| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Corresponding author: Catherine J. Kirkness, RN, PhD, Room T617, Biobehavioral Nursing and Health Systems, 1959 Pacific NE St, Box 357266, University of Washington, Seattle, WA 98195-7266 (e-mail: kirkness{at}u.washington.edu).
 |
Abstract |
|---|
 Top Abstract ObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
• Objective To evaluate the effect of a highly visible CPP display on immediate and long-term functional outcome in patients with traumatic brain injury.
• Methods A total of 157 patients with traumatic brain injury at a level 1 trauma center who had invasive arterial blood pressure and intracranial pressure monitoring were randomized to beds with or without an additional, prominent continuous CPP display. Primary end points were scores on the Extended Glasgow Outcome Scale (GOSE) and Functional Status Examination (FSE) 6 months after injury. Secondary end points were GOSE scores at discharge and 3 months after injury and FSE score 3 months after injury.
• Results Although GOSE and FSE scores at 6 months were better in the group with the highly visible CPP display, the differences were not significant. Slope of recovery for GOSE and FSE over all follow-up time points did not differ significantly between groups. However, the intervention’s positive effect on odds of survival at hospital discharge was strong and significant. Within a subgroup of more severely injured patients, the intervention group was much less likely than the control group to have CPP deviations.
• Conclusions The presence of a highly visible display of CPP was associated with significantly better odds of survival and overall condition at discharge.
|
Notice to CE enrollees: A closed-book, multiple-choice examination following this article tests your understanding of the following objectives:
To read this article and take the CE test online, visit www.ajcconline.org and click "CE Articles in This Issue."
|
Although nurses are aware of the importance of keeping CPP above a critical threshold, this awareness is generally in the context of 30-minute or hourly recordings of CPP. Current clinical bedside monitoring systems can display digital or analog "snapshots" of the immediate CPP, but the monitors are usually placed out of the clinician’s sightline, the snapshots are often small, and the CPP is often separated into an ABP display and an ICP display. Furthermore, current monitoring systems do not continuously display the trend in CPP over the past minutes or hours. Instead, a record of the trend must be printed out, often away from the patient’s bedside. Thus, changes in CPP that occur on a minute-to-minute basis, in relation to care, may not be visible or easily retrievable by critical care nurses to evaluate the effects of ongoing care.
| Even transient decreases in CPP may contribute to secondary brain injury.
|
Trends of declining CPP may not be recognized until CPP decreases below a critical level, thereby precluding nurses from preventing a serious decrease in CPP and requiring rescue rather than prevention. Therefore, we reasoned that highly visible, prominent displays of CPP information might enhance visibility and awareness of transient alterations in CPP related to care or of early trends of declining CPP. Immediate, sustained, and highly visible CPP information feedback would allow more rapid interventions such as elevation of a patient’s head to decrease ICP or adjustment in doses of sedatives or vasopressors to prevent or minimize suboptimal CPP. Therefore, we would expect to see fewer episodes of suboptimal CPP in patients with this highly visible display. To the extent that suboptimal CPP mediates secondary brain injury, we would expect to see improved short- and long-term functional outcomes in patients managed by using this display.
|
Objective |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
|
Methods |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
In addition to ICP management with sedatives and mannitol or furosemide, CPP management included elevation of ABP by administering fluids and pharmacological agents as needed, generally to maintain CPP above a threshold of 60 mm Hg. Patients’ outcomes were assessed at the time of discharge from the hospital and 3 and 6 months after injury by trained interviewers who were unaware what type of CPP monitoring each patient had (group assignment). Approval for the protection of human subjects was received from the University of Washington’s institutional review board; written consent for follow-up was obtained from the patients and/or their surrogates as appropriate.
Intervention and Randomization
Consecutive eligible patients admitted to the intensive care units during the study period were randomized to beds with either a highly visible active CPP display computer (intervention) or a blank CPP display computer (control) placed on a stand at the head of the bed. The active display consisted of a bar graph reflecting CPP during a half-hour period, with bars displayed in different colors depending on whether CPP was greater than or less than a threshold of 70 mm Hg (Figure 1
). The threshold of 70 mm Hg was selected on the basis of the recommendation from the guidelines for the management of severe TBI that were current when the study began.17 The active CPP display also included the numeric CPP value. The display was updated every 5 seconds. The control display showed a black screen with the message "Program Running. Please Do Not Adjust" (Figure 1).
