Dr Cochran: Hello, this is Amalia Cochran, the web and social media editor for JAMA Surgery, and I am speaking today on the topic of effective delirium prevention in older persons after elective surgery with Drs Christine Thomas --
Dr Cochran: -- and Alba Sanchez.
Dr Cochran: Welcome to both of you. I always start my interviews asking why a question was of particular interest to the researchers. Can you tell me more about why you and your collaborator group became interested in further studying postoperative delirium in the elderly?
Dr Thomas: Well, the collaborator group is a group of psychiatrists and also geriatricians who often meet the delirium in their consultation liaison services, and the outcomes are often very problematic in old age. So, patients lose competence and life quality. That's why delirium has become of increasing interest to us. As delirium is preventable, it was very important for us to study a group of patients that we can assess before they really suffer a delirium. So, we decided to study elective surgery patients, as this is an ideal situation to assess delirium incidence and duration as well as patient individual risk factors before surgery.
Dr Cochran: On behalf of our readers and/or our listeners who are not familiar with the German healthcare system, how did you select the university hospitals and tertiary care centers that were included as sites for this study? Were there major centers who declined to participate or who did not meet your criteria?
Dr Thomas: Yes, actually, we started off having a collaboration between me, as a neurologist geriatric psychiatrist, who had moved and got a new position in southern Germany, and a colleague, a geriatrician who is very much involved in the state geriatric medicine network, and the state finances this geriatric network, which consists of 15 major centers and 10 sub-centers. So, we got this whole group together and discussed how this study could be conducted in southern Germany, and at that point, it became clear that some centers did not have the according structure to be involved in the study. Most centers that declined had problems getting the personnel that could do the training and teaching and recruiting of patients and collecting the data, and some centers also did not have the high amount of longer surgeries in old and very old patients. So, it was those three reasons. So, for instance, one big university hospital in Heidelberg, for instance, did not have a structure where the geriatricians were at the hospitals together with the surgeons. The geriatricians had a different facility outside of the big university hospital. That was one reason not to be eligible. Another reason was that two other major centers just did not find the opportunity to get enough staff to do the research, even if the money was provided by our grant.
Dr Cochran: So, all very practical and pragmatic reasons for this selection of sites, it sounds like. I always like to give our listeners the opportunity to learn more about study design, particularly for prospective trials like yours. Can you share with us why you chose a stepped-wedge cluster design for this study? What benefits were there for this design?
Dr Sanchez: The stepped-wedged cluster randomized control trial design is being used with increasing frequency, especially in the area of health-related research, and this design has a number of advantages. For example, this approach and intervention is introduced over a number of time periods with the order in which each cluster begins to receive an intervention is determined at random. So, this randomized order of introduction of the intervention which perceived as fair. And so, in the classical form of the design, all clusters begin in the control condition and will receive the intervention by the completion of the trial. All clusters are actively taking part in the trial at this time period in the intervention will be made available for all the clusters by the end of the trial. So, this design provides intervention to all clusters, and intervention is never removed once introduced. Avoiding these controversial situations in which control will have no intervention, and that will be a benefit from an ethical point of view. Sometimes, it's not possible to implement intervention in half of the cluster simultaneously because of radical reason, for example, this was the case in our study, or because of financial reasons.
In these cases, the stepped-wedge design offers a solution and is not necessary to enter all the clusters at the same time. Also, the recruitment of clusters might be easier within this design, because every cluster will receive the treatment within the trial. Another strength of the design is that the clusters act as their own controls, because they receive both the control and the intervention conditions. Thus, the intervention effect can be estimated from both between clusters and within cluster comparison. But, of course, this design has its own challenges that should be considered before deciding whether it is the best option. For example, this design can require repetitive training activities in the different centers of clusters and an increase in the workloads for intervention teams over time, as more clusters start the intervention. Although, in our study, it was more an advantage, as it was not possible for us to introduce intervention in all centers at the same time. Also, a sustained engagement with clusters or centers that are in the control groups will be necessary to avoid the dropouts.
One issue could be also the duration of the trial that will be longer than in a parallel design trial if steps have minimum length due to the need to wait for intervention to take effect. And as we know, longer duration trials are more susceptible to contamination due to external events. Another consideration is the potential for loss of power at a center with a delayed intervention effect. Such delays can occur to the lack between the introduction of intervention and its effect on the outcomes. And, finally, this design only works when the same number of patients are recruited in each step. For example, we had to recruit 50 patients in even intervals, but we have difficulties for recruiting some of the patients in some periods. For example, in the first three months of the study. A previous study that uses this type of design reported that some clusters experienced major difficulties in including patients leading to excluding some clusters from participation, and that resulted in failing to include similar numbers of patients in intervention and control conditions. But, in general, when well conducted stepped-wedge designs are taken evidence is of higher quality than results obtained from observational or nonrandomized studies, and, in my opinion, despite its limitations, the stepped wedge design is the ideal design for our situation where we want to have interventions in each center, and we could not introduce interventions at the same time.
