adapted management can be better planned on the basis of these data.
Propofol alone should be the preferred sedation for the vast majority of endoscopy patients.
In the majority of cases in most countries, GI endoscopic procedures are carried out under sedation. Sedation increases the patient’s comfort, and many examinations and interventions only become possible through the use of sedation. Especially for prolonged endoscopy procedures and technically demanding therapeutic interventions, sedation is considered indispensable. On the other hand, sedation is associated with an additional risk of cardiopulmonary complications, almost irrespective of the agents used. The various national and international gastroenterology societies have therefore published guidelines on the use of sedatives and recommendations for a standardised approach including involvement of anaesthesiologists.1–6 Many of the recommendations given in the guidelines are based on limited evidence, as the data published so far on sedation-associated complications are mostly retrospective. Due to the well-established low rate of sedation-associated complications, only large studies (>100 000 endoscopies) allow for more realistic estimates.
In the ProSed 2 Study presented here, incidence and type of sedation-associated complications were prospectively recorded in a multicentre setting including more than 310 000 GI endoscopy procedures. An important secondary aim of the study was to identify risk factors for the development of sedation-associated complications.
Patients and methods
From December 2011 to August 2014, all endoscopic procedures conducted in 39 participating hospitals were consecutively recorded during various time periods; only hospitals with more than 2500 annual endoscopies and those which used the E&L Medical endoscopy documentation system were included. After study start, additional study centres were recruited, and none of the study centres discontinued participation until study end in August 2014. Online supplementary figure 1 shows the distribution of the study centres in Germany. The electronic case report form (CRF) included obligatory inputs for study parameters precluding finalising the endoscopy report without data input on sedation. Beyond this fact, no monitoring of patient and study data was performed.
Supplementary file 1
The study was carried out in accordance with good clinical practice requirements. All of the state medical councils for the participating research centres (in the states of Bavaria, Baden-Württemberg, Berlin, Brandenburg, Bremen, Hamburg, Lower Saxony, North Rhine–Westphalia, Rhineland-Palatinate, Saxony-Anhalt, Schleswig-Holstein and Thüringen) expressed their agreement with the ethics committee’s vote.
Outcomes and definitions
Main outcome was the rate of acute sedation-associated complications during the study period. Severe/major complications were defined as need for intensive-care unit admission, intubation, resuscitation and/or death. Minor sedation-associated complications were defined as paradoxical reaction (with sedative administration leading to increased restlessness in the patient), respiratory depression (sO2 <90% > 10 s), hypotension (drop in systolic blood pressure of >25%), or bradycardia (drop in heart rate of >20%) or tachycardia (>100 bpm). It was possible for other minor complications to be added in a free-text field—for example, the occurrence of paravenous injection into tissue. In addition, management of the above complications was documented (administration of an antidote, transfer to monitoring ward, securing airways, mask ventilation, endotracheal intubation, increasing of the oxygen supply, free-text field). Of note, mask ventilation due to desaturation was included into minor complications, and this changed only with endotracheal intubation.
Deaths were regarded as sedation-associated if it was the administration of sedative drugs that triggered the fatal complication or if the occurrence of the complication was not explainable and was typical of a complication caused by sedative agents—for example, hypotension or apnoea—or if the sedation was suspected to have led to additional deterioration in a situation that was already critical. This assessment was left to the individual examiner and not discussed in a study monitoring board.
The electronic documentation system used for the study (E&L Medical Systems, Erlangen, Germany) includes an electronic record for storing data on sedation and sedation-associated complications as obligatory input; only entering these data allowed for completion and storage of the endoscopy report. The sedation record was electronically linked to the data for the endoscopic examination.
To further characterise and assess complications and the respective risk factors, the following details given by the examiner had to be documented in the record:
Sedative agents and dosages used.
The occurrence of major (see above) and minor complications (none, paradoxical reaction, respiratory depression with sO2 <90% > 10 s, hypotension with a drop in systolic blood pressure >25%; drop in heart rate >20%; other).
Measures taken when complications developed (antidote administration, transfer to monitoring ward, securing airways, mask ventilation, intubation, increase in oxygen administration, other); and the outcome (no deficit, deficit, death).
Patient (risk) factors: age and sex, American Society of Anesthesiologists (ASA) classification (ASA 1–ASA 5).
