Decision fatigue in hospital settings: A scoping review

Abstract

Background

“Decision Fatigue” (DF) describes the impaired ability to make decisions because of repeated acts of decision‐making.

Objectives

We conducted a scoping review to describe DF in inpatient settings.

Methods

To be included, studies should have explored a clinical decision, included a mechanism to account for the order of decision making, and be published in English in or after the year 2000. Six databases were searched. Retrieved citations were screened and retained studies were reviewed against the inclusion criteria. References of included studies were manually searched, and forward citation searches were conducted to capture relevant sources.

Results

The search retrieved 12,781 citations, of which 41 were retained following screening. Following review, 16 studies met the inclusion criteria. Half were conference abstracts and none examined hospitalists. Emergency medicine and intensive care settings were the most frequently studied clinical environments (n = 13, 81%). All studies were observational. The most frequently examined decisions were about resource utilization (n = 8, 50%), however only half of these examined downstream clinical outcomes. Decision quality against prespecified standards was examined in four (25%) studies. Work environment and patient attributes were often described but not consistently accounted for in analyses. Clinician attributes were described in four (25%) investigations. Findings were inconsistent: both supporting and refuting DF's role in the outcome studied.

Conclusions

The role of clinician, patient, and work environment attributes in mediating DF is understudied. Similarly, the context surrounding the decision under study require further explication and when assessing resource use and decision quality, adjudication should be made against prespecified standards.

INTRODUCTION

“Decision fatigue” (DF) is a term coined by psychologists to describe “the impaired ability to make decisions and control behavior as a consequence of repeated acts of decision‐making”. ¹ Humans may have a finite cognitive reserve for making decisions, and its depletion results in impaired executive function and decreased self‐control as decisions are made—ultimately influencing subsequent decisions. DF may lead to avoidance of decision‐making, impulsivity, reliance on mental shortcuts and a default to decisions that appear to be safer. ² , ³ , ⁴ The greater the difficulty of the decision, the more DF an individual may face. ⁵ The practice of medicine is cognitively intense, with internal medicine clinicians making an average of 15.7 decisions per encounter and hospitalists getting interrupted once every 8 minutes. ⁶ , ⁷ There are therefore profound implications in healthcare should decision quality erode with repeated decision‐making.

Previous studies have demonstrated the negative consequences of DF in outpatient settings with higher rates of inappropriate antibiotic prescribing, higher likelihood of opioid prescribing, lower rates of breast and colon cancer screenings, and lower rates of diagnostic test ordering for patients with appointments later in the day. DF has also been postulated to contribute to the difference the time of day makes in surgeons' decisions to operate and even hand hygiene compliance among healthcare workers. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³

Acuity among hospitalized patients has been rising and hospital care has been transformed by the advent and rapid growth of hospitalists—clinicians who assume care for patients during their hospitalization. ¹⁴ , ¹⁵ Care in hospital settings differs from that in outpatient medicine, prompting our question: “what is known about decision fatigue in hospital settings”? As the state of the evidence is evolving, the nomenclature is not defined, and the populations of patients and healthcare workers in hospital settings are heterogenous, the question lends itself well to scoping review methodologies. While hospital‐based clinicians are susceptible to multiple forms of fatigue which may intersect and overlap with each other—including physical fatigue related to sleep deprivation and long working hours, compassion fatigue, and alarm fatigue—our focus in this work is on decision fatigue. To describe what is known about clinical outcomes related to DF in hospital settings and serve as a scaffold for future work, we undertook a scoping review.

