Sedation, analgesia, and delirium management in Portugal: a survey and point prevalence study

Objective To establish current Portuguese critical care practices regarding analgesia, sedation, and delirium based on a comparison between the activities reported and daily clinical practice. Methods A national survey was conducted among physicians invited to report their practice toward analgesia, sedation, and delirium in intensive care units. A point prevalence study was performed to analyze daily practices. Results A total of 117 physicians answered the survey, and 192 patients were included in the point prevalence study. Survey and point prevalence studies reflect a high sedation assessment (92%; 88.5%), with the Richmond Agitated Sedation Scale being the most reported and used scale (41.7%; 58.2%) and propofol being the most reported and used medication (91.4%; 58.6%). Midazolam prescribing was reported by 68.4% of responders, but a point prevalence study revealed a use of 27.6%. Although 46.4% of responders reported oversedation, this was actually documented in 32% of the patients. The survey reports the daily assessment of pain (92%) using standardized scales (71%). The same was identified in the point prevalence study, with 91.1% of analgesia assessment mainly with the Behavioral Pain Scale. In the survey, opioids were reported as the first analgesic. In clinical practice, acetaminophen was the first option (34.6%), followed by opioids. Delirium assessment was reported by 70% of physicians but was performed in less than 10% of the patients. Conclusion The results from the survey did not accurately reflect the common practices in Portuguese intensive care units, as reported in the point prevalence study. Efforts should be made specifically to avoid oversedation and to promote delirium assessment.

To improve our practice, it is important to understand our current practices so that we can identify the best targets for improvement. Other surveys conducted in different countries analyzed their own specific practices concerning analgesia, sedation and delirium (ASD) in critically ill patients. (9)(10)(11)(12)(13)(14)(15)(16) All of them gave information about explicit targets for quality improvement in each specific country.
In the United Kingdom (17) and Germany, (18) studies with designs similar to ours were conducted, with two different evaluations. They compared the physicians' analgesia, sedation and delirium practices reported in a survey with a subsequent daily clinical practice analysis. In both countries, the information reported in the survey did not accurately reflect the clinical practice observed.
The aim of this study was to characterize analgesia, sedation, and delirium practices in Portuguese ICUs, particularly the adherence to international recommendations, provide specific targets for improvement, and establish the priority for further research and national recommendations.

National survey
We conducted a MEDLINE search using the keywords "analgesia", "sedation", "delirium" and "ICU" to identify the most important aspects in the literature regarding this area. The survey had five parts: 1) professional profile and ICU characterization (8 questions); 2) generic information about guideline application and follow-up (4 questions); 3) sedation practices (9 questions); 4) analgesia practices (4 questions); and 5) delirium, rehabilitation/mobilization and sleep improvement (17 questions) (Appendix 1S -Supplementary material). The questions referred to participants' perceptions of their ICU daily practice. The format of the questionnaire was tick boxes and blank spaces for completion, with an average duration of 15 minutes, and questions were not mandatory. To ensure that the survey was robust, sensitive, and reliable, it was presented to the GASD (Analgesia, Sedation and Delirium Study Group). This group is a critical care research group that includes intensivists and ICU trainees who commented on, discussed and approved the final version.
Between 1 September 2016 and 30 April 2017, the survey was distributed through a link sent by e-mail to all ICU physicians registered in the Portuguese Intensive Care Society, irrespective of their working place being the National Health Service or a private setting. Neonatal and pediatric ICUs were excluded. The survey was designed using a web-based provider (MedQuest) accompanied by a cover letter, which informed the respondent of the details required for its completion.
The survey did not contain any data that could identify the respondents. Participation in the survey was anonymous, voluntary, and noninterventional.
To reinforce participation, the survey link was sent two times during this period. There was a limitation in the survey response; only the professionals who had not responded to the first link could open and fill the second survey link that was sent (to avoid duplicate answers).

