Palliative Care Pilot Assesses Decision Support Tool for Sedating Medications

A multicenter mixed-methods pilot investigation conducted across four specialist palliative care centers in Germany has evaluated the clinical feasibility, acceptability, and perceived appropriateness of a paper-based clinical decision support tool designed to guide healthcare professionals in administering potentially sedating pharmaceuticals. Developed within the broader framework of the iSedPall research project, the instrument—designated by researchers and clinical teams as the warning list—synthesizes evidence-based pharmacological data and expert consensus to address longstanding clinical uncertainties surrounding the sedative effects of medications routinely prescribed in end-of-life care. While existing medical guidelines and professional recommendations in palliative medicine have historically focused on intentional palliative sedation administered specifically to relieve refractory suffering, unintentional sedation occurring as an adverse secondary drug effect has received comparatively little formal systematic attention in clinical workflows. The newly piloted decision support tool seeks to establish baseline reference points and dosing cut-off values that indicate when mild or deep sedation should reasonably be expected in patients, thereby providing clinicians with structured reference data to enhance clinical assessments, reduce unintended sedative complications, and assist communication among interdisciplinary care teams.

To rigorously investigate how the warning list performs in real-world healthcare settings, the research team implemented a comprehensive mixed-methods research design that integrated three distinct evaluative components across four participating pilot centers in Bavaria, Germany, comprising two hospital-based inpatient palliative care units and two specialized palliative home care teams. Registered prospectively in the German Clinical Trials Register under registration identifier DRKS00027241, the overall investigation incorporated a large retrospective cohort study conducted through structured medical chart review, an online survey distributed to clinical staff across the participating institutions, and six qualitative focus groups conducted with multidisciplinary healthcare workers. Quantitative data gathered from the chart abstraction and survey instruments were systematically processed using descriptive statistical analyses in IBM SPSS Statistics, while qualitative narrative data derived from audio-recorded focus group interviews underwent structured qualitative content analysis facilitated by MAXQDA analytical software. This multifaceted empirical design allowed researchers to cross-examine documented patterns of clinical utilization against professional perceptions of clinical utility, operational relevance, daily workflow integration, and potential clinical misinterpretations.

The retrospective cohort component reviewed the detailed medical records of 1,189 consecutive patients receiving specialized palliative care across the four participating centers between May 2023 and January 2024. The abstracted patient population exhibited an age range extending from 27 to 100 years, with a mean chronological age of 75 years and a standard deviation of 12.7 years. Demographic characteristics documented in the dataset indicated that 46 percent of the cohort, representing 547 individuals, were female, and 69.1 percent, accounting for 822 patients, carried a primary oncological diagnosis. Dedicated study nurses extracted data from patient files concerning which specific sedating pharmaceuticals were prescribed, whether designated cut-off values for expected light or deep sedation were reached during treatment courses, whether the clinical warning list was formally documented as having been utilized, and patient discharge or mortality outcomes. Attending physicians categorized each patient into one of five predefined clinical groups based on treatment intent and sedation status, utilizing the validated Richmond Agitation-Sedation Scale modified for palliative care to establish sedation depth.

Pharmacological documentation demonstrated that approximately two-thirds of the studied cohort, totaling 804 patients or 67.6 percent of the 1,189 reviewed individuals, were administered at least one potentially sedating medication cataloged within the clinical warning list during their clinical care. The reference tool itself encompasses ten established palliative pharmaceuticals—specifically diazepam, lorazepam, midazolam, levomepromazine, haloperidol, melperone, pipamperone, propofol, dexmedetomidine, and trazodone—alongside phenobarbital, which had been originally planned as an active component during early tool formulation. Across the four pilot facilities, seven of the eleven cataloged substances were actively administered to patients in routine practice, with lorazepam recorded as the most frequently prescribed agent, followed by levomepromazine and midazolam. Conversely, four cataloged medications—propofol, dexmedetomidine, trazodone, and phenobarbital—were not administered to any patients within the analyzed clinical cohort during the study window, reflecting specific prescribing patterns and setting-specific clinical limitations observed within inpatient units and specialized outpatient home care teams.

