A search of the literature on the management of severe traumatic brain injury (sTBI) will reveal evidence-based guidelines and consensus-based recommendations supporting intracranial pressure (ICP) monitoring ( Table 61-1 ). These documents support the predominant position of academic neurotraumatologists in highly resourced medical environments that successfully lowering elevated ICP improves recovery by attenuating the morbidity associated with intracranial hypertension. The publication of a randomized controlled trial (RCT) that questions the efficacy of our current use of monitored ICP, as well as the economic and clinical challenges associated with ICP monitoring in less affluent environments, invites a review of the current evidence surrounding the necessity and utility of monitoring ICP in managing sTBI.
| BTF Guidelines for Adults |
|
|
| Milan Consensus Conference |
| Diffuse brain damage |
|
|
| Traumatic brain contusions |
|
|
|
| After decompressive craniectomy for intracranial hypertension (secondary DC) |
|
| After evacuation of intracranial traumatic hematomas (primary DC) |
|
∗ Comatose patients defined as patients without eye opening, not obeying commands, and not speaking understandable words after hemodynamic and respiratory stabilization and in the absence of anesthetic or paralyzing agents.
The utility of any monitor lies in its interpretation. Most ICP data are treated as end-hour ICP, reflecting the prior 60 minutes, although few studies actually specify their collection technique. End-hour ICP may be the instantaneous ICP value; a nurse’s subjectively derived value representing the prior hour; or, more rarely, an average that is based on some algorithm (often unspecified). None of these represent the way ICP is used clinically, which generally reflects instantaneous values, trends, responses to stimulation, spontaneous fluctuations, and the effect of treatment. More recently, higher resolution (real-time) ICP data have been collected, allowing averaging and trending as well as alternative analytic methods (e.g., area under the curve [AUC], ICP variability). The overall lack of rigor in the current literature has greatly hampered studies of the prognostic value of ICP, its optimal treatment threshold(s), the efficacy of treatment on altering outcome, and the definition of a “dose” of intracranial hypertension. Finally, the lack of natural history studies means that all ICP studies are performed with data from patients concomitantly treated at some threshold (most frequently 20 mm Hg). This confounds toxicities of treatment with the detrimental effects of intracranial hypertension, in particular greatly hampering analysis of the predictive value on ICP elevation and the determination of physiologic treatment thresholds. This is the evidentiary environment within which one approaches an analysis of the role of ICP monitoring in sTBI management.
Intracranial Pressure and Prognosis
In addition to the previously mentioned standardization considerations and the lack of natural history studies, the analysis of ICP as a predictive variable is confounded by irregularities in populations studied (e.g., excluding patients for “futility”), uncontrolled variability in management approaches, treatment toxicities, and mixed injury types.
Most studies evaluate the prognostic utility of ICP based on a set threshold. In general, such analyses support that intracranial hypertension is predictive of increased mortality. When morbidity is evaluated with mortality included, intracranial hypertension correlates with poor outcome (e.g., Glasgow Outcome Scale score [GOS] 1 to 3 or Extended GOS [GOS-E] 1 to 4), although this correlation appears to hold better for diffuse injuries than mass lesions. If morbidity is analyzed separately from mortality, then intracranial hypertension is frequently not predictive of poor-grade survival. A recent systematic review of ICP and outcome reported that the degree of intracranial hypertension (especially >40 mm Hg) was associated with unfavorable outcome if death was included but not for survivors alone. It was concluded that the pattern of ICP elevation and ICP that is refractory to treatment were more powerful predictors than peak values or threshold violations.
Analyzing the “refractoriness” of ICP involves evaluating the correlation between outcome and the response of intracranial hypertension to treatment. All such studies have used a 20–mm Hg treatment threshold and have reported significantly higher mortality for intracranial hypertension refractory to treatment. The systematic review of Treggiari et al. concluded that the odds ratios (ORs) of mortality and poor survivorship were significantly associated with intracranial hypertension refractory to treatment.
Overall, the absolute value of ICP appears to be of marginal utility as a prognostic variable, primarily acting as a marker of disease severity in terms of mortality. The pattern of resistance to treatment, particularly a refractory course, offers more prognostic power and is relevant to the critical value of quality of survival. However, its value as an independent predictive value remains unclear.
