Traumatic brain injury (TBI) is a common cause of death and disability worldwide.17 Despite some claims to the contrary, no clear decrease in severe TBI-related mortality or improvement in overall outcome has been observed over the past 2 decades.24
Recently, many studies have shown that the presence and frequency of high intracranial pressure (ICP) are predictive of the outcome of severe TBI.18,23 Thus, ICP monitoring has been widely used in the management of severe TBI. Indeed, the Brain Trauma Foundation (BTF) guide-lines recommend ICP monitoring in all patients who have suffered severe TBI (Glasgow Coma Scale [GCS] score < 9) with a CT scan revealing intracranial pathology (Level II recommendation) or in severe TBI patients with a normal CT scan but with 2 or more of the following risk factors: age older than 40 years, unilateral or bilateral motor posturing, or systolic blood pressure lower than 90 mm Hg (Level III recommendation).4,6 Despite these recommendations, significant variability remains in the practice of ICP monitoring.5,19,22,26,29 T his m ay r eflect c onflicting or absent clinical evidence regarding the benefit of ICP monitoring on outcomes in TBI patients.
Although a trial of ICP monitoring in patients having suffered TBI (a large multicenter randomized study) was published in 2012, controversy remains due to issues with the design of that trial.7 That study included only a subset of patients in South America, where the use of ICP monitoring is generally low.25
Although a Cochrane review on the topic was published recently, no study was analyzed because only randomized trials were included.9 Without considering observational studies, the review was unable to provide any conclusion as to the value of ICP monitoring in patients with TBI. Thus, we conducted a meta-analysis of ICP monitoring studies, including observational studies, to evaluate the evidence of a relationship between the use of ICP monitoring and outcome in patients with TBI.
Methods
Data Sources and Searches
We searched MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials (Central) from their inception to October 2013 for relevant studies using appropriate combinations of MeSH terms and key words, including intracranial pressure, intracranial hypertension, ICP monitor, intracranial pressure monitoring, TBI, traumatic brain injury, brain injury, and craniocerebral trauma. We performed this relatively wide search to include the maximum number of relevant patients.
Study Selection
Randomized clinical trials and prospective cohort, retrospective observational cohort, and case-control studies that compared ICP monitoring with no ICP monitoring for the treatment of TBI were included in our analysis. Studies that did not include patients who did not undergo ICP monitoring as the comparator were excluded. Also excluded were articles not containing original research (e.g., narrative reviews, editorials, and case reports). Studies included had to report at least one point of mortality in an ICP monitoring group and a no–ICP monitoring group.
Data Abstraction
According to preestablished eligibility criteria, 2 reviewers (Q.Y. and X.W.) independently reviewed all citations and selected eligible studies using a standardized data abstraction form. Disagreements were resolved by consensus. Data were extracted for 1) study characteristics (author and year of publication, country of origin, study design, and sample size), 2) patient demographics (age and sex) and baseline characteristics (severity of injury), 3) treatment details (ICP monitoring or not, type of ICP monitor used, and duration of ICP monitoring), and 4) study outcomes (mortality, intensive care unit [ICU] length of stay [LOS], and hospital LOS in different groups). In instances of duplicate reporting, we used data from the study that included the largest number of patients or, when available, individual patient data from each study. We contacted authors for clarification on study samples or for missing data.
Methodological Quality of Included Studies
We used the criteria for reporting observational studies proposed in the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement to complete a methodological evaluation of the included observational studies.32
Statistical Analysis
Most statistical analyses were performed using the Review Manager Software (version 5.2, Cochrane Collaboration). All tests were 2-sided, and the significance level was set at 0.05. Statistical analyses for continuous variables or dichotomous variables were conducted. Mean differences (MDs) were used for the analysis of continuous variables (ICU LOS and hospital LOS), and odds ratios were used for dichotomous variables (mortality). The distribution of ICU LOS and hospital LOS were right-skewed and we thus log-transformed them to yield a normal distribution.3 A log-normal distribution facilitated the analysis and presentation of outcomes between groups. The association between ICP monitoring and TBI mortality was expressed as odds ratios with 95% confidence intervals. If the value “1” was excluded from the 95% confidence interval, the point estimate of the odds ratios was considered to have reached statistical significance (p < 0.05). The generic inverse variance method was used to analyze the odds ratio. The I-squared (I2) statistic was used to measure the extent of inconsistency among the results.15 Heterogeneity was detected using the chi-square test. Because the chi-square test lacks power when the number of studies is small, we considered significant heterogeneity to be present when both the chisquare value was within the 10% level of significance (p < 0.10) and the I2 value exceeded 50%.