Analysis of dosing thresholds revealed considerable variations among individual pharmaceuticals regarding the proportion of patients who reached established cut-off benchmarks for anticipated sedation. Threshold values corresponding to expected light sedation were reached by at least some patients across all seven administered drugs, encompassing between 0.1 percent and 13.0 percent of treated individuals depending on the particular agent. In contrast, threshold doses corresponding to expected deep sedation were attained for only three specific medications: lorazepam, midazolam, and levomepromazine, occurring in 0.3 percent to 4.8 percent of patients receiving these therapies. Notably, midazolam emerged as the pharmaceutical agent for which both light and deep sedation cut-off thresholds were reached most frequently across the entire patient cohort. More than half of all evaluated patients, representing 699 individuals or 58.8 percent of the total sample, received potentially sedating drugs without documented sedation, falling into groups where sedation was neither intended nor observed or where sedation had been contemplated but ultimately not realized.

Clinical utilization of the warning list exhibited substantial variation depending directly on the treatment category to which patients had been assigned by their attending physicians. Patients who experienced intentional sedation initiated specifically to relieve refractory distress and those who developed sedation as an adverse drug reaction without the feasibility of adjusting dosages each accounted for less than 5 percent of the total cohort. Within these two distinct categories, high proportions of patients reached the light sedation cut-off value, comprising nearly 90 percent of the intentional sedation group and approximately 85 percent of the adverse reaction group, while roughly one-third of patients in each group reached the threshold for deep sedation. Furthermore, these two groups demonstrated the highest mortality rates among all categorized patient subsets. Crucially, the retrospective review established that clinical staff utilized the decision support warning list most frequently when managing patients who developed sedation as an adverse secondary drug reaction, confirming that the tool functioned in practice precisely where diagnostic ambiguity regarding drug-induced consciousness impairment was most pronounced.

In parallel with the chart review, the investigators gathered quantitative and qualitative feedback directly from healthcare practitioners through an online survey conducted between September and October 2023 via the Unipark survey platform. Distributed by email to approximately 100 multidisciplinary staff members across the four clinical centers, the questionnaire yielded completed responses from 33 healthcare workers, including 9 physicians, 17 registered nurses, and 7 practitioners representing other palliative care disciplines. Among the 28 respondents who answered questions regarding active tool use during the pilot implementation phase, nearly half, comprising 13 professionals—specifically 7 physicians and 4 nurses—reported having utilized the warning list in their clinical work. When asked about long-term adoption, 24 participants provided definitive feedback, with one-third of respondents, representing 8 clinical staff members including 4 physicians and 3 nurses, indicating an intention to continue consulting the warning list in daily practice following the formal conclusion of the research study.

To contextualize these survey metrics and explore professional experiences in greater depth, researchers conducted six qualitative focus groups between October 2023 and January 2024, involving 22 palliative healthcare professionals across interview sessions lasting between 70 and 110 minutes. The focus group cohort comprised primarily experienced physicians and nursing staff, the majority possessing more than five years of specialized professional experience, with eight clinicians participating both in their respective center-specific group and in an aggregated setting-specific discussion. Using semi-structured interview guides informed by preliminary survey findings and the Consolidated Framework for Implementation Research, the qualitative analysis identified four dominant thematic categories: overall acceptability of the warning list, perceived clinical relevance, practical contexts of routine use, and operational user-friendliness. Participating clinicians reported that the fundamental objective of the decision support list was clearly understood, with multiple practitioners noting that reviewing defined medication ranges helped them establish renewed mindfulness regarding pharmacological boundaries and dosage thresholds.

Qualitative commentary emphasized that the decision support tool delivered practical benefits primarily in staff education, team communication, and interdisciplinary clinical case conferences. Physicians and nursing staff observed that the reference list provided a concise educational overview that was particularly valuable when onboarding new colleagues and establishing a shared baseline understanding of sedative liability. During routine team meetings and morning handovers, multidisciplinary teams utilized the documented cut-off values to review night-time as-needed medication orders, evaluating whether cumulative dosages had crossed defined limits and achieving consensus on whether observed decreases in patient responsiveness constituted drug-induced sedation. In at least one documented instance, a senior physician recounted that cross-referencing patient medication against the warning list prompted the clinical team to recognize that a patient’s somnolence was drug-related, which subsequently led to a transparent, reassuring explanatory discussion with the patient’s family members regarding treatment side effects and clinical status.