Use of Intracranial Pressure Monitoring and Outcome
Intracranial Pressure-Monitor-Based Management Protocols
The association between monitoring ICP and outcome has been evaluated as validation of its use. Studies from single centers not using monitoring showing no associated difference versus historical controls from monitoring centers have been too flawed to be conclusive. Small prospective and larger retrospective single-center reports of the influence of instituting sTBI protocols focused around ICP and cerebral perfusion pressure (CPP) management have uniformly demonstrated increased efficiency (e.g., decreased ventilator days, decreased number of treatments) and generally supported associated improvements in short-term outcomes. Care system modifications ranged from preprinted orders, through explicit algorithms, to formal care pathways. However, one small prospective study reported that initiating an sTBI protocol was not associated with improved outcome and that the significant benefits on management efficiency were statistically independent of whether patients underwent ICP monitoring. A similar, larger, two-paper retrospective series reported that the significant improvements in mortality associated with protocol initiation were independent of ICP monitoring, which was significantly associated with increased use of treatment modalities and intensive care unit (ICU) length of stay.
In aggregate, it appears that, if associated with adequate attention to protocol compliance, monitoring of deviations, and definition of the interventions (preprinted orders, flowcharts, management protocols, care pathways), the literature supports that the establishment and enforcement of protocols/care pathways focused on ICP-monitor-based management of sTBI patients can be expected to generally result in decreased resource use and improvements in short-term patient outcome. However, the degree and even direction of the specific contribution of ICP monitoring to these improvements is unclear.
Center-Based Studies
Multicenter studies of the association between ICP monitoring and outcome can be divided into center-based and patient-based approaches. Center-based studies focus on “aggressive care” as associated with more frequent ICP monitoring. Bulger et al. analyzed prospective data from 33 level I/II trauma centers, classifying those who monitored ICP in more than 50% of patients that met the Brain Trauma Foundation (BTF) guidelines monitoring criteria (GCS ≤ 8 and an abnormal admission computed tomography [CT] scan) as aggressive. Only 36% of centers met these criteria. This designation strongly covaried with the availability of traumatic brain injury (TBI)-related resources and personnel and with treatment intensity. Overall hospital mortality was significantly lower at “aggressive” centers, suggesting that this ICP monitoring frequency-based definition of aggressive care strongly covaries with practices supportive of improved survival.
This study contrasts with the report of Cremer et al., which retrospectively compared two level I trauma centers, one of which treated suspected intracranial hypertension based on imaging and clinical examination (ICE), the other predominantly directing care based on ICP monitoring (67% of studied patients). The monitoring center used significantly more resources, but there was no difference in survival-to-discharge between centers.
The contrasting results of these two frequently quoted, center-based studies suggest that, although the frequency of ICP monitoring may be a useful index of beneficially attentive care in multicenter studies, it should not be used in isolation as a marker of effective care of sTBI patients. Mauritz et al. found that the frequency of ICP monitoring varied according to center size (increasing from small to medium centers, then decreasing for larger institutions) and severity of injury (increasing, then decreasing as severity increased). It also varied by age. If aggressiveness of care is to be usefully studied, then it clearly warrants a more complex definition derived via multivariate analysis.
Patient-Based Studies
Multicenter, patient-based studies have analyzed large databases to investigate the association between ICP monitor insertion and outcomes. Two large studies have been reported based on general trauma databases. An analysis of 5507 patients with abbreviated injury scale (AIS) head scores greater than 3 from the Ontario Trauma Registry found that 9.8% were monitored, with a very wide center-specific range (0.5% to 21.4%). Multivariate analyses controlling for AIS head score, injury severity score (ISS), and injury mechanism indicated that ICP monitoring was associated with significantly improved survival. However, different results were gleaned from sTBI patients (GCS ≤ 8 and an abnormal CT) with ICU stays of 3 or more days from the National Trauma Data Bank. Worse risk-adjusted hospital mortality and discharge functional status as well as increased complications (pneumonia, renal failure, and infections) were reported for 708 monitored patients as compared with 938 nonmonitored patients. Only 43% of patients meeting the BTF guidelines criteria for ICP monitoring were actually monitored.
Although these reports presumably reflect “real-life” treatment of TBI patients at general trauma centers, they lack the neurologic indices necessary for rigorous adjustment for TBI severity. Their disparate findings under these conditions suggest that the use of ICP monitoring as a quality benchmark should be considered relative, not absolute.
Prospective, TBI-specific databases provide the best data for rigorous risk adjustment. Farahvar et al. prospectively analyzed collected data from 1307 sTBI patients who received treatment for intracranial hypertension within 48 hours of injury, 1083 (83%) of who were monitored. Nonmonitored patients were significantly older and had significantly more pupillary abnormalities. Controlling for age, GCS score, CT abnormalities, pupil abnormalities, and hypotension, multivariate logistic regression modeling of 2-week mortality in adults revealed a strong trend toward reduced risk for patients having ICP monitoring (OR, 0.64, 95%; confidence interval [CI], 0.41 to 1.00; P = .05).