Risk factors of the endoscopic procedure: Type of procedure (oesophagogastroduodenoscopy (OGD), colonoscopy, endoscopic ultrasonography (EUS), endoscopic retrograde cholangiopancreatography (ERCP), enteroscopy); diagnostic or therapeutic procedure; emergency versus elective procedure; the duration of the examination in four categories (<10 min, 10–30 min, 30–60 min, >60 min),
Hospital and staff risk factors: level of care of the participating hospital (primary care, secondary and tertiary referral hospitals), the number and specialisation of the medical staff involved in the procedure such as endoscopists and endoscopy assistant (two-person team) as well as the presence of additional staff responsible only/mainly for administering and monitoring sedation (three-person team), namely endoscopy assistant (nurse assisted endoscopy sedation, or nurse-assisted propofol sedation/nurse-administered propofol sedation (NAPS) if applicable), physician/gastroenterology fellow or gastroenterologist with special training in intensive care medicine, anaesthesiologist.
The study data were transferred from the centres to the study centre electronically and in pseudonymised form. When a major complication occurred, the study protocol explicitly permitted detailed querying by the study’s directors regarding the way in which the complication was managed and the subsequent disease course in the study centre concerned. All authors had access to the study data and reviewed and approved the final manuscript.
Depending on scale and distribution results are shown either as arithmetic mean with SD, median with 25%–75% quartiles or frequencies with proportions. Accordingly, a two-tailed non-paired t-test, Mann-Whitney U test or Fisher’s exact test was used to compare groups. Multiple analysis was done by using a binomial logistic regression model to identify risk factors for sedation-associated complications. Pairwise potential relationships between risk factors were considered by including the corresponding interaction term into the multiple model. A P value <0.05 was considered to be statistically significant. Analyses were conducted by using R V.3.3.2 (The R Foundation, Vienna, Austria).
Patient and procedural data
A total of 39 hospital gastroenterology departments in Germany participated in the study. The centres documented examinations over a median period of 24 months (range 1–32 months). A total of 368 206 endoscopy procedures was recorded between December 2011 and August 2014. Each of the centres included a mean of 9441 endoscopies (range 818–28 094) (figure 1); centres for which comparatively small numbers of endoscopic procedures are included (<1000 endoscopies) only participated in the study for a few weeks, but these centres were all also endoscopically active departments (with more than 2500 endoscopies per year). The hospitals levels of care provision was primary care (n=19), and secondary (n=12) and tertiary referral centres (n=8); 31 of the 39 centres had haemodialysis facilities, and one hospital had a solid organ transplant centre. Online supplementary table 1 provides the characteristics of patients and endoscopy procedures relative to the hospital’s level of care provision
Supplementary file 2
Of the endoscopy procedures 10.9 % were carried out without sedation (n=40 280). The 25 complications observed in this group were classified as minor in 22 patients (0.05%), and major complications occurred in 3 patients (0.007%). One patient died periprocedurally (0.002%). There were no differences in the median age of patients who received sedation and those who underwent examinations without sedation and those who did not. More patients examined without sedation were in ASA classes 3–5 in comparison with those who received sedation (27.2% vs 21.7%). Among the unsedated patients, emergency indications were more frequent (11.5% vs 5.2%), therapeutic procedures less frequent (15.6% vs 21.4%) and examination times shorter (<10 min: 4.67% vs 26.1%). There were no major differences with regard to the type of procedure between these two groups.
Certain data such as gender were missing for 13 736 patients (4.2%) due to a compatibility problem in the electronic querying system. Therefore complete data on analysed variables were available in 314 190 (85.2%) procedures, and these are the basis of the following analyses including the multiple logistic regression analysis (figure 2). Patient sex ratio was equal (49.9% men, 50.1% women), the mean age was 65 years (range 0–104 years). Seventy-nine per cent of the patients (248 774) were in ASA groups 1 or 2. Since the study focused on acute sedation-associated complications on a procedural basis, multiple counts of single patients was not specifically analysed per endoscopy; (>1 endoscopy in the same patient).
Procedures included OGD 52.6% (n=165 269); colonoscopy 32.9% (n=103 367); EUS 7.2% (n=22 515); ERCP 6.7% (n=20 967); and small-bowel endoscopy 0.7% (n=2072). Small-bowel endoscopy included both push enteroscopy and also balloon enteroscopy. Differentiated analysis of the percutaneous endoscopic gastrostomies (PEGs) was also not possible, as only 107 were recorded, while most of the PEG procedures that were carried out were documented as therapeutic OGDs. Therapeutic procedures were performed in 21.4% (67 385) of the endoscopies under sedation, and 5.1% (16 148) were classified as emergency endoscopies. The duration of the examination was between 10 min and 30 min in most cases (<10 min, n=82 276; 10–30 min, n=186 395; 30–60 min, n=38 349; >60 min, n=7080).