Point prevalence study
A point prevalence study (PPS) was performed on the 26 th of January 2018 in adult ICUs. We excluded high dependency units and cardiac and cardiothoracic ICUs. Invitations were sent to the directors of 20 ICUs in Portugal from north to south, and 17 agreed to participate. The study was approved by the Research and Ethics Committee of Centro Hospitalar de Lisboa Ocidental. Each hospital received ethics approval from its local Institutional Review Board and was granted a waiver of informed consent for this observational, minimal risk study. No incentives were offered to participants. No financial rewards were granted to participating centers.
The participating ICU team was asked to complete a short data collection form (Appendix 2S -Supplementary material) for each patient (≥ 18 years) in the ICU between 00:00 and 24:00h on the 26 th of January 2018. The exclusion criteria were death during the study period and withholding or withdrawing life-sustaining treatment decisions.
In each ICU, there was one physician dedicated to conducting the study, checking the eligibility, and collecting all the data during the study day (24 hours). Most of the data were collected directly from the patients' records; there was no intervention in clinical practice, no additional assessments, and no changes in the normal routine in the ICU. The collection form was returned on paper. All data were entered into a dedicated electronic database exclusively created for this study and managed by the authors who ensured its confidentiality. Patients were given a code number to secure their identification.

Data and statistical analysis
Standard descriptive statistics were used as appropriate, and variables were reported as numbers (%). Continuous variables are expressed as the mean ± standard deviation, and categorical variables are expressed as n (%). As the number of respondents varied across the questions, with some missing answers, the proportions displayed in the results section and tables were not constant. Statistical analysis was performed using IBM SPSS Statistics v26.0 (IBM, Somers, NY, USA).

National survey
The overall response rate was 28% (117/418 physicians). The majority of respondents (91.2%; 103/113) worked exclusively in the ICU and specialized in intensive care medicine (68%; 78/114), with a mean age of 47 years and a mean of 12 years of ICU practice. The ICUs had a mean of 13 beds (5 beds level 2 and 10.7 beds level 3) and were predominantly distributed in the north and center of the country (84%), and most of them had a mixed medicalsurgical patient case mix (89.8%) ( Table 1). The existence of protocols for ASD was considered useful by 95% (104/110) of the physicians, but less than 50% referred to having such protocols in their ICUs ( Table 2).

Point prevalence study
Twenty ICUs were invited to participate in the PPS. A total of 17 ICUs were included, from the North, Center and South of Portugal, with a total of 189 beds. We included 192 patients (37.5% females and 62.5% males). The number of patients was higher than the number of beds available since, in some cases, during the study duration, a patient was discharged from the ICU and another was admitted to the same bed ( Table 3).
Most of the ICUs were mixed, just one was specific for neurocritically ill patients, and the main type of admission was medical (61.5%). The mean Acute Physiology and Chronic Health Evaluation (APACHE) II and Simplified Acute Physiology Score (SAPS) II scores at admission were 19 and 42, respectively. The Sequential Organ Failure Assessment (SOFA) score at admission was 7.3 and on the study day was 5.9. Fifty-six percent of the patients were under invasive mechanical ventilation. There were some missing values in the case report form that are mentioned in the results presentation (Table 3).
A sedation assessment was performed in 88.5% of patients (170/192), at a minimum of 1 time per day and a maximum of 17 times/per day. The most commonly used scale was the RASS (58.2%; 99/170). The Glasgow coma scale (GCS), (26) was used alone in 18.2% (31/170) and in association with the RASS scale in 15.9% (27/170) ( Table 5).
To assess the adequacy of sedation (oversedation and undersedation), we compared the sedation target for a specific day with the sedation presented by patients. Figure 3 shows the variation in sedation level, in RASS points, between the target and the measured RASS.   Results presented as n (%).*Presented as the mean ± standard deviation, only if present in more than 10 patients. In 147 patients, 57% (n = 84) achieved the sedation target desired for that day, 32% (47 patients) were oversedated compared with the targets, and 11% (16 patients) were undersedated.
All ICUs used pain scales, and furthermore, some applied more than one scale in the same patient ( Table 5). The BPS (19) was the most used scale (45.7%; 80/175), followed by the Numeric Rating Scale (NRS) (27) (
Concerning the 8.3% that were assessed for delirium, the method used had a significant percentage of missing values (91.7%). Assessment with CAM-ICU was reported only in 5 patients. Subsyndromal delirium was not assessed in the ICUs. Physical restraints were used in 18 patients (9.4%).

