Activity Transcript

Susan P. McGrath, MD: Hello, I'm Dr Susan McGrath, director of the Surveillance Analytics Core in the Analytics Institute and the Department of Anesthesiology at Dartmouth Hitchcock Medical Center in Lebanon, New Hampshire in the United States. Welcome to this program titled "Moving Toward Consensus: Assessment and Monitoring for Opioid-Induced Respiratory Depression" (OIRD).

Joining me today are Dr Frances Chung, who is a professor in the Department of Anesthesia and Pain Medicine at the University of Toronto, in Toronto, Ontario, Canada, and Dr Kai Zacharowski, who is a professor of anesthesiology and intensive care medicine and the director and clinical head of the Department of Anesthesiology, Intensive Care Medicine, and Pain Therapy at the University Hospital Frankfurt, Goethe University, in Frankfurt, Germany. Welcome.

We know that adequate management of postoperative pain is important to ease patient suffering. We also know that controlling pain is directly related to increasing functional mobility, reducing length of stay, and increasing patient satisfaction. Opioids are frequently used to manage acute postoperative pain despite being associated with adverse events (AEs), including nausea, vomiting, sedation, and respiratory depression. Respiratory depression is an extremely important side effect of opioids and is the primary cause of death associated with opioid-based analgesics. The seriousness of OIRD was illustrated in an analysis of closed malpractice claims, where 55% of OIRD events resulted in death, another 22% in severe brain damage. Additionally, most of those events—97%—were judged as preventable with better monitoring and response. So preventing OIRD requires a nuanced assessment, with a need to have pain management and balance that with the potential for oversedation and these serious AEs.

In this program, we're going to discuss ways of improving early detection of OIRD, and we'll start with Dr Zacharowski. Can you discuss the issue of respiratory depression associated with opioid use postoperatively?

Kai Zacharowski, MD, PhD, ML, FRCA, FESAIC: Well, thank you for this question. Respiratory depression often occurs without an obvious overdose of opioids. And residual anesthesia plus opioid analgesics can lead, obviously, to respiratory depression. And drug-related factors (eg, analgesic doses) and/or patient-related factors (eg, presence of comorbidities) can enhance the central nervous system (CNS) depressant effects of opioids, and opioid accumulation occurs or excessive duration of action occurs. Various factors play an important role in this, I call it almost a disease, because people are dying.

And I would like to discuss a clinical case, an example I heard some years ago, and it's quite heart touching because it illustrates the big problem of OIRD. It’s about a young man, a 21-year-old young man, who’s healthy, and he went to the hospital for a very basic procedure. He had a tonsillectomy. And this young man is a tall, strong guy, and he usually can take a lot of pain without any problems. He does not take any analgesic drugs in daily life. And he was told by his doctor to take an opioid analgesic postoperatively for pain, because pain was to be expected. So he followed this advice. He went to bed, and several hours later, his wife tried to wake him up. He was unresponsive to cardiopulmonary resuscitation, and he died. So this is a sad story, and it shows us that this can happen even to young, healthy people. It can happen to anybody.

Dr McGrath: What a tragic case, and one I'm sure we all agree should just never be something that happens in medicine. Dr Chung, can you talk about specific factors that have been shown to increase the risk of OIRD and perhaps give us another example of a case?

Frances Chung, MBBS, MD, FRCPC: Yes. I would like to give you an example of a medical legal case that I'm aware of. This is a 50-year-old woman, with a body mass index (BMI) over 40 kg/m², and she has a history of sleep apnea and noncompliance with continuous positive airway pressure (CPAP). She underwent repair of the shoulder rotator cuff, and she had general anesthesia plus an interscalene block. She was discharged to home, and her son found her dead at 10:30 AM [the next morning] in her bedroom. She had a respiratory arrest and could not be resuscitated, sadly. And she had only taken 2 oxycodone, 5 mg each, the entire time at home.

We did a narrative literature review and found there are some significant risk factors for OIRD, such as elderly, female sex, presence of sleep apnea, chronic obstructive pulmonary disease (COPD), cardiac diseases, hypertension, diabetes, neurological diseases, renal diseases, morbid obesity, and if the patient has 2 or more comorbidities, as well as the use of patient-controlled analgesia (PCA), different routes of opioid administration, and a sedative being administered at the same time.