METHODS

This manuscript follows the PRISMA for scoping reviews (PRISMA)‐ScR reporting guidelines. ¹⁶

Eligibility: Participants, concept and context

A protocol was created using the Johanna Briggs Institute template. ¹⁷ All studies in English were included. Studies published before 2000 were excluded as the emergence of the hospitalist model in 1996 transformed inpatient care and we wanted this review to reflect the current state. ¹⁸ Experimental, quasi‐experimental, analytical and descriptive observational, and qualitative studies were included. Studies were included if the care setting was in the hospital regardless of the clinician or patient group studied. As the concept of interest was DF in clinical settings, included studies had to assess the possible impact of DF on a decision and account for the order of decision‐making in some way (e.g., order of rounding, time elapsed during a shift, etc). Without the explicit evaluation of decisions, the impact of DF on outcomes by time of day is imputed and complicated by factors related to physical fatigue, staffing, and resource availability that differ between regular and off‐shift hours. Accordingly, studies investigating complication rates, morbidity, or mortality by time of day without specifically assessing decision‐making along the path of causality were excluded. To focus on the consequences of DF in real‐life scenarios, simulation studies were excluded as were studies without patient care related outcomes. An initial limited search of MEDLINE was undertaken to identify articles on the topic and to assist with the creation of the protocol.

Search strategies

To identify potentially relevant documents, the following bibliographic databases were searched on March 11, 2024: MEDLINE (Ovid), Embase (Ovid), Cochrane Library (Cochrane), CINAHL (Ebsco), PsycINFO (Ebsco), and Scopus (Elsevier). Using database filters, searches were limited to the English language, and to the years 2000–2024. The Cochrane search hedge for Human studies was also applied to limit studies to those with human subjects. Search strategies were drafted by an experienced librarian (J. C. S.) and further refined through team discussion. The MEDLINE search strategy was reviewed by another librarian before execution using the Peer Review of Electronic Search Strategies (PRESS) Checklist. ¹⁹ The text of each database search strategy is found in Supporting Information S2: Appendix A and can also be accessed via the data availability statement at the end of the article. The results from all databases used were aggregated and de‐duplicated using the Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia (Available at www.covidence.org). The electronic database search was supplemented by manually searching references of included full‐text articles and by forward citation searching the 13 articles in Web of Science on June 9, 2024 (no limits used, full list of Web of Science databases listed in Figure 1.

Figure 1.

PRISMA flow diagram. Source: Page MJ, et al. BMJ 2021;372:n71. doi: 10.1136/bmj.n71. This work is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

Data extraction

Following a pilot test, titles and abstracts were screened by two independent reviewers for assessment against the inclusion criteria. Following screening, sources considered potentially relevant were retrieved as full‐text manuscripts when available and all potentially relevant sources were assessed against the inclusion criteria by two or more independent reviewers. At this stage, reasons for exclusion of sources of evidence were recorded. Disagreements between the reviewers at each step of the selection process were resolved through discussion. A data extraction form was created and piloted. Following iterative improvements, the form was used by reviewers to summarize details about the participants, concept, context, study methods and key findings relevant to the review question.

RESULTS

The initial search resulted in 12,781 citations. Following the removal of 3104 duplicates, 9677 studies were screened for inclusion and 9636 were found to be ineligible. Of the 41 studies retained for further review (available in Supporting Information S2: Appendix B), 28 were excluded, leaving 13 citations. Three further citations were identified by manual review of references of included sources. When we forward citation searched the 13 citations, we identified a full‐length manuscript corresponding to an included abstract. We replaced the abstract version with the full‐length manuscript resulting in a final set of 16 ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ (Figure 1).

Results are summarized in Table 1. Half (n = 8, 50%) of the studies that met the inclusion criteria were conference abstracts while the other half were full length peer reviewed manuscripts.

Table 1.

Summary of included articles.

