Reporting on Neurological Decline as Identified by Hourly Neuroassessments

Abstract

BACKGROUND:

Hourly neurological examinations (“neuroassessments”, “neurochecks”) have been widely used to monitor for a decline in neurological status, allowing for timely intervention. There are, however, limited data behind this common practice. The objective of this study was to identify how frequently neurological decline occurred across various diagnoses, and whether that decline 1) was identified by a scheduled neurocheck, and/or 2) altered management.

METHODS:

Cross-sectional survey performed in a neurological intensive care unit at a tertiary care academic medical center. Clinical neuroscience nurses caring for patients with hourly neurological assessments completed a brief survey at 12-hour shift completion.

RESULTS:

Data were collected from 212 nurse’s shifts. Neurological changes were identified by nurses in 14% (n=30) of shifts. The neurological change was identified during a scheduled neurocheck 67% of the time, with the detection of neurological changes more likely to occur during a scheduled neurocheck than at other times (p<0.05). There was no change to the care plan in 55% of the cases of neurological decline. Patients with subarachnoid hemorrhage were more likely to have a decline detected.

CONCLUSION:

Findings suggest that many patients undergo hourly neurological checks without ever identifying a neurological deterioration. In many instances of neurodeterioration, there was no change to the treatment plan pursued. Primary diagnoses and neurological changes may not be entirely independent, and therefore, hourly neurological examinations may have greater yield in some diagnoses than others. Replication is warranted with a larger sample to evaluate the risks and benefits of neuroassessments.

INTRODUCTION

Following acute neurological injury, frequent neurological examinations (neuroassessments, neurochecks) are often ordered by practitioners and performed serially by nurses to monitor for a change in neurological status.¹ These neuroassessments utilize a standardized approach to examine patients; they should evaluate level of consciousness, language and speech, cranial nerves, movement and sensation to uncover neurological deficits that impact patient care.^2,3 There is evidence to suggest that neurochecks are useful to guide timing of interventions, particularly neurosurgical interventions.⁴ However, these frequent neuroassessments can sometimes continue unabated for days on end.^5,6 There also appears to be some thought that frequent neurochecks contribute to poor outcomes such as short term cognitive dysfunction, as providers report increased willingness to discontinue hourly neurochecks in delirious patients.⁷ Given the relationship between in-hospital delirium and increased morbidity, increased mortality and worsened long-term cognitive outcomes,^8–11 risk factors for delirium such as sleep disruption need to be weighed against the potential benefit of more granular neuromonitoring.

The ideal frequency of neurochecks, as well as the ideal duration of frequent neurochecks is unknown.¹ Indeed, determination of an ideal frequency and duration are intrinsically related to whether neurochecks are achieving the goal of identifying neurological deterioration and/or altering management. As such, the objective of this study was to identify whether changes in neurological status are identified during formal scheduled hourly neuroassessments (scheduled) versus at other times during nursing encounters with the patient (other neuroassessment), and whether these recognized changes in neurological status result in a change in management across a range of diagnoses.

METHODS

This cross-sectional survey-based study was performed in a neurological intensive care unit (ICU) at a tertiary care academic medical center. Clinical neuroscience nurses from this single ICU were invited to complete a survey consisting of yes/no, multiple choice and write-in questions at the end of each shift caring for patients with hourly neuroassessments (see Supplemental Digital Content). Both day and night shift nursing participation were encouraged.

Survey questions were initially developed using a working group model in consultation with a member of the group with experience in survey methodology (JD), and subsequently validated for content and face validity using a volunteer cohort of physician and nursing providers. To homogenize results we attended to the size and length of question and answer choices. Answer choices were randomized when the order of the answer choices was irrelevant. Given the repeated need for nurse participation, minimizing the duration of the survey was necessary to improve initiation and then completion of the survey. The ideal time felt to balance the need for repeated participation with construct validity was <2 minutes per survey. Surveys were maintained through the SurveyMonkey^® web-based platform, and used branching logic to provide the next relevant question based on the answer to the previous question. Participant demographic information was not collected.

The survey opened in summer 2022 and remained open for eight months until the target number of surveys (>200) with complete data on hourly neuroassessments had been met. Although a power analysis could not be performed given the limited literature on neurochecks, we identified our sample size based on literature review with special attention paid to studies examining the utility of serial neuroassessments.¹² Participants were asked to complete the survey at the end of each of their 12-hour shifts to reflect the events of the shift they just worked. Surveys could be completed once per shift using a desktop computer or a smartphone. A neurological change was defined as any neurological worsening, and all reported instances of neurological decline were also confirmed by physician team members.

Clinical neuroscience nurse recruitment was performed using a multifaceted approach: 1) initial group email from nursing managers, 2) flyers posted in high visibility areas (e.g., break rooms, staff bathrooms) in the neurological ICU, 3) weekly email reminders from the study team, 4) reminders sent via the electronic health record to the charge nurse to encourage participation of bedside providers caring for patients ordered for hourly neurochecks. Recruitment for survey participation was encouraged with raffle incentives; completion of the survey was not required for raffle entry.