DISCUSSION
The clinical practice reported in the survey did not accurately reflect the clinical practice reported by the PPS. Oversedation both identified in the survey and the PPS, is still present in Portuguese ICUs, with a high percentage of benzodiazepine use. Delirium and subsyndromal delirium daily assessment with validated scales was not performed even though the survey reported delirium assessment in more than 70%.
Analgesia with opioids was the first reported choice, but acetaminophen was the leading drug, as described in the PPS.
Daily sedation monitoring, a fundamental approach for sedation management, is a common practice in Portuguese ICUs, reported in the survey (92%) and confirmed by the PPS (88.5%). Despite this high percentage, there is still space for improvement, and an effort should be made for all patients to have access to sedation monitoring as a means to avoid oversedation and related complications. The RASS scale was the most frequently used and reported sedation scale, perhaps because of its ease of use, with positive scores for agitation and negative scores for sedation ( Figure 4).
The GCS was reported in the survey as part of the sedation assessment (15.3%) and was also used in clinical practice (reported in PPS as a sedation assessment toll in 18.2%). It was not possible to determine if it was used as a sedation scale or for consciousness evaluation (our study included 7.3% of neurocritical patients). The GCS should not be used as a sedation scale because it has more inconsistent interobserver reliability, particularly in nonverbal intubated patients, compared with validated sedation scales. (26) Other studies have used this scale for sedation, but it has always been pointed out as a study limitation. (6,30) With the PPS, we concluded that 32% of patients were oversedated and 11% were undersedated when compared with the sedation targets defined for that specific day (Figure 3). Therefore, 57% had an adequate RASS level. Interestingly, in the survey, the percentage of oversedation reported by physicians was 46.4%, which was even higher than we observed in the PPS. This is an improvement target, as oversedation is related to worse ICU outcomes.
When we considered sedation, propofol was the most frequently reported sedative both in the survey and the PPS. Benzodiazepines, namely, midazolam, were reported in PPS in 27.5%. When comparing our results with data from other countries, this is a positive trend toward the best clinical practice, as the target should always be toward light sedation. (31) Reports around Europe show differences in benzodiazepine use. For example, midazolam is more frequently used in some countries, such as Spain, (15) France, (14) Poland (11) (respectively: 16.7%; 75%, >90% of patients), and less frequently used in the UK and Nordic countries, (13) where propofol occupies the leading position (in 80-98% of cases). Dexmedetomidine has been available in Portugal since 2011 but was used in a smaller number of patients (4.6%), possibly due to cost issues or lack of experience. (32) Pain assessment is common in Portuguese ICUs (the survey reports 93% and the PPS 91.1%). It should be generalized to all patients, as pain control is a main goal in ICU care. The importance of pain control is well demonstrated in the eCASH (early Confort using Analgesia, minimal Sedatives and maximal Human care) (7) concept, where effective pain relief is the first priority associated with minimal sedation. Opioids remain the mainstay for pain management; however, because of safety issues (such as sedation, delirium, respiratory depression, ileus and immunosuppression), there is a trend toward the use of a multimodal analgesia approach. This strategy allows opioid sparing and analgesic effectiveness improvement with other agents, such as acetaminophen, ketamine and nonsteroidal anti-inflammatory drugs (NSAIDs). Nevertheless, it is important to keep in mind that the best strategies should always be individualized.
There were some differences in the first analgesic option between the survey and the PPS. In the survey, opioids were the preferred analgesic, followed by acetaminophen. In the PPS, the most frequently used analgesic medication was acetaminophen, in monotherapy or in association with opioids.
In the PPS, opioids were the second most commonly used analgesic agents. Locoregional analgesia use was still marginal in most of our ICUs.
In the PPS, all participating ICUs assessed pain with validated scales, a fundamental and priority approach for patient best care. The BPS was the most frequently used scale (45.7%; 80/175), followed by the NRS (20.6%; 36/175), the VAS (9.7%; 17/175) and the ESCID (5.7%; 10/175). Similar reports were seen in the survey, with BPS, VAS and NRS being the most used scales (Figure 4). In some studies, pain evaluation is dependent on patient collaboration, leaving noncollaborating patients without evaluation. In France, pain assessment in communicative patients was reported to occur in approximately 70% of cases and in noncommunicative cases, 30%. (14) A particularity of our study was the diversity of pain scales used, with the behavioral assessment tools being preferentially listed for patients unable to self-report pain, such as BPS. This was not the focus of this study but could perhaps be explained by the difficulty in pain measurement and the attempt to choose the most reliable and valid pain assessment method for each patient.