Mauritz et al. attempted to model the decision to monitor by creating an ICP-monitoring propensity score derived on the basis of severity of injury indices. They applied this score to 1856 prospectively studied sTBI patients from 32 ICUs. When adjusted based on this model, they found no significant independent association between ICP monitoring and risk-adjusted discharge mortality. As noted previously, the frequency of ICP monitoring in this study varied according to age, center size, and severity of injury.
Meta-analysis of Patient-Based Studies
Stein et al. performed a meta-analysis of 127 sTBI patient-based studies, analyzing the influence on outcome of aggressive treatment based on ICP monitoring frequency. Their analysis was study-based; they did not use pooled data. They reported significant independent associations of improved outcome and decreased recovery associated with aggressive treatment.
Randomized Controlled Trials
The BEST TRIP (Benchmark Evidence from South American Trials: Treatment of Intracranial Pressure) trial, a recent RCT, compared the outcomes of patients managed according to a protocol based on monitored ICP versus a group treated for intracranial hypertension based on serial ICE without implanted monitors. Both groups were aggressively resuscitated and managed in small ICUs according to specified protocols by intensivists with special interest in neurotrauma, who did the serial examinations themselves. They reported no significant difference at 6 months in a composite outcome score combining mortality, morbidity, functional outcome, and neuropsychological testing. ICP monitor-based treatment was associated with significantly fewer ICU days of treatment for intracranial hypertension and 50% fewer individual ICP treatments. Both groups had equal incidences of neurologic deterioration.
Smith et al. performed a smaller RCT in which patients with ICP monitors were randomized to ICP-based mannitol administration with escalation to high-dose pentobarbital for refractory intracranial hypertension versus scheduled mannitol regardless of monitored ICP, escalated only for neurologic worsening. There was no difference between groups in dichotomized GOS scores at 1 year. ICP was 5.5 mm Hg higher in the group treated based on monitored ICP.
All of the these nonrandomized studies analyzing the association between ICP monitoring and outcome suffer from the inability to describe and control for the individual decision making related to the insertion of an ICP monitor in a particular patient. Each such decision reflects an unquantified admixture of individual evaluations, physician preferences, physician policies, institutional policies, and other approaches. The wide center-specific monitoring range in the Lane study, the low compliance with the BTF monitoring guidelines in the Shafi paper, and the low fraction (36%) of level I and II centers that inserted ICP monitors in more than 50% of patients meeting the BTF guidelines criteria for monitoring in the Bulger report highlight the variations in practice associated with this decision making. Age and severity of injury have been reported to be greater in patients treated for intracranial hypertension but not monitored when compared with those monitored. The size of the managing center may also influence monitoring decisions. The critical relationship between the decision to monitor, the perceived prognosis, and the intention to manage intracranial hypertension is uncontrolled in these studies. For instance, it should not be expected that an sTBI patient who is not monitored (and therefore managed expectantly) because of a perceived poor prognosis would have the same outcome as an intensively managed patient who is not monitored based on institutional practices. The vagaries in the association between ICP monitoring and outcome reflected in these studies likely reflects their inability to control for such decision making.
The process of randomizing the use of ICP monitoring removes the above decision making-related vagaries from the study. In addition, the specification of treatment protocols for all randomized groups defines the level of aggressiveness independent of the monitor. Neither of the RCTs supports an association between improved outcome and implantation of an ICP monitor if all patients receive intensive treatment of intracranial hypertension.
Of note, all of these studies examine ICP monitoring in the composite group of patients with sTBI rather than just the subset with established intracranial hypertension. Nevertheless, if aggressiveness of care is actually beneficial to improving outcome from sTBI in general, it does not appear that ICP monitor insertion is a sensitive or specific benchmark or quality assurance indicator.
Related posts:
What Lessons Have Intensivists Learned During the Evidence-Based Medicine Era?
What Is the Role of Invasive Hemodynamic Monitoring in Critical Care?
Should Fever Be Treated?
Are Computerized Algorithms Useful in Managing the Critically Ill Patient?
Inhaled Vasodilators in ARDS: Do They Make a Difference?
How Can We Monitor the Microcirculation in Sepsis? Does It Improve Outcome?
Full access? Get Clinical Tree