Descriptive statistics were used to summarize survey data. An answer to whether a neurological change occurred during the nursing shift was required for data to be included in analysis. A two-sample test of proportions was conducted to determine if the detection of neurological status changes was significantly greater during the hourly assessments than during the non-scheduled assessments. Chi-squared analysis was used to compare differences between categorical variables. Fisher’s Exact Test was used as an alternative to Chi-squared analysis when one or more of the categories had a frequency <5. A post-hoc analysis of residuals was conducted to determine which differences between categories were significant. All statistical analysis was performed by one author (PJL).

The study protocol was reviewed and exempted by the Institutional Review Board (IRB). Informed consent was obtained from all participants, though a waiver of documented consent was granted by the IRB. All anonymized data not published within this article will be made available by request from any qualified investigator.

RESULTS

Survey completion rate was 86%. 212 surveys provided data for the main question of whether a neurological change occurred and were therefore included in analysis. Thirty surveys (14%)—representing 30 nursing shifts—identified a neurological decline in a patients’ clinical status. Of these thirty neurological changes, 67% (n=20) were identified during a scheduled neuroassessment; the remaining 10 (33%) neurological changes were instead identified during other informal care interactions with the patient. This finding represents a significantly higher chance of identifying a neurological change during a scheduled neurocheck than at other times (p < 0.01).

Among the patients with neurological changes identified on a scheduled neurocheck (n=20), 11 (55%) surveys reported no change to the patient’s treatment plan, while nine (45%) reported a change in management. Changes to the treatment plan included: changes in medication (e.g., ordered or discontinued; n=8, 89%), orders for imaging (e.g. head computed tomography; n=8, 89%), changes in neurological examination frequency (n=3, 33%), procedures performed (e.g., intubation; n=2, 22%), and ‘other’ alterations (e.g., electroencephalography; n=1, 11%). More than one change may have occurred per patient.

Among the 212 surveys, diagnoses for patients receiving hourly neurochecks included: post-operative elective craniotomy (n=47, 22%), subarachnoid hemorrhage (n=46; 22%), ischemic stroke (n=30, 14%), intracerebral hemorrhage (n=24, 11%), ‘other’ (n=21, 10%), post-operative elective IR (n=16, 8%), status epilepticus (n=10, 5%), post-operative elective spine surgery (n=7, 3%), brain tumor without surgery (n=6, 3%), and was not reported in 5 instances (n=5, 2%).

In post-hoc analysis, it was noted that certain diagnoses were related to observing a neurological change (χ² = 17.04; p < 0.05; Table 1). Patients diagnosed with subarachnoid hemorrhage were more likely to have a change in clinical status (residual 3.10), whereas elective post-operative craniotomy patients were less likely to have a change in clinical status (residual −2.21). Although a patient’s diagnosis was associated with a change in clinical status, diagnosis was not associated with an alteration to the treatment plan (χ² = 6.36; p = 0.38).

Table 1.

Frequencies, percentages, and adjusted residuals (in parentheses) for change in clinical status by diagnosis. Adjusted residuals greater than ±1.96 (bolded) represent dependency between diagnosis and observing a clinical change.

	No change identified			Change identified
Diagnosis	f	%	Residual	f	%	Residual
Post-operative craniotomy	45	95.7%	(2.21)	2	4.3%	(− 2.21)
Subarachnoid hemorrhage	33	71.7%	(− 3.10)	13	28.3%	(3.10)
Ischemic stroke	26	86.7%	(0.14)	4	13.3%	(− 0.14)
Intracerebral hemorrhage	21	87.5%	(0.25)	3	12.5%	(− 0.25)
Other	20	95.2%	(1.30)	1	4.8%	(− 1.30)
Post-operative IR	12	75.0%	(− 1.30)	4	25.0%	(1.30)
Status epilepticus	8	80.0%	(− 0.55)	2	20.0%	(0.55)
Post-operative spine	7	100%	(1.10)	0	0.0%	(− 1.10)
Brain tumor without surgery	6	100%	(1.01)	0	0%	(− 1.00)
Not reported	4	80.0%	(− 0.38)	1	20.0%	(0.38)

DISCUSSION

This study is a preliminary attempt to describe how frequently neurological deterioration is identified by formal neuroassessments, whether deterioration impacts treatment plans, and if so, which changes to the treatment plan are pursued. We found that 14% of shifts included an identification of a neurological decline, and 67% of these changes were identified during a scheduled hourly neurocheck. However, in patients who had neurological deterioration, changes to the treatment plan were pursued 45% of the time. When pursued, the most common changes to the treatment plan included medication changes and/or obtaining additional imaging.