Delirium assessment presented the most divergent results between the survey and the PPS. The authors believe this is one of the procedures with more opportunity for improvement and further interventions.
In the survey, physicians described delirium as an important cause of mortality and morbidity, and 70% reported assessing it daily (most of them using CAM-ICU). This was not translated into practice, as with PPS, there was a small percentage of delirium assessment by validated scales (< 5%) and a high percentage of missing values. The CAM-ICU was the most used and reported scale in both studies (Figure 4). Hypoactive delirium is frequently not identified without a routine and validated screening tool, and even hyperactive delirium can be undiagnosed or misdiagnosed. (8) Different studies show that the use of validated scales along with education improves the ability to detect delirium in the ICU. (33,34) We assume that the lack of a validated delirium assessment is responsible for the low delirium prevalence (7.8%), contrary to the majority of the studies with a higher prevalence of delirium (approximately 31.8%). (35) The authors believe that there are several reasons that explain the low adherence to formal delirium screening. On the one hand, in our survey, more than 60% of respondents considered delirium diagnostic scales difficult to apply. On the other hand, and according to a survey performed by Ely et al. (36) in 2001, the majority of health care practitioners believed delirium was a prevalent problem; nonetheless, protocols for managing delirium were scarce. In our survey, even though the existence of protocols for ASD was considered useful by 95% (104/110) of the physicians, fewer than 50% described having such protocols in their ICUs (Table 2).
According to PPS, physical restraints were used in 9.4% of patients. The PPS did not address the rationale for physical restraints in intubated and nonintubated critically ill patients or the type of physical restraints.
There is a lack of information about its use, safety, efficacy, and outcomes in critically ill adults. (6) Restraint use varies widely between 0% in some European countries and more than 75% in North America. Prior to this study, the only Portuguese information about restraint usage was 87% in the ICU. (37) Our study presents some limitations. First, the time elapsed since the study took place. These study started before the publication of the 2018 Pain, Agitation/Sedation, Delirium, Immobility (rehabilitation/mobilization) and Sleep (PADIS) guidelines. (6) At that time, the PAD-Pain, agitation/sedation and delirium guidelines were the main references for clinical practice. (38) Second, surveys are frequently used to establish current clinical practice in a variety of health care settings; they are lowcost, quick, and easy to conduct, although achieving a high response rate and reliable information can be challenging. In this survey, there was a low participant response rate (28% response rate), similar to other surveys (French -18.1%; (14) Poland -37,8% (11) ). We must assume some degree of bias, as studies relying on self-report regarding clinical practices overestimate the use of evidence-based medicine compared with real-life practices.
Aside from these expected limitations, there are strengths to report: 1) To the best of our knowledge, this is the first study about analgesia, sedation and delirium practices in Portuguese ICUs; 2) Considering the extensive drawbacks of surveys, we also performed PPS regarding the same aspect of the survey. With the survey, we aimed to address the perception of analgesia, sedation and delirium practices among intensivists and with the PPS the actual practices.
3) Despite a low individual response rate in the survey, the respondents represented most of the ICUs from all the different regions of Portugal.

CONCLUSION
To the best of our knowledge, this is the first Portuguese national study that encompasses both a national survey and a point prevalence study that provided a detailed perspective of the Portuguese approach concerning analgesia, sedation and delirium. This study emphasizes the need for widespread educational efforts for the implementation of evidence-based strategies for analgesia, sedation and delirium management in Portuguese intensive care units.

AUTHORS' CONTRIBUTIONS
MC Paulino and I Pereira interpreted the final results and drafted the manuscript. All authors developed the initial idea and planned the study, reviewed successive versions, and approved the final version of the manuscript.