Recently, we did another systematic review and meta-analysis of 8600 citations, and we found 12 observational studies on OIRD. There were about 4000 patients who suffered from OIRD. So actually, the incidence of postoperative OIRD is very high. From this meta-analysis, it is 5 cases per 1000 anesthetics administered, and 85% of these cases occur within the first 24 hours postoperatively. And the morphine-equivalent daily dose of postoperative opioid is higher in those who had OIRD. If the patient had preexisting cardiac diseases, 70% higher risk; if the patient had pulmonary disease, 2-fold higher risk; and if the patient has sleep apnea, 40% higher risk. But there was no significant association between OIRD and age, gender, BMI, or American Society of Anesthesiologists (ASA) status in this meta-analysis.

However, in reviewing the literature on case reports and deaths in patients with sleep apnea in the perioperative period, we did find there's a clear dose-response pattern. If the patient had less than 10 mg, there is no increased risk of death. But if the patient had 10 to 25 mg, the risk is increased by 50%. If it is more than 25 mg of opioid, the risk is increased by 3-fold.

Dr McGrath: Frances, can you talk about the prediction of OIRD in patients monitored by capnography and oximetry in the PRODIGY trial?

Dr Chung: Oh yes. The PRODIGY trial was a prospective observational trial, and they blindly used capnography and oximetry in these patients. It was an international trial at 16 sites. A respiratory depression episode was defined as a respiratory rate of 5 breaths per minute or less, an oxygen saturation of 85% or less, and end tidal carbon dioxide of less than 15 mm Hg or more than 60 mm Hg for 3 minutes or longer. And then they did a risk prediction tool using a complicated multivariable logistic regression. So essentially, they found if the patient is less than 60 years old, there is no risk. They did a number to total this PRODIGY score, and if 60 to [less than] 70 years old, they assigned an 8; more than 70 to about 80 years old, 12 points; 80 years old or older, 16 points; male is assigned 8 points; opioid naive, 3 points; sleep apnea or disorders, 5 points; and chronic heart failure is 7 [points]. If the patient scored less than 8 [points], they are a low-risk patient. The medium-risk patient has a score of 8 to 14, and the high-risk patient has a PRODIGY score of 15 or greater.

Dr McGrath: Very interesting. So at Dartmouth Hitchcock we had a sentinel event, similar to the 2 that have been described by Kai and Frances, in the early 2000s. And that event led to some recognition that OIRD is a preventable situation, and our hospital made a commitment to eliminate unwitnessed arrests from our hospital, in total. This was a young patient admitted to a general care unit after having minor surgery. The patient had a history of anxiety and preexisting conditions. They were assessed, and it had been determined that they had little risk postoperatively of having an AE, and they received medication for anxiety and opioids for pain. And all of their initial postoperative assessments were normal. And the patient was found dead in bed by the nurse upon entering the room to do a vital signs assessment some hours later. So again, this is one of those situations that is, in our opinion, and I think in many others' opinion, a preventable event.

We made some observations based on this that includes the time factor of these events, that they happen within minutes and not hours, and so intermittent monitoring is not a highly accurate way of picking up on these events. That postoperative anesthesia risks are really underemphasized in relation to the potential impact, which here was death. That there was a lot more technology present in the general care setting, resulting in a lot more complexity in those environments, including use of PCA in the general care setting. These things create an opportunity for overload and, of course, alarm fatigue, which I will talk a little bit more about in just a moment. Nursing workload at the time had been increasing steadily, and there were fewer resources available in the general care setting to care for patients.

This situation has only gotten worse over the 15 years since we first had this observation and will continue to be a problem in this particular setting for the foreseeable future. Prevention of the circumstances leading to deterioration is not necessarily possible. And as Frances discussed and Kai mentioned, there are a lot of preexisting conditions and other circumstances that lead to deterioration. Rapid response teams, that are there to rescue patients, are only part of the solution. And part of the reason there is because it takes 6 to 8 minutes, on average, for those teams to reach the bedside once they've been notified, and patients can deteriorate much more quickly than that, often with poor outcomes.

And so what we decided to do was implement a continuous monitoring system in our general care setting, and there were a lot of design factors that we had to consider and trade-offs based on technology and organizational constraints that we had to figure out. And some of those included things like which patients to monitor, and we decided to monitor all of our inpatients because the evidence suggests that other approaches, including use of risk scores or selection based on procedures or medication delivery, would miss some of those patients. The sensitivity and specificity there are just not as high as we would like to get every patient to be safe. We also tried to figure out how often to monitor patients. In this case, we elected to monitor patients for their entire stay, for essentially the same reasons. Our goal was to completely eliminate unwitnessed arrests in this setting, and other options, such as intermittent monitoring, which doesn't account, again, for the speed with which these events occur, and time-limited monitoring, say for the first 24 hours postoperatively, still leave some patients exposed to the risk of respiratory depression and can result in patient harm.