Despite broadly favorable reception regarding educational utility, focus group participants identified distinct operational barriers and usability concerns that complicated daily reliance on the reference document. Several physicians noted that because the warning list derives its cut-off dosing thresholds from a theoretical standard patient, it necessarily lacks the nuanced complexity required to evaluate fragile palliative patients with multimorbidity, fluctuating organ function, and concurrent polypharmacy. Experienced physicians expressed a preference for relying on their own clinical intuition and specialized pharmacological knowledge rather than consulting a simplified standardized document. Conversely, other clinicians described the paper tool as cumbersome and time-consuming, pointing out that because dosing values on the list were expressed in daily cumulative amounts, clinicians frequently had to perform manual mathematical conversions to evaluate divided twice-daily or thrice-daily medication regimens within pressurized clinical schedules.

Focus group participants also identified a significant clinical risk inherent in the tool’s formatting, cautioning that less experienced practitioners might misinterpret the warning list’s cut-off thresholds as recommended starting doses for intentional palliative sedation rather than upper boundaries indicating potential adverse drowsiness. Clinicians illustrated this potential misunderstanding by pointing to cataloged medications such as pipamperone, an antipsychotic included on the warning list because of its potential sedative side effects but clinically unsuitable as an agent for inducing intentional therapeutic sedation. To mitigate this hazard, investigators noted that separate standardized dosing recommendations specifically governing the initiation of intentional palliative sedation had been developed as a complementary companion tool within the wider iSedPall intervention package, ensuring that clinicians maintain a clear distinction between avoiding unintended adverse sedation and administering planned therapeutic symptom control.

Discussions regarding future implementation revealed widespread interest in integrating the warning list directly into electronic health records and digital prescribing systems, though clinicians and researchers weighed distinct trade-offs associated with digital automation. Participating professionals suggested that embedding automatic alerts within digital patient charts could instantly warn prescribers when cumulative midazolam or lorazepam doses crossed critical thresholds, prompting real-time clinical reflection without requiring manual calculations. However, the study authors highlighted that digital clinical decision support introduces recognized challenges documented across medical literature, including alert fatigue resulting from excessive automated warnings and the risk of diverting clinical focus away from holistic bedside patient evaluation toward rigid electronic data entry. Consequently, the authors emphasized that maintaining a portable paper version or pocket card remains highly relevant, particularly in outpatient home care settings and bedside team discussions where digital terminals are unavailable or impractical.

The authors acknowledged several key methodological limitations that contextualize the study findings, foremost being the restriction of the pilot investigation to four palliative care centers located exclusively within Bavaria, Germany, which naturally restricts the immediate generalizability of the findings across diverse healthcare environments and national palliative care models. Furthermore, the small absolute sample size of the voluntary online survey and the inclusion of only two brief questions dedicated specifically to the warning list within a broader survey battery restricted the breadth of quantitative survey conclusions. Nevertheless, the study demonstrated that the warning list is acceptable, feasible to implement, and clinically relevant, providing valuable pilot evidence that structured decision support can effectively heighten professional awareness of sedative adverse drug reactions while underscoring that standardized reference tools must always support, rather than supplant, individualized multidisciplinary clinical judgment.

Subject of Research: Clinical decision support for potentially sedating medications in specialist palliative care

Article Title: Clinical decision support for potentially sedating drugs in palliative care: a mixed-methods pilot study

Article References: Krauss, S. H., Odierna, B., Kolmhuber-Seibold, S., Rémi, C., Bausewein, C., Klein, C., Ostgathe, C., Kremling, A., Schildmann, J., Schildmann, E., & the iSedPall study group (2026). Clinical decision support for potentially sedating drugs in palliative care: a mixed-methods pilot study. BMC Palliative Care, 25(1), Article 288. https://doi.org/10.1186/s12904-026-02346-z

Image Credits: AI Generated

DOI: 10.1186/s12904-026-02346-z

Keywords: palliative care, clinical decision support, potentially sedating drugs, iSedPall project, adverse drug reactions, retrospective cohort study, mixed-methods research, unintentional sedation, lorazepam, midazolam, levomepromazine, chart review, focus groups, electronic health records, RASS-PAL scale, palliative home care, inpatient palliative care, clinical pharmacology