|
Outcome Assessment
The primary end points were the scores on the GOSE and the FSE, assessed 6 months after the injury. Secondary end points were the GOSE score at discharge and 3 months after injury and the FSE score 3 months after injury. The GOSE extends the widely used Glasgow Outcome Scale18 to an 8-point scale by dividing each of the severe-disability, moderate-disability, and good-recovery categories additionally into an upper and a lower category.19,20 Scores on the GOSE range from 1 (dead) to 8 (upper category of good recovery); higher scores reflect better function. The FSE is used to assess changes that occur as a result of acute brain injury in relation to physical, social, and psychological functioning and financial independence.21,22 For each of 10 functional status categories, scores range from 0 (no change in functioning from before brain injury) to 3 (complete dependence). These scores are summed for a total score from 0 to 30; higher scores reflect poorer functioning. An FSE score of 31 is assigned to patients who have died. Thus, mortality is included in the scoring of both functional outcome measures. Scores on both the GOSE and the FSE are assessed by using standardized interview formats.
Measurement of Other Variables
Demographic (age, sex), diagnostic (computed tomography, magnetic resonance imaging), and management (medical, pharmacological, or surgical intervention) data were recorded from medical records. Postresuscitation Glasgow Coma Scale (GCS-PR) scores were recorded to reflect the severity of head injury.23 To account for other trauma-related injuries, scores on the Injury Severity Scale were also obtained.24,25 ABP was monitored via an intra-arterial catheter connected to Transpect pressure transducers (Abbott Laboratories, Abbott Park, Ill). ICP was measured via a Camino transducer-tipped catheter (Integra LifeSciences, Plainsboro, NJ) inserted intraparenchymally. The display computer received analog input from the bedside monitoring system (Spacelabs Medical, Redmond, Wash), sampling the signals at a rate of 100 Hz. CPP was calculated as mean ABP minus mean ICP. ICP and ABP data were saved to the display computers in 5-second summaries.
Statistical Methods
All data analyses were conducted by adjusting for age, sex, and severity of injury (GCS-PR and Injury Severity Scale) because of the strong influence on survival previously established for these variables. The Student t test and 
2 test were used to compare baseline characteristics. Both GOSE and FSE scores were treated as interval variables for the main statistical analyses. Logistic regression, with adjustments for age, sex, and severity of injury, was used when the GOSE score was analyzed as a categorical, dichotomous outcome (dead versus alive). We analyzed the overall intervention effect over time by using growth curve analysis26 based on hierarchical linear mixed modeling.27 This method allowed us to characterize the typical within-subject pattern of the outcome scale scores over time.
|
Results |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
. Most clinical and demographic factors were similar in the control and intervention groups. However, the percentage of patients with GCS-PR motor scores of 4 or less was significantly higher in the intervention group (P = .02), reflecting a greater proportion with the most severe head injury in this group. Scores on the Injury Severity Scale did not differ significantly between groups. Falls were a more frequent mechanism of injury in the intervention group than in the control group.
|
, Table 2). Adjusted FSE scores at 6 months were lower in the intervention group, also reflecting better overall function, but again the difference was not significant between the intervention group and the control group (Figure 3, Table 2).
|
|
|
, Table 2). FSE scores at 3 months after injury were nearly identical in the 2 groups (Figure 3, Table 2).
Change in Function Over Time
At discharge from the hospital, the median GOSE score of survivors reflected severe disability (unable to function for 24 hours without at least some assistance). By 6 months, the median GOSE score of survivors reflected functional status at a level of moderate disability, indicating independence in personal care but ongoing disability and inability to return to prior work. The FSE score is obtainable only at 3 and 6 months and shows a similar but more detailed pattern of increasing independence but not full recovery. When these outcomes were tested with hierarchical linear modeling, no significant difference was found in the slope of the recovery trajectory during the 6 months after injury, according to either GOSE or FSE scores (P = .88 and P =.64, respectively; Figures 2 and 3; Table 3).
|
). The data on discharge disposition suggest that this finding largely reflects the lower number of in-hospital deaths in the intervention group. A total of 17% of the control group died in the hospital versus 10% of the intervention group; 8% of the control group was discharged to home versus 13% of the intervention group. Similar percentages of each group were discharged to rehabilitation (50% control, 47% intervention) or skilled nursing facilities (23% control, 27% intervention).
|
).
|
However, exploratory analyses of the role of CPP thresholds and the percentage of time that CPP was less than the threshold values revealed associations between outcome and visibility of CPP feedback. Analyses of receiver-operating-characteristic curves were used to maximize sensitivity and specificity of specific CPP cutoff points in relation to survival. We used these cutoff points to specify lesser and greater percentages of time that CPP decreased below 70 mm Hg and 60 mm Hg in the first 4 days of monitoring. (See the Appendix for further details about specifying cutoff points.)