Title (first author's last name) Reference number)	Publication year	Type	Setting	Study design	Study period	Variable to capture order of decisions/time	Decision(s), and outcome(s) studied	Results	Work environment, clinician or patient attributes described
Influence of external factors on frozen section performance (Wisell) 20	2013	Abstract	US, Pathology	Retrospective	6 months	Time of day	Diagnostic discrepancies based on final diagnosis) and deferrals	973 frozen sections. Bimodal peak in rates of discrepancies and deferrals. First between 10–11 a.m. following the busiest period and again at the end of the day.	Work environment: numbers of cases per hour.
Impact of emergency department surge and end of shift on patient workup and treatment before referral to internal medicine: A health records review (Charbonneau) 21	2018	Manuscript	Canada, Emergency Medicine	Retrospective	2013– until target number of patients for each diagnosis reached	End versus middle of shift	Completeness of workup before referral to internal medicine for admission, admission rates, redirection to other services, diagnostic disagreement Outcomes: patient mortality, 30‐day return to ED	308 patients, No differences in completeness of workup by end of shift or surge	Work environment: process of admission to the hospital Patient attributes: age, sex, comorbidities, medications
Relationship between provider fatigue and pulmonary embolism evaluation in the emergency department (Graves) 22	2018	Abstract	US, Emergency Medicine	Retrospective	4 years, 2013–2017	Time in shift “late“ versus “early”	Adherence to guideline directed evaluation with d‐dimer in low‐risk patients based on simplified revised Geneva score	532 patients, No difference in guideline adherence by “early” or “late” time in shift	None
Influence of ward round order on critically ill patient outcomes (Evans) 23	2018	Abstract	Australia, Intensive Care Unit	Retrospective	1 year, 2014	Bed assignment (first three beds vs. last three beds on unit equated to rounding order)	Decision to extubate Outcomes: ICU length of stay, hospital length of stay, ICU mortality, duration of mechanical ventilation	681 patients, No differences in outcomes based on rounding order	None
The impact of decision fatigue on ancillary test ordering in the emergency department (Petrie) 24	2019	Abstract	US, Emergency Medicine	Retrospective	6 months, 2016	Hour of shift	Laboratory and imaging orders per clinician over shift duration	Increased utilization of lab tests, decreased rates of CTs as shift progresses. Increased rates of lab tests and CTs associated with total and hourly active patient burden.	Work environment: time to patient room assignment, numbers of patients seen Patient attributes: triage acuity
Case start time affects intraoperative transfusion rates in adult cardiac surgery, a single‐center retrospective analysis (Addis) 25	2020	Manuscript	US, Anesthesiology	Retrospective	3 years, 2013–2016	Start time	Administration of any (packed RBCs, platelets, plasma, cryoprecipitate) allogenic blood products and administration of individual blood products	1421 cases Increasing probability of transfusion of packed RBCs with later start times, no differences in transfusion of other products	Patient attributes: initial hemoglobin, age, sex, body mass index, body surface area, comorbidities, case duration, pump duration, aortic clamp duration Clinician attributes: surgeon and anesthesiologist identity
Later emergency provider shift hour is associated with increased risk of admission: A retrospective cohort study (Tyler) 26	2020	Manuscript	US, Emergency Medicine	Retrospective	6 years, 2010–2016	Hour of shift patient first assessed	Likelihood of admission	294,031 encounters Higher likelihood of admission in last hour, last quarter and in later hours of shift	Work environment: annual patient volume Patient attributes: age, sex, Emergency Severity Index
Decision fatigue in the Emergency Department: How does emergency physician decision making change over an 8‐h shift? (Zheng) 27	2020	Manuscript	Canada, Emergency Medicine	Retrospective	2 years, 2014–2016	Hour of shift patient first assessed	Proportion of patients with consultation requests and/or imaging orders Outcomes: ED length of stay, return to ED within 72 h	87,752 patients No differences in consultations, consultations resulting in admission, or return visits for patients discharged without consultation by hour of shift. The rates of CT head and CT abdomen and ED LOS declined while the rate of CT chest did not change as the shift progressed.	Work environment: annual patient volume, layout of department, shift assignment Patient attributes: age, sex
Patterns of decision fatigue during rounds in the medical intensive care unit (Auriemma) 28	2020	Abstract	US, Intensive Care Unit	Prospective observational	61 days	Direct observation	Change in plan versus maintenance of status quo	17 intensivists, 354 patients, 7646 decisions. At patient level: later decisions more likely to result in change of plan, at day level later decisions more likely to maintain status quo	Clinician attributes: gender, experience Patient attributes: APACHE score