Dr Cochran: That makes a lot of sense that this was an ideal design for your study in particular. Prospective trials like this almost inevitably encounter some form of challenge while they are ongoing, and it sounds like you already described some of the challenges with patient enrollment that occurred early on. Were there other challenges with enrollment or conducting the study that came up for you, and in contrast to that, was there a part of the work that was easier than expected?
Dr Thomas: At first, it was rather difficult to recruit the patients early in the patient's process through the whole surgical procedures. We had thought that it might be ideal to contact the patients maybe about a fortnight before the surgical date, but that was oftentimes not possible. It was just not possible to get the connection early on as the surgeons often did not inform us early enough. Also, we had a lot of data security problems, as we were not allowed to see the OR schedules to review them for eligible patients. So, those practical things were really difficult. Also, I have to tell you that in the German healthcare system, all of the expenses are covered. So, the patients do not have a financial interest in taking part in a study. So, it's nothing we can offer them apart from some little gift or something to be a part of our study. It's their altruistic approach that brings them to be part of a study. We are not allowed to pay them money for ethical reasons. In the US, that's different, because oftentimes patients have some cost that they have to pay out of pocket, and in research environments, often these costs are then covered by the research facility.
Dr Sanchez: No, for example, another challenge was that at that at 2 months follow-up (t10), some patients were in rehabilitation at that time, so we couldn't reach them, and when the t10 assessment was missing, we tried to do the best to get them back into the follow-up assessment at the t11. Some doctors could not be reached in the follow ups, and in general the collaboration with the doctors didn't work as we had thought. So, as I introduced in the previous question, in the first interval, the first three months when more of the centers were in the control group, there were some problems related with the recruitment numbers, and it was necessary to have two steps. We needed to recruit 50 patients per center in each period, but one of the centers failed to reach their required number of patients in the first interval, and to solve this problem, we decided to widen the range a bit, being possible to include in each period 5 patients above the range (55), or below the range (45).
Dr Cochran: So, I would love it if you all would share with me one or two findings that perhaps surprised you when you started to look at the data as it was analyzed.
Dr Sanchez: I think that the first surprise for me was when we calculated the delirium incideces rate at each center, and we realized that there was too much difference between the centers, and this difference will be related to the differences in the type of surgeries in each center. Then, we examined the incidence rates, and we saw that intervention worked for patients that are undergoing noncardiac surgery but not for patients undergoing cardiac surgery. So, we realized that the incident rates in patients undergoing noncardiac surgery were significantly lower in the intervention group comparing with the control group, but these results were not present in the cardiac surgery group, and we confirmed this in the main analysis when we stratified the sample into cardiac and noncardiac surgery in the model 3 in the GEE analysis. And we saw that intervention on delirium were unrelated in the patients undergoing cardiac surgery, but for patients undergoing noncardiac surgery, the intervention was significantly inversely associated with delirium.
Dr Thomas: I was really astonished that the variability of the delirium incidence was so high within the centers but also within the different steps or periods, so this was very interesting to us. Also, it was very interesting to find that the amount of patients that had special risk factors, for instance, being frail or being cognitively impaired, was higher than we had thought. That's also very interesting. And that was very much in contrast to the numbers of diseases that we had asked them to give us beforehand as a medical history. Oftentimes, they did not name, for instance, dementia as one of their diagnoses, but when we tested them, they were cognitively as impaired as a demented patient.
Dr Cochran: Well, and this was something that I was particularly interested in. So, I'm gratified to hear that you were as fascinated as I was, apparently, that the level of baseline cognitive frailty that you identified in the patient population, which was approximately 75% of the study group patients, I believe, seems really high, versus what I want to expect it is in the general population. Is this because we're under-evaluating this in the general population, or am I simply too optimistic about the youthfulness of people over the age of 70?
Dr Thomas: Well, I think it's a very special situation, as Dr. Sanchez already has pointed out, we had to assess those patients mainly on the day before their scheduled operation, and this was a very critical time for them. They just had a change in environment, so they had been admitted to the hospital. Often, they were in pain. If you think of a patient that needs replacement of a hip, well, of an arthritic hip, for instance, so they were in pain. They had pain medication at high levels, and all these factors are of high importance and reduced cognitive reserve that the patients that are older than 70 years normally have. We only studied patients that are older than 70 years. They just have not as many reserves as younger patients, so they more easily get cognitively impaired because of their surroundings. Being a psychiatrist, I also have to claim that, as they await their surgery the next morning, there's also a kind of fear that they will have to cope with, which will also reduce the cognitive reserve. So, that's the reason why I think the high amount of patients with cognitive impairment can be explained. However, we have to say that we used the MoCA, the Montreal Cognitive Assessment, and we did not use the Mini-Mental State Examination. The MoCA is more, well, it has more domains that are tested, and it's a more difficult test. So, this is also, so, we did apply a more correct test, I think, on those patients. We were, of course, very interested in changes in their cognitive abilities. That's why we needed a more differentiated test rather than the Mini-Mental State Examination that is not as differentiated, and the cutoff is lower.