In cases of heterogeneity, summary measures of the effect of ICP monitoring for mortality were obtained by conducting a random-effects meta-analysis of the best-effect estimate available from each study, which assumes that studies were a random sample of a hypothetical population of studies and assigns a weight to each study, taking into account variance within and between studies. When an adjusted odds ratio was available from the paper, it was taken as the best estimate; otherwise, the crude estimate was used. The extent to which the combined risk estimate might be affected by individual studies was assessed by consecutively omitting each study from the meta-analysis. Subgroup analysis considering more homogeneous sets of studies was used as an additional sensitivity test. Publication bias (linked to the fact that negative trials are cited less frequently and are therefore more likely to be missed in the search for relevant studies) was assessed using the “funnel plot” technique, based on a graph plotting effect estimates against sample size. Additionally, Egger regression and the Begg and Mazumdar methods were used to evaluate publication bias using Stata software (version 10, Stata Corp.). Multiple meta-regression was performed using Stata 10 to identify the factors causing the heterogeneity.
Quality of Evidence
Studies were graded and assigned a quality rating with respect to the key question according to the CEBM (Centre for Evidence-Based Medicine) criteria. Studies were graded from Level 1 (strongest evidence, such as reports with clinical decision rules and high-quality validating studies) to Level 5 (weakest evidence, often expert opinion).14
Results
Study Selection
In total, 7175 studies were identified and screened for retrieval using the strategy described above. After screening the title and abstract, 7047 studies were excluded and 128 were retrieved and subjected to detailed evaluations. Based on the inclusion and exclusion criteria, 114 of those studies were excluded, and thus 14 studies were included in the meta-analysis (Fig. 1). All included studies were published, peer-reviewed papers.
Identification process for eligible studies. Of the 7175 studies initially identified from our electronic search, 14 met the inclusion criteria and were included in this meta-analysis.
Study Characteristics
Thirteen studies were observational cohort studies and 1 was a randomized controlled trial. Tables 1 and 2 describe the 14 eligible studies according to their design. Nine studies conducted a multivariate analysis with an adjusted odds ratio of mortality in the ICP monitoring group compared with the no–ICP monitoring group. The randomized controlled trial was conducted in South America. Of the observational cohort studies, 7 were conducted in North America (US or Canada), 2 in Austria, 2 in China, 1 in Saudi Arabia, and 1 in the Netherlands. Nine studies contained information on in-hospital mortality, 4 on ICU mortality, 3 on 6-month mortality, 2 on 2-week mortality, and 1 study contained information on 28-day mortality. Six studies contained enough information on ICU LOS to be included in the meta-analysis, and 4 contained information on hospital LOS.