Weekend extubation rates differ between weekends of intensivist continuity vs non‐continuity (Macauley) 30	2020	Abstract	US, Intensive Care Unit	Retrospective	4 years 2016– 2019	Continuity versus discontinuity of care by intensivist over the weekend	Decision to extubate	190 weekends, 312 extubations. Noncontinuity weekends showed consistent trends toward more extubations	None
How are patient order and shift timing associated with imaging choices in the Emergency Department? Evidence from Niagara Health administrative data (Strobel) 31	2022	Manuscript	Canada, Emergency Medicine	Retrospective	6 years, 2013–2019	Order in which the patient was seen per physician (censored at 40), time in physician's shift that patient was seen (in 15‐min increments)	Imaging orders Outcomes: return to ED within 7 days	841,683 visits Probability of imaging increased later in shift, probability of imaging decreased as patient order increased, magnitude of change much larger by patient order than for time, no differences in return to ED based on imaging status	Work environment: annual patient volume, triage processes, hour of the day, day of week, year, site and number of patients seen Patient attributes: age, sex, triage acuity, ambulance use, presenting complaint Clinician attributes: identity
Effects of decision fatigue on throughput and resource utilization over 12‐h shifts in the emergency department (Treasure) 32	2022	Manuscript	US, two free standing EDs, Emergency medicine	Retrospective	5 months, 2020	First 8 h versus last 4 h of 12‐ h shifts	Decision to admit, time to decision to admit, Outcomes: door to doctor time, throughput time	9724 Fewer patients seen in the last 4 h of shifts. Lower admission rates and higher throughput time in last 4 h, (driven by site with higher patient load) No differences in decision to admit or door to doctor time	Work environment: annual patient volume, patients per hour
Association of time of day with delays in antimicrobial initiation among ward patients with hospital‐onset sepsis (Ginestra) 29	2023	Manuscript	US, Ward patients	Retrospective	2.5 years, 2017–2019	Hour of day with 7 a.m. as the reference for the start of the day	Initiation of antimicrobial therapy for hospital‐onset sepsis Outcomes: mortality rates	1672 patients Probability of initiating antimicrobials lowest at shift changes, declined throughout night shift, stable during day shift, no difference in mortality rates by time of sepsis onset	Work environment: year, time in year (quarter) service type, hospital Patient attributes: age, gender, severity of illness (SOFA score, mechanical ventilation, use of vasopressors, admission diagnosis
Does hour of shift affect emergency department provider decision to obtain a Computed Tomography scan? (Cebula) 33	2023	Abstract	US, Emergency Medicine	Retrospective	2 years, 2020–2022	Time in shift	CT scan ordering	87,989 visits Adjusting for acuity, patients seen in the last hour and last half of the shift were less likely to have CT scans ordered	Work environment: Shift at which the patient was initially evaluated Patient attributes: Emergency Severity Index, age, sex
Burden of decision making and cognitive function amongst high consequence decision‐makers in Intensive Care (Khalil) 34	2023	Abstract	UK, Intensive Care Unit	Prospective observational	Four 24‐h on‐call periods	Direct observation	Decision fatigue scale score, reaction times, quantification of decisions and interruptions	Decision fatigue scale score and reaction times increased post call. Mean number of decisions morning: 237 afternoon: 91, Mean number of interruptions morning: 40, afternoon: 11	Work environment: numbers of decisions, interruptions, staffing gaps on the unit Clinician attributes: age, seniority, number of days on the unit. Patient attributes: acuity
Impact of fatigue on emergency physicians' decision‐making for Computed Tomographic scan requests and inpatient referrals: An observational Study from a tertiary care medical center of the Sultanate of Oman (Al Arimi) 35	2023	Manuscript	Sultanate of Oman, Emergency Medicine	Retrospective	3 months, 2019	First versus second half of three 8‐h shifts (8 a.m.–3 p.m., 3 p.m.–11 p.m., 11 p.m.–8 a.m.)	CT scan ordering, referral for inpatient admission Outcomes: yield of CT scan, outcome of referral for admission	941 scans, 1048 admission referrals. No difference in CT scan ordering or yield by first or second half of shift, no difference in referrals by first or second half of shift but higher proportion of negative referrals in second half of afternoon shift	Work environment: total number of beds, occupancy, daily patient load, diurnal distribution of patients, structure of team