Additionally, we found that a patient’s diagnosis was associated with observing a change in neurological status. Specifically, patients who were status post elective craniotomy were less likely to have a neurological change, whereas patients with subarachnoid hemorrhage were more likely to have a neurological change, and some diagnoses had no clinical changes identified. This result suggests that the frequent neurological checks ordered across many diagnoses may be better utilized in specific diagnoses with a greater yield than others. Our study did not find a significant relationship between primary diagnosis and changes to the care plan in patients for whom a neurological change was identified.

These results are consistent with prior published data on neurochecks in stroke, where a retrospective study identified that a notable portion (25%) of neurological decline was identified outside of scheduled assessment times.¹² In contrast to the prior retrospective study focused on neurocheck success in identifying neurological decline after stroke, our present study noted a lower frequency of deterioration. The primary question asked by our study was how frequently neurological decline occurred across multiple diagnoses. Therefore, the different rates of neurological decline observed between the two studies may relate to the inclusion of seemingly stable elective post-operative patients in the current study.

Despite the important results presented herein, there are some limitations in this study to consider. Given the construct of the survey, it was impossible to determine when in a patient’s course their deterioration occurred (the temporal profile). For instance, did patients who experienced neurological deterioration do so early in their course, or after several days? This question is a topic area that has been explored previously in intracerebral hemorrhage,^4,13 and there is some literature to support frequent monitoring for short periods of time with diminishing return and increased risk for potential harm after 24–48 hours.^1,6 In addition, a single patient may have experienced neurological deterioration more than once. Furthermore, experience of the nurse completing the survey was not recorded. Nurses with more experience might have been able to identify neurological decline on a different temporal profile (e.g., earlier) or at a different rate (e.g., more often), though each instance of deterioration was validated by the medical team as well. In this study, it was assumed that there was high clinical concern for neurological deterioration even when the patient was ordered for ongoing hourly neurochecks. However, it is known from prior studies that this may not always be an accurate assumption.⁵

In any survey-based study there is a risk for self-selection bias. In this particular study, those who completed the survey (and those who did not) may have been motivated by personal reasons, leading to under- or over-reporting of neurological decline. This participation bias also may limit the generalizability of the results, though many diagnoses were included in the study population and represent a well-rounded neurocritical intensive care unit patient population.

Lastly, 212 twelve-hour shifts are represented by the data, but it is unclear whether this sample size is sufficient to draw definitive conclusions, especially given the heterogeneity of the sample. At the initiation of the study, data were sparse in the literature on neurochecks and neurodeterioration, providing little information on which to base a power analysis, and therefore we ultimately aimed for a sample size similar to prior studies.¹² In contrast to de Leon et al.,¹² our study represents nursing shifts rather than individual nurses or patients. We encourage further research utilizing a diverse population of patients with acute neurological injury, with variables to include hours from admission/surgery to deterioration, nurse identifier and patient identifier to potentially improve the generalizability of the data. A larger, multisite dataset is needed to examine causative relationships more definitively. It is not clear given the methods used in this study whether a change in treatment plan should have occurred in cases where it did not. Future study using ethnographic techniques by content experts is advised to overcome this methodological problem.

In this study, the scheduled, formal hourly neurochecks identified neurological decline in 20 (9%) instances. This finding suggests that there were a substantial number of patients for whom hourly neurochecks were ordered and performed but where patients did not experience a neurological decline. Furthermore, a change in neurological status resulted in a change in management only 45% of the time, suggesting that perhaps there is a way to be more selective in the patient population where hourly neurochecks are ordered. For instance, if a neurological change would not alter the care plan, then perhaps a patient could be spared hourly examinations.

Any care plan requires an assessment of risks and benefits. The benefit of frequent neuroassessments to avoid missing a neurological decline is paramount; however, the benefits are presumed in many situations, and the risks are largely unknown. This study adds to the literature the finding that a non-negligible percentage of neurological decline is noted outside of scheduled neurological assessment and/or does not change medical management. As the body of literature grows on the negative impact of sleep on physiological function,^14,15 there is increasing urgency for rigorous study of the risks and benefits of frequent neurochecks, as well as study of other biomarkers and modalities for predicting and monitoring for neurological decompensation, respectively.¹⁶ Additionally, hourly neurochecks are resource intensive; the burden to the nursing workforce in the context of the nursing shortage and burnout is also another consideration.¹⁷ Prior to changing practice, an evaluation of the risk of eliminating neurochecks in diagnoses where the yield is rare would need to occur. With this evaluation, the granular monitoring provided by frequent neurochecks would be weighed against potential iatrogenic negative impact on our patients’ recovery.

CONCLUSION

Our results indicate that 14% of nursing shifts witnessed neurological decline in a patient, with 67% of changes in neurological status identified on scheduled, formal hourly neurochecks. However, this change in neurological status resulted in an alteration to the treatment plan less than half the time. Further, the relationship between primary diagnosis and neurological changes suggests that hourly neurological checks may have greater yield for some diagnoses than for others. Further study is needed to delineate the ideal target population for neurochecks as well as the frequency and duration of serial neurological monitoring.