Sedation details and teams
The sedative agents were chosen in accordance with the usual procedure in each individual study centre (table 1, online supplementary table 2). Propofol was the preferred agent (used alone in 61.7% and in combination with midazolam in 23.0%).
Supplementary file 3
Sedation was administered by the two-person team (endoscopist/endoscopy assistant supervised by the endoscopist) in 50.8% of cases (n=159 718). In this group, a third person to take care of sedation was requested secondarily during the course of endoscopy (ie, after start of sedation) in 5341 cases (1.7% of all cases), consisting of a second assistant (0.8%), a second physician trained in intensive care medicine (0.9%); reasons were not prospectively documented, but were mostly sedation-related problems.
In the remaining cases (49,2%) a third person was primarily in charge of sedation (three-person endoscopy/sedation team). This consisted of NAPS in the vast majority of cases, namely 38.9% (n=122 074), a second specialist physician (with experience in intensive-care medicine) in 10.1% of cases (n=31 886) and an anaesthesiologist in 0.2% (n=512).
As mentioned above, 1.7% changed from the two-person to the three-person endoscopy/sedation team; thus final percentages were 49.1% and 50.9%.
In the 314 190 endoscopies under sedation, a total of 1054 complications occurred (0.3%). These were major complications in 0.01% (n=38), with a sedation-related mortality of 0.004% (n=13) and minor complications in 0.3% (n=1019).
Of the 38 patients who suffered a sedation-associated major complication, 28 were transferred to the intensive-care unit and 26 also required intubation. Thirteen of the 38 patients died (0.004%) (online supplementary tables 4 and 5). Nine patients of the 25 patients with a non-fatal major complication received resuscitation measures.
In 9 of the 13 patients who suffered a fatal complication, endoscopy was performed as an emergency procedure, in the majority of cases (n=7) as emergency OGD for haemodynamically significant upper GI bleeding. In one other patient, OGD was carried out in a case of anastomotic leakage with septic mediastinitis. Among the five fatalities occurring during ERCP, one was an emergency procedure. The mean patient age was higher (75 years, range 52–90 years) and the rate of ASA class >2 cases (85%) were higher than in the entire group.
Minor complications are listed in table 2. To manage the complications in 283 patients, oxygen administration was increased, and 100 patients had the airways mechanically secured (with Guedel’s/Wendel’s tube, Esmarch’s manoeuvre). Nineteen patients required mask ventilation due to desaturation (0.006%). Antidote administration was carried out (with naloxone or flumazenil) in 34 patients.
Risk factor analysis
Selected multiple logistic regression analysis (after exclusion of non-significant variables by univariate analysis (online supplementary table 5) and multiple logistic regression (online supplementary table 6)) confirmed the eight factors as risk factors for sedation-associated complications in GI endoscopy (table 3): one was patient-associated (ASA class >2), four were procedure-related (emergency/therapeutic endoscopy, type and duration of endoscopy), two had to do with hospital logistics (hospital care level, two-person versus three-person team) and one was medication-associated (propofol monosedation with lowest risk).
Supplementary file 6
Supplementary file 7
Of the other factors, age, sex and case volume of the hospital were not significant factors in the multivariate analysis.
Sedation has been established for GI endoscopic procedures in most countries. In addition to increasing the patient’s comfort, sedation has enabled to carry out technically demanding and time-consuming endoscopic interventions such as ERCP and small-bowel endoscopy. It probably also increases the quality and safety of the endoscopic procedure as well.7–9
These advantages of sedation are associated with an increase in cardiopulmonary risks. Many specialist societies have therefore published guidelines on sedation in endoscopy including involvement of anaesthesiologists.1–6 The quality of the evidence for the recommendations is limited by a paucity of large and prospective studies; most of these had been published years ago, and it can be assumed that especially with agents such as propofol, experience has been accumulating, so that application and complications may have reached a steady state. To the best of the authors’ knowledge, there have only been four studies in the literature on sedation-associated complications in GI endoscopy that have included more than 100 000 endoscopic examinations, and only one of these had a prospective design (table 4).10–13 The present study, including more than 310 000 endoscopies, is the largest prospective, multicentre study conducted to date to record the incidence of and risk factors for sedation-associated acute complications in GI endoscopy at present. A special aspect of the study is that completeness of patient inclusion during the study period in each centre was ascertained by an obligatory data input field in the endoscopy report.