We also considered how to monitor patients and what sensors to use. So we considered things like reliability, validity, and patient acceptance as factors in our selection of sensors. We looked at several options, including pulse oximetry, capnography, and bioimpedance. We looked at display of trends, how we would use statistical features, potentially, in notifying clinicians of deterioration, and how we would integrate this information into scores. Ultimately, we decided to use pulse oximetry, as we saw it as the best fit for our circumstances at the time that we were implementing the system. We still use it all these years later and find that it has very high patient acceptance. And we've had good results implementing with pulse oximetry, although we know other organizations have implemented using other technologies as well.

Another thing that we considered was how to enunciate and draw attention to patients who were deteriorating. We focused on task-tailored information for bedside nurses. They are the primary observers of patient state in this particular setting. So everything that we did was designed to help them and build something that they could see as being a tool for them to use at the bedside to detect deterioration. And we decided to focus on severe and sustained events, rather than transient events or events that weren't potentially leading to patient harm or death. This is a really important point with regard to alarm management.

As we know, alarm fatigue is a widely publicized and very impactful issue in medicine, especially in the medical/surgical setting, where event prevalence is low and resources are very limited to respond to alarms. And so these things have a big impact on how alarm systems should be designed in this particular setting. This point was emphasized in The Joint Commission's 2013 Sentinel Event Alert, where they noted that there were thousands of alarm signals occurring daily in a single hospital unit, and no intervention was needed for 85% to 99% of these signals. And so this and numerous other publications of safety events have demonstrated that clinicians can become desensitized to alarms rather easily, and it has very little to do with wanting to respond to alarms. A lot of this has to do with cognitive ability. And so again, special attention is required, and there have been numerous approaches proposed to address this particular problem, such as providing specific guidelines for alarms in different settings and tailoring settings to patients who need to be monitored.

In our case, we've developed what we call a surveillance approach and used very wide alarm settings. We tried to reduce noise using delayed enunciation and averaging of signals. We also use directed notification of nurses at the bedside, rather than broadcast alarms. And again, focus on detecting serious and sustained events, rather than events that aren't so serious. We have shown, in a recent publication, that this surveillance approach can be quite effective at reducing alarms in the general care setting. We were able to show that an application of an oxygen saturation (SpO2) threshold of 80% vs 88%, which is a typical SpO2 alarm setting in the critical care setting, produced a reduction of 88% in alarms. So 88% reduction in alarms, moving from 88% SpO2 down to 80%. The addition of a 15-second annunciation delay in SpO2 alarm reduced those alarms by an additional 71%, when using an alarm threshold of 80%. So this is an enormous difference in alarm sounds in this particular setting.

We also looked at the combined impact of alarms from different parameters and saw that the frequency for every 24 hours was about 4.2 alarms for SpO2 in pulse rate using our surveillance configuration, at 83 alarms for a typical critical care monitoring setting where SpO2 is set at 88%, and 320 alarms when looking at condition monitoring, which is a configuration where we set in our simulation's alarms at 90% for SpO2. So this, again, is a very large difference in alarms, and we have found that the settings that we use, what we call surveillance settings, have been very effective in reducing OIRD events.

We did a 10-year retrospective review after implementing the system. We conducted a systematic analysis looking at our inpatient data, followed by a chart review for individual patients, and looked for signs of patient death or permanent harm related to the administration of sedatives and analgesics in the general care setting. What we found in over 111,000 patients is that when monitoring was in place and used, we did not have a single death over this 10-year period from OIRD. We did have 1 patient death due to OIRD in a unit where surveillance monitoring was available. However, the patient wasn't being monitored at the time that the event occurred. In unmonitored patients, which was about 15,000 patients over the course of a 29-month period of incremental implementation of the system, we had 3 patients die from OIRD. So this is a really big difference. It's a 19-fold difference per 100,000 patients at risk, between those patients who were monitored and not monitored with respect to these OIRD events. This equates to a death rate, when surveillance monitoring was used or available vs not, of .0009% vs .02%, and that was a statistically significant finding. So moving on now. Kai, can you talk a little bit about your thoughts on the progress that could be made with the use of continuous monitoring in the postoperative period?

