Postoperative delirium assessed by post anesthesia care unit staff utilizing the Nursing Delirium Screening Scale: a prospective observational study of 1000 patients in a single Swiss institution

In this study, the presence of delirium in patients who were recovered in a PACU after a non-cardiac surgical intervention was prospectively studied. The instrument utilized for the detection of delirium was the Nursing Delirium Screening Scale (NU-DESC; with a delirium defined as a score of ≥ 2 points). NU-DESC was conducted by the nursing staff in the PACU prior to the transfer of the patient to the surgical inpatient ward.

The most prominent result of the study was that delirium assessment using the NU-DESC test was very easily feasible in this setting. There was a relatively low incidence of delirium with only 4.3 %, compared to some previously cited rates [2, 13, 17]. Even when just considering the at-risk subgroups, the rates of delirium remain low. There are a number of possible reasons for these findings, which are discussed below.

The delirium that occurs in the hospital is still a poorly understood phenomenon. In recent years, however, awareness of the problem has increased and the research surrounding delirium was intensified [2, 3, 10, 12]. As a consequence, many risk factors that can lead to delirium have recently become known [7, 9, 10, 18]. Elderly patients undergoing major surgery are particularly at risk [7, 9, 10, 18, 19]. It also appears justified to distinguish between cardiac surgery and non-cardiac surgery, since the former is associated with a greater risk for both the incidence of delirium as well as the development of long term brain dysfunctions [1, 10, 20, 21]. Unfortunately, delirium is not always a reversible condition. It can be associated with serious long-term consequences, which affect not only cognitive performance, but also may result in prolonged stay in the hospital and intensive care unit, higher costs, and even increased rates of complications and mortality [1, 4, 5, 21–25].

In our study, we found a relatively low incidence of delirium. The most obvious reason for that seems to stem from our patient population. The incidence of delirium can reach rather high incidences in specific sub-groups. In certain patient populations in intensive care units it can reach levels of up to 80 % [26]. Elderly orthopedic trauma patients have been reported to have a frequency of delirium of almost 50 % in the PACU [19, 27]. In contrast to these sub-groups of at-risk patients, our patient population was rather young and healthy, as is reflected in their low ASA physical status scores. Radtke and colleagues [13] reported an incidence of 25 % for delirium in their recovery room using the NU-DESC. In their series, only 75 % of all patients had an ASA physical status of either I or II. In contrast, our series included 90 % of patients with an ASA physical status of I or II, likely reflecting the difference in patient population between a tertiary care hospital (Charité, Berlin, Germany) and our institution. Interestingly, in a recent study by Card [12] et al. a very similar rate of delirium to our study was reported using the CAM-ICU test, which may be more specific but less sensitive than the NU-DESC in a PACU [15].

Another difference between our work and the mentioned studies using the NU-DESC score lies in the fact that in our study the assessment was done by regular nursing staff of the post anesthesia care unit rather than specifically trained psychologists, psychiatrists, or study nurses [13]. Especially item 5 of the NU-DESC data collection sheet (Table 1) asks for a comparison between the given patient and a reasonable, adequately acting normal person. PACU nursing staff may be more likely to gauge someone who is not yet fully oriented or still slightly sedated postoperatively as normal, than an independent investigator who is not familiar with this postoperative setting. Accordingly Haenggi and colleagues noted that the exact assessment and documentation of the patient’s sedation scale should be a key component when studying postoperative delirium [11]. In a study with patients in an intensive care unit, 53 % of the patients were found to be delirious when screened only with the CAM-ICU score. This number was significantly reduced to 31 % when the more sedated patients (Richmond Agitation Sedation Scale, RASS score of −2 or −3) were excluded [11].

Even if not quantified systematically, the acceptance of the additional task of delirium screening by the team staff was high, and with the test being in fact performed in about one minute, on average, the additional burden was not a topic.

The overall utility of the NU-DESC can also be questioned. There is the notion that early detection of delirium could lead to more effective treatment and the reduction of the negative consequences,[14, 28, 29] despite there being no uniform consensus on the exact treatment modalities [2]. The detection of delirium remains challenging because of the variety of manifestations: a clear confusional state with agitation on one end of the spectrum, and a hypoactive state on the other [12]. Accordingly, the diagnosis should not be supported on an individual estimation, but must be assessed using an appropriate assay [13, 17, 26, 28, 30]. The NU-DESC has been validated against more extensive tests and impresses mainly with its simple implementation, making it well suited for routine use in a PACU [13, 16]. Applying the NU-DESC to a given patient requires not more than 1 min [16]. However, there are some inconsistent data reported about sensitivity and specificity of the test [13, 17]. A recent review of different delirium bedside screening tools characterized the NU-DESC as highly sensitive (>95 %) but considerably less specific (>70 %) [15]. Taking this into account, it is difficult to predict in what manner using the NU-DESC test could induce bias regarding the incidence of delirium, when compared to other tests.

It would be tempting, yet unrealistic to assume that our anesthetic management contributed to the relatively low incidence of delirium in our patient population. However, in contrast to our standard, i.e. maintaining anesthesia using Propofol, Radtke and colleagues [13] chose volatile anesthetics for over half of their patients, and a study by Neufeld and colleagues [17] used volatile anesthetics in over 90 %,. Since the anesthetic management for our patients did not follow a given study protocol, we abstain from further speculation and a more detailed analysis at this point. The same limitation stands true for the role of pharmacological premedication with benzodiazepines given to our patients prior to anesthesia, which was not controlled as would have been under study conditions.

In addition, the duration of stay in the PACU and thus the time between the end of anesthesia and the assessment of the NU-DESC score may play a roll. This time was almost twice as long in our study, compared to that of Neufeld et al. who did not assess their patients prior to transfer to the surgical inpatient ward; they rather did so when reaching an Aldrete score of 9 points, which was the case after about 45 min [17]. When considering the initial recovery period after general anesthesia as a form of delirium, it is not surprising that the incidence of delirium decreases with increasing time after the anesthesia [11].

The NU-DESC score is potentially too simple to be used as a widespread screening tool in post anesthesia care units, especially when dealing with patients that are slow in their reactions. The clinical benefit of the NU-DESC could be increased by virtue of more intense training of the PACU nurses in order to achieve a more uniform assessment. It has been described for intensive care units, that regular feedback on the incidence of delirium, completeness of patient evaluations, and occasional reference comparison to the assessment of an uninvolved psychiatrist helped the successful implementation of the screening for delirium [28, 31]. The same approach could be supportive for PACUs, but more studies in the immediate postoperative setting are warranted. At the same time a longitudinal extension of the study that would investigate the course of patients with delirium after their transfer to the surgical inpatient ward could contribute further valuable insight into the disease. In fact, the relevance of PACU delirium screening will have to be examined in the context of its actual association with adverse outcomes in future trials.

A limitation to our study is that the timing of the delirium screening is somewhat vague. Choosing fixed time intervals from admission to the PACU, and probably adding repeated assessments would have made our data more objective and reproducible. However, our study was thought to be only a first step in a broad implementation of delirium screening in postoperative patients. Hence, the feasibility of the test and estimating the proportion of patients with a need for follow-up was our first priority.

We did not attempt to verify known predictors for developing delirium postoperatively. It also must be mentioned that the comparison of patients with delirium and the patients without delirium lacks an appropriate power analysis.