Due to the low complication rates and in particular the low rate of complications associated with death, studies on this scale are necessary in order to generate reliable data. In our study, the overall rate of all sedation-associated complications was 0.3%; the rates of major complications and mortality were 0.01% and 0.004%, respectively.
Various limitations of the present study should also be noted. Delayed complications in particular (eg, complications developing in the recovery room or later on the ward or after discharge) were not systematically documented, although some of them may have been included in the present study. In addition, it could be speculated, whether all minor complications were fully recorded in the study. For example, paradoxical reactions to sedation were much higher in a recent study using benzodiazepine-based regimens.14 Another limitation due to the methodology is the absence of standardisation both in relation to the choice and administration of sedative agents and also in the staffing modalities used for monitoring. On the other hand, however, the present study represents real-life conditions and therefore provides valid data on the safety of sedation in GI endoscopy, at least in everyday hospital work.
Furthermore, in contrast to other studies, mask ventilation as a therapeutic management of desaturation was not rated as a major complication. This might, at first sight, appear to reduce the comparability of the study data. If desaturation leading to mask ventilation were to be classified as a major complication, the rate of major complications would increase from 0.01% to 0.02%.
The results of other large studies on sedation in endoscopy published yielded different results in some respects (table 4). The first large retrospective study (n=324 737), by Sharma et al in 200710 used sedation based on benzodiazepines, supplemented with an opiate if necessary without detailed data about regimen and dosage. The rates of sedation-associated complication rate and mortality were 0.9% and 0.008%, respectively, substantially higher than in the present study. As compared with our study, this paper included more patients in ASA classes 1 or 2 (89.9% vs 79.2%) and also included more colonoscopies (70.3% vs 32.9%), both factors associated with lower complication rates.10 On the other hand, the largest multicentre study and analysis on sedation-associated complications in GI endoscopy focused on propofol (n=646 080)11 reported on overall complication and mortality rates of 0.1% and 0.0006%, respectively, lower than in our and other studies. Here, some methodological limitations such as retrospective data collection from papers and some centres may have led to underestimation of complications rates. Patient selection to exclude higher ASA grades may have been an additional factor.
Two larger German studies on sedation-associated complications of GI endoscopy were published recently: in a prospective registry of complications from 191 242 endoscopies performed under propofol sedation in 36 hospitals, complication and mortality rates were 0.04% and 0.003%.10 In our own ProSed 1 Study12 using a retrospective, multicentre design (n=388 404), using propofol-based sedation or benzodiazepine-based sedation, these rates were 0.01% and 0.004%, respectively. Results of these studies, which were performed in similar settings as the present prospective work, were confirmed by our study.
The results of the multivariate risk factor analysis also deserve some comments: as expected,10–13 complications increased with a higher ASA Score (ASA class >2). The increased risk for the development of sedation-associated complications during emergency endoscopy has also previously been described.12 13
The choice of sedation regimen in the present study was at the choice of each individual study centre. Still, propofol was used in the vast majority of cases, either alone (62%) or in combination with midazolam (22.5%). Propofol has been in the focus of numerous studies, mostly with lower case numbers.9 15–26 Rates of severe complications were low, ranging from 0 in ASA classes 1 and 2 patients in 200717 to 0.016% in 2014;20 none of the patients required hospitalisation or suffered long-term injury. Differences in the complication rate as compared with the studies mentioned above are probably explained by patient selection (outpatients) and the types of examination involved (mainly diagnostic colonoscopy). The present study was not a randomised trial; given this limitation, it suggests that propofol sedation may be associated with the lowest complication rate in comparison with other sedation regimens. Notably, sedation regimens including an opiate were associated with doubling of risk, even more so when combining midazolam and opiates. This has never been shown before.
This is in contrast to other papers; a meta-analysis by Qadeer et al 27—which included 12 randomised studies and 1161 patients on the complication rate of propofol versus benzodiazepine-based sedation in combination with an opiate—showed no overall difference between the two sedation regimens. The only exception was propofol sedation for colonoscopy which was associated with lower OR for cardiopulmonary complications. However, due to the low complication rates, only very large randomised studies would allow statistically valid conclusions on different complication rates.