Dr Zacharowski: Yeah, thank you very much, Sue, and also for the great work you have done. Continuous monitoring, I think this is part of the solution. It is something which improves not only outcomes, but increases patient safety. And I have a clear opinion that I'm a huge fan of this. On the other side, no monitoring in the world can help if someone is not reacting to the alarm, and you have outlined this already. And it's not reacting to the alarm, switching it off, it's a little bit more. Doing something—and that could be waking up the patient or supplying extra oxygen.

So I have this sort of, almost a dream idea. I would suggest that when prescribing opioids, you should go along with a continuous monitoring solution during treatment. So you can't see one thing and forget the other one. So if you prescribe opioids, you want to reduce pain, suffering of a patient, then also you should prescribe some sort of monitoring, especially when patients are going home and out of the hospital frame, and that I would really urge. We need to have a change in thinking, not prescribing. We need to monitor what we are prescribing somehow, in some manner.

Dr McGrath: Are there any other things that the community might attend to or think about doing to address this problem of respiratory depression that hasn't been solved, that you can think of?

Dr Zacharowski: Yes. It's teaching, teaching, teaching, creating awareness, huge awareness. And I think if a patient is being operated on, we have to include the family, include the people who are looking after the patient, especially when the patient's at home. Awareness of the problem and to understand what to do if something happens. And this is probably something we have not considered enough, and we can do far more about this. And the second thing is we have to urge industry to invest more in this field. And in particular, I mean monitoring, but also personal devices could be of help, maybe like a smart watch, phone, whatever, where we could have additional monitoring options, and also warning on an automatic phone call, and someone on the other line helps them with advice. But I think there's lots of things which can be done in the future. However, first we have this huge problem, people dying, and now we start thinking. It's sad, but there's hope for the future, and hope for our future patients.

Dr McGrath: Great, those are excellent recommendations. I also think that there are some advances in monitoring technology that are going to be important, if those recommendations are going to be adopted for the prevention of OIRD events, particularly in the home environment. I know that the advent of COVID and the resulting impact on healthcare resources prompted a lot of innovation in remote monitoring systems that allowed patients to be monitored at home in the last few years. For instance, in the United Kingdom, the National Health System implemented a home monitoring program using intermittent pulse oximetry monitoring. That program demonstrated reduced mortality for those patients vs the mortality rate from the first wave of COVID. So, there was some progress made there with remote monitoring. In the US, University Hospitals in Cleveland and the University of Colorado Health also used continuous pulse oximetry with remote reviewing by clinicians, and so this study showed a reduced length of stay for patients who were discharged with pulse oximetry, with no increase in readmissions. And the study at University Hospitals in Cleveland showed predicted reductions in mortality and a cost savings overall for treating COVID patients [with remote monitoring]. I also know that some of these advances that were made during the COVID period have made their way into OIRD-specific monitoring systems. There's a system that's available that's very similar to the approach we used in the inpatient setting with our SafetyNet that combines continuous monitoring with tailored alarms and notification escalations for specific conditions, including OIRD. I think that that system's been improved for use in Europe, and again, it's very similar to what we've done in the inpatient setting that we've had very good success with. The pairing of tailored alarm management with care management models, providing advice or assistance from a medical professional in the home setting will be really important to implementing the recommendations that were just mentioned.

So Frances, can you talk a little bit about the location and timing of critical postoperative events in patients with obstructive sleep apnea (OSA)?

Dr Chung: The Society of Anesthesia and Sleep Medicine and the American Society of Anesthesiologists established a confidential death or near-death registry for sleep apnea a few years ago. And over 3 years, there were 66 cases of death or near death, all critical events that were reported from multiple hospitals across the USA. Known sleep apnea was diagnosed in 86% of these patients, and 17% were unrecognized sleep apnea. And 96% of these patients received opioids within 24 hours before the critical event; 66% also received a sedative. Nine percent of death or near death occurred in the postanesthesia care unit (PACU); 14% occurred in the step-down unit or intensive care unit (ICU). In the ward, the incidence was 56%, and surprisingly, 21% of these death or near-death [events] occurred at home.