When the full sample (N = 157) was considered, patients randomized to the control group were more likely to have deviations to less than 60 mm Hg than were patients in the intervention group (44.2% control vs 30.8% intervention, 2=2.96, P =.08). The trend was less strong for mean CPP less than 70 mm Hg (39% control vs 31% intervention, 2 = 0.81, P = .37). The relationships of deviations to less than the CPP threshold to randomization group were much stronger in the more severely injured subgroup. When the relationship of CPP level at specified cutoff points at values less than the threshold values was explored for the subgroup with GCS-PR motor scores of less than 4, a clear advantage was apparent for those in the intervention group. In the control group, 63% with low GCS-PR motor scores had a mean CPP less than 60 mm Hg for a critical period, compared with 19% in the intervention group (2 = 7.29, P = .007; Figure 4). Similarly, 50% of the control group had CPP deviations to less than 70 mm Hg for a critical period, compared with 14.3% in the intervention group (2=5.54, P=.02).
|
These data suggest that the difference in the proportion of patients within the more severely injured subgroup whose CPP decreased to values less than the threshold values may have been a factor in the greater survival rate of patients in the intervention group. This survival effect does not appear to be due to shifting of the poorest outcomes after discharge, because few survivors were in a vegetative state in the intervention group, either overall or in the most severely injured subset. Further, survivors in the intervention group were represented at 3 and 6 months in the highest functioning levels as well as among the more severely disabled.
|
Discussion |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
The mechanism by which patients in the intervention group were in better condition at discharge appears to be mediated in part through CPP, at least for the most severely injured subgroup. The influence of physiological parameters on outcome may be most strongly apparent in the early phase after injury, particularly affecting mortality. The influence on long-term outcome of survivors may be less apparent. Strong evidence, both from previous studies28–30 and this one, indicates that CPP level as a major predictor of outcome. In the only study31 in which CPP was actively manipulated through aggressive management of blood pressure, the intervention group had a designed increase in CPP but also had an increase in the incidence of acute respiratory distress syndrome and no benefit in long-term outcome.
In our study, we manipulated the visibility of CPP information, with the assumption that the response of nurses to this CPP information would be reflected by the use of traditional management strategies: reduction of ICP with sedation, diuretics, and body position, and support of blood pressure with vasoactive drugs. Any brief periods of CPP at values less than the threshold value remained visible to nurses and physicians up to one half hour with the active display. This visibility of CPP levels over time may have encouraged greater attention to overall management. By contrast, this information was lost to caregivers unless the electronic record was consulted when only the blank display and the usual bedside display were present.
The primary limitation of this study was the inability to measure clinicians’ behavior in response to changes in CPP displayed on the active monitors. Therefore, we can only speculate about whether nurses, physicians, and respiratory therapists used the information that was visible to them to tailor the care patients received. Kenner32 conducted a small thesis study of nursing actions in response to changes in CPP while the present study was in progress. She included 5 patients with the active CPP display, 6 with the blank display, and 6 with the unit’s usual bedside display. Because most episodes of sustained decreased CPP occurred when a nurse was absent from a patient’s room, the information about CPP was lost to clinicians whose patients had the usual monitoring display.32 Kenner did not query nurses about whether or how they were using information visible on the monitor when CPP was within normal ranges.
| The use of a real-time display of CPP resulted in better patient condition at the time of hospital discharge.
|
With the data available, we cannot be certain of the mechanism behind the improved survival in the intervention group. In addition, we were unable to measure the many potential factors beyond the severity of the initial injury and the hospital course, such as social support and access to rehabilitation, that may have played an important role in influencing outcome at 3 and 6 months in TBI survivors in this study.
Overall, mean CPP levels during the study were almost 80 mm Hg; CPP therefore was not less than the display threshold of 70 mm Hg most of the time. Because of the complex pathophysiology of brain injury—perhaps unless excursions to extreme ranges occur—monitoring and management of any single parameter would be insufficient alone to result in a significant independent improvement in outcome, particularly long-term outcome. Further, the increasing use of decompressive craniectomy in neurosurgical practice maintains CPP at or above minimum desired levels in patients so treated, leaving a smaller subset for whom the careful tracking and adjustment of CPP is more crucial. In such cases, the persistence of the visible display of decreases in CPP for up to a half hour in the active display group may have provided important cues to the nursing and medical staff.
Although the threshold in the active display was set to 70 mm Hg, analysis of the data showed that the most critical cutoff point for CPP was 60 mm Hg, consistent with the level recommended in the 2003 revision of the guidelines.33 Because the percentage of time at values less than the threshold value that was related to poorer survival was small, setting the display threshold higher than 60 mm Hg to warn that CPP is approaching the critical threshold may be useful. This adjustment would allow time to carry out interventions before CPP actually decreased to less than that threshold.34
|
Conclusions |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
Appendix: Details of Statistical Methods
Minimum sample size at 80% power to detect a clinically meaningful effect size of improvement of 1 level on the Extended Glasgow Outcome Scale (GOSE), with entry covariates of injury severity, age, and sex, was calculated as 128 (64 in each treatment group). Over-sampling was done to ensure sufficient power if sizable numbers of participants were lost to follow-up and to allow subgroup analysis.