Dr Sanchez: Yes, and just to add that is true that the percentage of people with cognitive impairment is high for the reason that Dr. Thomas has said, but also, only around 2% of the patients have the diagnosis of dementia according to the Charlson Comorbidity Index, and the median for the MoCA score was two to four for both groups at baseline, which would be only a slightly lower than is usually found in the general elderly population.
Dr Thomas: However, in Germany, there was a big study in 2016, the General Hospital Study was its name, and they assessed a point prevalence of cognitive impairment in patients that are admitted to the hospital over 70 years of age, and at least they did not all assess surgical patients but also internal medicine or urology or dermatology patients, and they found an amount of 40% of patients over the age of 70 years to be cognitively impaired. So, I think it's a a valid number.
Dr Cochran: When we were talking about surprises in the findings, you mentioned the difference that initially showed up and then subsequently played out with the cardiac surgery population versus the elective general surgery or elective orthopedic surgery population. Do you have any thoughts on why the interventions may not have been as effective in the cardiac surgery population?
Dr Thomas: Well, our intervention is a nonpharmacological intervention, and we made a lot of effort to have an optimal intervention with optimal modules. However, all of them were nonpharmacological, and they were not applied during the surgery, so what we think is that it was just not enough for those cardiac patients to have those nonpharmacological interventions, and I think we need more interventions in the perioperative phase. We do know that those patients have more changes in the blood-brain barrier. We do know that they are highly vulnerable regarding changes in their blood pressure, regarding anemia, regarding the depth of the anesthesia. So, I think we have to control these parameters, too. That's a high challenge, also, because it's very difficult to do this in a multi-center situation, as there are always very specific approaches in the surgeons, so they want to, of course, we have a state-of-the-art, but also, we have different opinions on different things. What you need and how deep the anesthesia has to be to have the best situation, so it's very difficult to get them all on the same protocol level. So, this is something we have to approach in the future. There is no study by now that really can prove that nonpharmacological interventions can prevent delirium in cardiac surgery. There is some evidence that dexmedetomidine, for instance, used on ICUs rather than other sedating medication and pain medication is of advantage for these patients.
Dr Cochran: Great. Were there any notable ways in which the excluded patient population who did meet enrollment criteria differed from the included patients?
Dr Sanchez: There were 840 patients that we contacted, and they have not given the consent to participate in the study, but since they have not given their consent, we do not have the data to analyze them. As Dr. Thomas explained before, German health system covers the cost of the surgery and the hospital stays, and in Germany the interest to be in the study, it's much less than it can be in the USA, because the patients don't have this financial incentive. And then, there was some patients that met all inclusion criteria but were located at a distance to the center which is about 120 kilometers and did not have advanced dementia, and these patients were recruited for the PAWEL-R study that investigated clinical, psychosocial, biological risk factors of delirium. And those patients have baseline characteristics similar to those of the main part of the study in terms of age, gender, education, and presented comorbidities. Only the percent of baseline cognitive impairment was slightly lower since the patients with advanced dementia were not included.
Dr Cochran: One of the pressing clinical questions that occurred to me as a surgeon that works in a specialty that is more
emergency based, is if we should consider using similar nonpharmacologic interventions that is those that you used for the study, in our patients who are having urgent or emerging surgery?
Dr Thomas: Certainly, yes. I'm very sure that our intervention including the education of the ward staff and introducing an additional delirium expert nurse and the delirium prevention team was the model that we provided will be very helpful and will be able to prevent delirium in patients with emergency surgery, as well, at least as well as they do for elective surgery patients. We all know that patients with an emergency surgery have a much higher risk of being delirious, and some of them are already delirious before the surgery even has started.
Dr Cochran: And then my closing question for both of you is, do you have plans for future extension of this important work?
Dr Thomas: Yes, we do. There are two things that we are working on at the moment, apart from of course having a lot of data still left to evaluate. First thing, we want to stretch out our study on emergency surgery cases but also on huge internal medicine situations like pneumonia or urinary tract infections with delirium in old age and things like that, and, also, we want to stretch it out to other regions in Germany, three very different regions where we also have smaller hospitals and rural areas in it, and we already applied for a grant also from our nationwide grant service there to see how these patients will profit from our approach. Moreover, we changed our educational approach now, as we got more experience with online learning, because this took a lot time to teach everybody personally. So, we now started to use e-learning elements in our education, which seems to be an important way to get this into a regular assignment of hospitals.
Dr Cochran: Dr. Christine Thomas and Dr. Alba Sanchez, I would like to thank both of you for taking time to speak with me today about effective delirium prevention for older persons after elective surgery.
Dr Thomas: Well, thanks to you for your interest.
Dr Sanchez: Thank you very much, yes.
Dr Cochran: This episode was produced by Daniel Morrow at the JAMA Network. The audio team here also includes Jesse McQuarters, Shelly Steffens, Lisa Hardin, Audrey Forman, and Mary Lynn Ferkaluk. Robert Golub is the executive deputy editor for digital media. To follow this and other JAMA Network podcasts, please visit us online at JAMAnetworkaudio.com. Thank you for listening.