How about the timing of these events? In the PACU, the death or critical event occurred 2 hours after anesthesia. In the step-down unit and ICU, it occurred 13 hours after. In the surgical ward, the event occurred at 16 hours, and at home, it occurred 27 hours after anesthesia. So death or severe brain damage was more likely to occur in the surgical ward and at home, and a critical event was more likely to occur in the PACU, the ICU, or step-down unit. When patients were monitored at the time of the event, they were 6 times more likely to have critical events than death or near-death events. So monitoring at the time of the event would result in patients having critical events, and they can be resuscitated and may be okay. So knowledge of the timing of the respiratory event is very important. And actually, the medium time to the initial OIRD episode was about 9 hours postoperatively. The peak occurrence of the first respiratory depression event usually occurred between 2:00 PM and 8:00 PM on the day of surgery. However, the peak time of all respiratory depression events occurred from 2:00 AM to 6:00 AM in the early morning.

Dr McGrath: That distribution of events is very interesting. I think that you've also done some post hoc analysis looking at home oximetry to identify sleep apnea patients taking opioids. Can you comment on that, please?

Dr Chung: Yes. We did a post hoc analysis looking at home oximetry to identify sleep apnea in patients taking opioid analgesics. And we found the predictive performance of the oxygen desaturation index from oximetry and evaluated that against the apnea hypopnea index from polysomnography, which is sleep study. And the results were excellent. We showed that the area under the receiver operating curve (AUC) for the oxygen desaturation index to predict moderate to severe sleep apnea was about 82%. For severe sleep apnea, the AUC curve was 87%. So essentially, we were able to conclude that at-home oximetry was very useful to identify sleep apnea patients, especially those who are unrecognized or undiagnosed sleep apnea patients, because these patients may be at greater risk to opioids when they're discharged home together with the opioid for their pain.

Dr McGrath: Yeah. And equally important in the inpatient setting because we know that many patients who have these events have undiagnosed OSA. So being able to diagnose at home is very important. And with the greater and faster discharge of patients on opioids to home settings, this work is really impactful. Kai, can you talk a bit about some of the guidelines and tools that are available now to address OIRD?

Dr Zacharowski: Sure. There are obviously some scales for assessing sedation available. For example, the oversedation risk criteria, which assigns points to predictors that cannot be modified (eg, patient characteristics) and those that can, like treatment choices, and it takes them into account—multiple risk factors. It stratifies patients as low, moderate, or high risk for OIRD and, obviously, oversedation. There's another scale, which is the Pasero Opioid-Induced Sedation Scale, which is designed to monitor sedation during opioid administration, and it classifies patients’ alertness vs drowsiness.

And there are also guideline recommendations. And this is interesting, because there's a shift toward opioid-sparing and multimodal analgesia intraoperatively. And also, pain relief should be optimized in the PACU, and immediate release opioids are preferred for managing postoperative pain. And it says to avoid use of benzodiazepines and gabapentinoids with opioids and to educate patients. This is what I always say, educate patients on the safe use of analgesics and appropriate disposal of unused medications. And we cannot say it often enough: do not use alcohol while taking opioids.

In 2018, The Joint Commission released updated pain assessment and management standards, which basically covers: facilitates practitioner and pharmacist access to the Prescription Drug Monitoring Program databases, and works with clinical staff to identify and acquire treatment needed to monitor patients at high risk for AEs from opioid treatment. Also, has criteria to screen, assess, and reassess pain which are consistent with the patient's age, condition, and ability to understand. And develops pain treatment plans, based on evidence-based practices and the patient’s clinical condition, past medical history, and obviously pain management goals. And finally, monitors patients identified as being high risk for AEs related to opioid treatment. And, what I said earlier, educates the patient and family on discharge plans related to pain management.

Dr McGrath: Great, thank you for sort of closing the loop there. We've talked about research today. We've talked about technology implementation. We've talked about guidelines for opioid administration. So just to summarize, our aim today was to create awareness of OIRD and to educate patients and medical professionals regarding this condition, its causes, and guidelines to address it. We also discussed multiple tactics that are available to reduce OIRD events, and these can be used in both the inpatient and home setting. And again, with trends in healthcare, we think that managing this condition in the home setting is going to be an ever more important thing to address.

So this discussion also emphasized the need to use a systems approach to design patient safety systems, such as OIRD event prevention. This is really evident, as we can see that many of these tactics by themselves—guidelines, medication administration information, technical pieces, scoring, monitoring systems—all need to be integrated into a system that includes education, as Kai was kind enough to remind us of many times, of the patients, the staff, and medical professionals about this and how these various components can work together to prevent these events from happening.

I'd like to thank Frances and Kai for joining us today and for this great discussion. And thank you for participating in this activity. Please continue on to answer the questions that follow, and complete the evaluation.

This transcript has not been copyedited.