Thirteen (81%) of the included studies were set in either the United States or Canada. The remaining three were conducted in either Australia, the United Kingdom or the Sultanate of Oman. Nine (56%) addressed care provided in Emergency Department (ED) settings, while the next most common study setting was intensive care units (ICUs) (n = 4, 25%). Only one (6%) addressed general medicine patients in ward settings.

All studies were observational, with two using prospective data gathering, while the remainder (n = 14) were retrospective evaluations, often of large administrative data sets. Most retrospective studies (12/14, 86%) used actual time of day, or time since the start of the workday or shift, as a marker for when clinical decisions were made. Many studies used shift start times to further categorize time into “early” or “late” periods. One specifically included the order in which the patient was seen in the analysis. Both prospective studies were in ICU settings and used direct observation of clinicians to quantify and characterize decisions. Retrospective evaluations included data spanning 5 months to 6 years. One prospective observation followed physicians for four on‐call days and the other reported 61 days of observations.

The most commonly studied decisions pertained to resource utilization. Eight (50%) studies reported rates of resource utilization including the use of consultations and referrals for admission through the ED, laboratory or imaging use. Four of these eight investigations simultaneously examined the relationships between the decision and subsequent outcomes (e.g., rates of return to the ED, yield of imaging studies or the outcome following referral for inpatient admission). Decision quality measured against prespecified standards was reported in four (25%) investigations and included outcomes such as completeness of evaluations before referral for inpatient admission, guideline concordant test utilization, initiation of antimicrobials in patients with sepsis or diagnostic accuracy. Studies directly observing clinicians prospectively reported changes in the Decision Fatigue Scale and reaction times and whether decisions resulted in changes in the care plan or continuation of the status quo.

Some aspects of the work environment or patient attributes were described in most (n = 13/16, 81%) studies, however, it was not always clear whether these were accounted for in analyses. Fewer (n = 4/16, 25%) reported clinician attributes and none examined possible clinician specific effects such as experience on decisions.

Findings and conclusions regarding the role of DF in the decisions and outcomes studied were inconsistent even when studies were conducted within similar clinical settings and with large sample sizes—with some detecting and others refuting the deleterious consequences of DF.

DISCUSSION

To our knowledge, this is the first scoping review that specifically examines DF in inpatient clinical settings and may help to advance the knowledge base and create a future research agenda.

Almost all investigations addressed care provided by clinicians in ED or intensive care settings with a single study examining patients in a general ward setting. While the impact of fatigue and workloads has been widely studied in different disciplines, we were surprised that we did not find studies in inpatient settings that addressed the concept of DF based on our criteria among hospitalists or other clinicians who work in demanding inpatient environments. Inpatient care continues to represent one of the highest expenditures in US healthcare, and examining DF and its impact in hospital settings may be consequential to quality, safety, efficiency, and wellness outcomes. ³⁶ , ³⁷ , ³⁸

Our review uncovered several methodological considerations for future work which may best be structured in the conceptual framework of DF previously outlined by Pignatello et al. In that framework, DF may be thought of as having “antecedents” that predispose to the development of DF, “attributes” that describe its impact on cognition, behavior, and physiology and finally the cumulative “consequences” of these phenomenon. ¹ The characterization of DF in inpatient care appears to be in its early stages—with all studies focusing on the detection of DF on cognitive or behavioral attributes and finding contradictory results.