Not unexpected, but shown for the first time, a substantial risk increase was observed by a prolonged duration of the examination, which was rather linear: as compared with short examinations (<10 min), OR for complications increased by 1.8, 3,8 and 7,9 for examinations taking up to 30 min, 30–60 min and longer than 1 hour, respectively. This information is crucial for planning sedation for longer and complex procedures. Similarly, therapeutic endoscopy independently creased the risk by the factor of 1.7. Similarly, Wehrmann et al’s 28 study showed a high sedation-associated complication rate in interventional endoscopy with an overall complication and mortality rate of 1.4% and 0.03%, respectively. The risk increase associated with OGD, ERCP and enteroscopy in comparison to colonoscopy and EUS is remarkable. Prone positioning and left-side positioning of the patient may possibly lead to more difficult respiratory conditions.
The literature published to date offers inconsistent data regarding the frequency of sedation-associated complications in specific levels of hospital care provision.10 11 29 Sharma et al found an increase of sedation-associated complication by the factor of 1.4 for procedures performed in veteran’s hospitals compared with those at a university hospital. In contrast, in our study the complication rate was more than 1.5 times higher in tertiary referral compared with primary care hospitals. It has to be mentioned that the different levels of invasiveness of the endoscopic therapeutic procedures were not documented in such detail to detect differences in complexity and risk level. Similar trends were observed when the sedation teams (two-person versus three-person teams) were compared in the multivariate analysis which showed significant risk increases when an additional person was involved dedicated to sedation. We think that this may rather mirror a presumed higher procedure risk leading to involvement of an additional person rather than the influence of the three-person team per se. It is also debatable whether ASA status as the only risk factor assessment scale used here, as in most other studies, adequately reflects patient risk. This is supported by other studies such as the study by Berzin et al 30 on sedation-associated complications during ERCP under anaesthesiological care: of 528 patients included, 3% had to undergo intubation.
With regard to sedation-associated mortality, 12 of the 13 patients who died had three or more of the risk factors identified. In the majority of the cases (n=8) the patients were examined for haemodynamically relevant upper GI bleeding. This analysis shows that sedation-associated complications need to be anticipated particularly in patients in whom several risk factors are present.
Finally, complications also occurred in the group of patients who underwent endoscopy without sedation, with 0.005% minor and 0.007% major complications and 0.002% mortality (n=1). The study was not set out to compare complication rates between sedated and unsedated endoscopy, but risk factors such as higher ASA class and emergency endoscopy showed that there was a higher risk profile in the group of unsedated patients.
In summary, the data confirm retrospective evidence from previous years with a large prospective, multicentre study design that sedation during GI endoscopy in endoscopy units with high throughput is carried out with a high degree of safety. One severe complication (serious adverse event) can be expected in 1 out of 10 000 endoscopies, and mortality rate was 5 per 100 000. We already provided indirect evidence for the relative safety of propofol monosedation. Our analysis of risk factors will enable more targeted sedation logistics especially in patients who are at increased risk. This will likely help to avoid and further reduce sedation-associated complications during GI endoscopy.
Supplementary file 4
Supplementary file 5
The authors thank E&L Medical Systems Ltd, Erlangen; Edgar Lehmann for his generous support for the project; Andrea Maybaum and in particular Dirk Saake for technical support and supervision of the data analysis.
Contributors AB planned the study, did the study organisation, was involved in data collection, prepared the manuscript, was involved in thestatistical analysis and submitted the study. CE planned the study, was involved in data collection and study organisation, prepared the manuscript and was involved in the statistical analysis. WR, CF, SP and AA were involved indata collection. AK, HM, HK, CS, AL, MP, J-UJ, CvT, ME, MS, MK, DS, WH, OE, OP, FK, BL, UW, DS, SB, SuB, VC, JFE, FLD, CB, TR, GH, MB, CRdM, GK, CP, CV, RK, WF, IK, JK, MB, NLS, AM, RV, RP, ME, WS, MD, WSch, AD and VS were involved in data collection and in manuscript creation. JL was involved in manuscript creation. GK and AK were involved in the statistical analysis. NB was involved in creating figures.
Competing interests None declared.
Ethics approval Ethics Committee of the Medical Council of the state of Hesse (FF 8/2012) on 13 April 2011.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.