The statistical package HLM 5 (Scientific Software International, Lincolnwood, Ill) was used to specify a within-subject linear growth model and an across-subject model. The within-subject model had 2 parameters: intercept and slope. GOSE score was assessed at all time points, but the score on the Functional Status Examination (FSE) was assessed only at the 3-month and 6-month time points because of categories such as financial independence. At the across-subject hierarchy, the 2 implied within-subject coefficients (intercept, slope) were considered to be dependent on the randomized assignment to the intervention or control group.
Also incorporated in the across-subject model were the demographic covariates age and sex and variables reflecting severity of the initial injury, including the motor component of the Postresus-citation Glasgow Coma Scale and the score on the Injury Severity Scale. The conclusions about the intervention effect over time are based on interpretation of the regression coefficients associated with randomization group and the intervention effect on the intercept (estimated outcome at discharge) and the slope (estimated linear change over time in 3-month increments). The time index was adjusted to reference the within-subject level-1 intercept to the hospital discharge time point for the GOSE analyses and to 3 months for the FSE analyses.
Although credible partial information about 6-month status (dead/alive or dead/bad/good outcome) was available for nearly all subjects, approximately 9% of the study participants were missing GOSE or FSE outcome information at either the 3-month or the 6-month follow-up time points. Multiple imputation methods35,36 were used to generate plausible values of the missing data consistent with the available partial information about each of these patients. Nine representative complete case data sets were created by using the MICE (Multiple Imputation by Chained Equations, van Buuren and Oudshoorn, TNO Prevention and Health Public Health Institute, Leiden, The Netherlands, 2000) software package. All the covariates to be used in the analysis model (age, sex, severity of initial injury), except design contrast factors (randomization group and cerebral perfusion pressure [CPP] level groups), and all available partial information about 6-month status, were incorporated into the imputation model. The Predictive Mean Matching tabular imputation algorithm37 was used to ensure that the values that were imputed for the strictly integer GOSE and FSE scales remained whole numbers within the appropriate range for each variable. A total of 21 iterations of the imputation method were executed to ensure convergence of the algorithm. The linear mixed model longitudinal growth curve analyses reported in this article were executed for each of the 9 resulting imputed complete case data sets, and the results were pooled by using the HLM 5 software package.
For analyses less tractable to multiple imputation methods, the 9 imputed data sets were aggregated into a single representative complete case data set by computing the median of each cell across the imputations. Taking the median of an odd number of imputations determined by using the Predictive Mean Matching algorithm for a given cell ensures that the replacement value will be a whole number within the appropriate range for the variable.
Coenrollment
A total of 58 subjects (37%) enrolled in this study were also enrolled in the Magnesium Sulfate for Neuroprotection After Brain Trauma study, a randomized, double-blind controlled trial to evaluate the effect of early administration of magnesium following traumatic brain injury on survival, posttraumatic seizures, and functional status and neurobehavioral functioning. Analyses were carried out to assess for any potential confounding related to coenrollment in the 2 studies. Assignment to the experimental or control arm of the magnesium sulfate study was not significantly different between the intervention and control arms of this study. The outcome analysis for this study was run with and without a dummy variable coding the magnesium sulfate study randomization group assignment. Results were almost identical and did not change any conclusions. Therefore, the magnesium sulfate investigators (and we) concluded that there was no confounding by that study. Participants in the magnesium sulfate study were included in all analyses reported here.
Cutoff Points for CPP Level Related to Survival
For the CPP level of 70 mm Hg, the cutoff point was specified at 31% of monitoring time in which CPP was less than 70 mm Hg. For the CPP level of 60 mm Hg, the cutoff point was 6% of monitoring time in which CPP was less than 60 mm Hg. Binary logistic regression was performed by using the dichotomous dependent variable survival (dead or alive) by the categorical lower or higher percentage of time that CPP was less than threshold values, again with each patient’s age, sex, and severity of the initial injury controlled for.
|
ACKNOWLEDGMENTS |
|---|
To purchase reprints, contact The InnoVision Group, 101 Columbia, Aliso Viejo, CA 92656. Phone, (800) 809-2273 or (949) 362-2050 (ext 532); fax, (949) 362-2049; e-mail, reprints{at}aacn.org.
|
REFERENCES |
|---|
|
TopAbstractObjectiveMethodsResultsDiscussionConclusionsReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