The primary antecedent of DF is the act of repeated decision‐making and congruent with our inclusion criteria, was captured in some way in all studies. However, contexts surrounding decision‐making that are important and contribute to the development of DF such as the density of decision‐making were often not reported. Many studies used the time of day as a surrogate for the order of decision‐making and while the field of chronobiology is evolving, when examining DF, future studies may need to additionally examine the physiological differences related to circadian rhythm disruptions and their impact on outcomes that may be superimposed on DF. These disruptions may impact both clinicians and patients. ³⁹ , ⁴⁰ Similarly, while patient acuity was reported in many investigations, clinician attributes were rarely described. Clinician attributes such as experience may interact with DF and should be accounted for. There are no predefined “consequences” or clinical outcomes that are known to reflect DF and as such pragmatic choices have been studied and have often included rates of resource utilization. However, the downstream clinical correlations of resource use were inconsistently provided and assumptions of over or under use were occasionally made without arbitration: for example, while some studies reported rates of return to the ED associated with the use or nonuse of specific investigations, others did not. Outcomes also appear to be impacted by the culture and practices of specific contexts. For example, in ED settings, a reluctance to hand off investigations to colleagues toward the end of shifts may impact the acuity of patients seen, the numbers of patients seen, and investigations ordered with shift progression—factors that are difficult to completely control for analytically in retrospective studies. As future investigators consider study designs to assess DF in hospital settings, rounding order should be accounted for—as inpatient clinicians are also likely to see their most unstable patients (who also require the most consequential decisions) earlier in the day.

Choosing outcomes for DF is further complicated by the observation that (a) repeated decision‐making may not always be deleterious and may in fact “tune up” decision‐making, making it more efficient and effective in some scenarios. ³¹ and (b) the perceived risk of the decision by the clinician is likely to mediate DF. This latter finding was demonstrated in the study in which with later surgical start times, transfusion requirements increased for packed red blood cells but not for any other blood product—perhaps reflecting the different perceived risks of transfusing each product. ²⁵ As the research agenda matures, our recommendations for future work in DF in inpatient settings are summarized in Table 2.

Table 2.

Summary of recommendations to consider in future work on decision fatigue in inpatient settings.

Antecedents	Attributes	Consequences
Standardization of terminology and conceptual frameworks across studies
Clinician (e.g., age, gender, experience, time of day, expanding beyond diurnal variations to assess the impact of week on/week off and other schedules)	Perception of risk of decision being studied (e.g., expense, patient harm, local constraints and priorities)	Arbitration to determine the appropriateness of decision studied against prespecified criteria
Patient (e.g., acuity, severity, patient order and decisions per patient, time of day)	Presence of decision support (e.g., decision aids, stewardship of resource use, robustness of evidence based impacting the balance between “judgment” and objective criteria)	Consider outcomes that include all missions including clinician wellness
Clinical environment (e.g., numbers and rates of decisions, interruptions, multitasking, handoff culture and impact on numbers of patients seen over time)

While almost all studies have described DF in a diurnal pattern, one reported lower extubation rates with continuity of intensivist care over the weekend—raising the question of DF over the course of a longer clinical stretch. ³⁰ This may be particularly salient and an area for future study for hospitalists who frequently work for a week at a time at most institutions.

Our work has certain limitations. Our focus on real‐world settings may have caused us to omit preclinical data related to DF. We excluded languages other than English which may limit the international generalizability of our findings. Our search strategy was optimized to retrieve research relevant to DF in clinical inpatient settings and our work is therefore not reflective of broader concepts that may be adjacent to DF such as fatigue or sleep deprivation. This limitation also underscores the need for better standardization and specificity of the terminology that may be used in future studies of DF.

CONCLUSION

DF is a complex phenomenon mediated by interactions between clinicians, patients, and the work environment. We are early in our understanding of DF in inpatient settings, and future efforts to establish whether it occurs, the decisions that are most sensitive to it, and how to ameliorate it are needed.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Supporting information

Supporting information.

Supporting information.

Perry K, Jones S, Stumpff JC, et al. Decision fatigue in hospital settings: A scoping review. J Hosp Med. 2025;20:385‐395. 10.1002/jhm.13550

DATA AVAILABILITY STATEMENT

The search strategies used in this study are openly available in searchRxiv: https://doi.org/10.1079/searchRxiv.2024.00724, https://doi.org/10.1079/searchRxiv.2024.00725, https://doi.org/10.1079/searchRxiv.2024.00726, https://doi.org/10.1079/searchRxiv.2024.00727, https://doi.org/10.1079/searchRxiv.2024.00728, https://doi.org/10.1079/searchRxiv.2024.00695.