Characteristics of the studies included in the meta-analysis
| Authors & Year | No. of Patients | Country | Study Design | Inclusion Criteria | Type of ICP Monitor | Level of Evidence (CEBM) |
|---|---|---|---|---|---|---|
| Lane et al., 2000 | 5507 | Canada | Retrospective observational cohort | AIS head >3 | Not specified | 4 |
| Mauritz et al., 2007 | 415 | Austria | Retrospective observational cohort | GCS <9 | Intraparenchymal 77%; EVD 10%; epidural 3%; unknown 2%; combined 8% | 4 |
| Mauritz et al., 2008 | 1856 | Austria | Observational cohort | GCS <9 | Not specified | 2b |
| Thompson et al., 2008 | 1776 | US | Retrospective observational cohort | Any TBI ICD-9 code; age 24–65 yrs | Not specified | 4 |
| Shafi et al., 2008 | 1646 | US | Retrospective observational cohort | GCS <9 | Not specified | 4 |
| Griesdale et al., 2010 | 171 | Canada | Retrospective observational cohort | GCS <9 | EVD 100% | 3b |
| Haddad et al., 2011 | 477 | Saudi Arabia | Retrospective observational cohort | GCS <9 | Not specified | 3b |
| Farahvar et al., 2012 | 1307 | US | Prospective, observational cohort | GCS <9 | Not specified | 2b |
| Biersteker et al., 2012 | 265 | Netherlands | Prospective, observational cohort | GCS <9 & meeting BTF criteria for ICP monitoring | Not specified | 2b |
| Talving et al., 2013 | 216 | US | Prospective, observational cohort | AIS head >3 & GCS <9 & meeting BTF criteria for ICP monitoring | Becker ventriculostomy EDMS II 59%; transducer-tipped pressure/temperature fiberoptic catheter Camino Advanced Monitor (bolt) 41% | 2b |
| Alali et al., 2013 | 10,628 | US & Canada | Retrospective observational cohort | AIS head >3 & GCS <9 & meeting BTF criteria for ICP monitoring | Not specified | 3b |
| Gao et al., 2013 | 36 | China | Retrospective observational cohort | GCS <9 | Not specified | 4 |
| Zeng et al., 2013 | 168 | China | Prospective, observational cohort | GCS ≤12 | Not specified | 4 |
| Chesnut et al., 2012 | 324 | Bolivia & Ecuador | Randomized-controlled trial | GCS <9 on admission or a higher score on admission that dropped to <9 w/in 48 hrs after injury | Intraparenchymal monitor 100% | 1b |
AIS = Abbreviated Injury Scale; CEBM = Centre for Evidence-Based Medicine; EDMS = External Drainage and Monitoring System; EVD = external ventricular drain; ICD-9 = International Classification of Diseases, Ninth Revision.
Outcomes in the ICP monitoring and no–ICP monitoring groups of the studies included in the meta-analysis
| Authors & Year | No. of Patients | Dead/Total Cases in ICP Group | Dead/Total Cases in No ICP Group | ICU LOS in ICP Group vs in No ICP Group | Hospital LOS in ICP Group vs in No ICP Group | Crude OR (95% CI) of Mortality for ICP Monitoring | Adjusted OR (95% CI) of Mortality for ICP Monitoring | Confounding Factors Being Adjusted |
|---|---|---|---|---|---|---|---|---|
| Lane et al., 2000 | 5507 | 153/541 (in-hospital mortality) | 1222/4946 (in-hospital mortality) | mean 9.7/4.3 | mean 44.0/22.8 | 1.23 (0.99–1.46), in-hospital mortality | 0.77, in-hospital mortality | Mechanism of injury; AIS head; ISS |
| Mauritz et al., 2007 | 415 | 78/264 (ICU mortality) | 50/151 (ICU mortality) | — | — | 0.85 (0.84–0.87), ICU mortality | Not specified | Age; ISS; GCS |
| Mauritz et al., 2008 | 1856 | 361/1031 (ICU mortality); 402/1031 (in-hospital mortality) | 281/825 (ICU mortality); 314/825 (in-hospital mortality) | 18 (8–31)/9 (5–17)† | — | 1.04 (0.87–1.25), ICU mortality; 1.04 (0.86–1.26), in-hospital mortality | Not specified | SAPS II (includes age, GCS); sex; AIS head; large volume replacement; isolated TBI; volume of TBI per center |
| Thompson et al., 2008 | 1776 | 181/616 (in-hospital mortality)* | 388/3635 (in-hospital mortality)* | — | — | 0.94 (0.79–1.12), in-hospital mortality | 0.94 (0.81–1.09), in-hospital mortality | ISS; GCS-M; midline shift; pupillary response; AIS head; sex; Charlson score |
| Shafi et al., 2008 | 1646 | 149/708 (in-hospital mortality) | 113/938 (in-hospital mortality) | — | 25 (0.8)/22 (0.7)‡ | 1.82 (1.40–2.38), in-hospital | 1.82 (1.30–2.54), in-hospital | ISS; RTS; AIS head; craniotomy; GCS-M; spine injuries;pre -morbid cardiac disease;ICU complications |
| Griesdale et al., 2010 | 171 | 28/98 (in-hospital mortality); 22/98 (28-day mortality) | 9/73 (in-hospital mortality); 9/73 (28-day mortality) | 14 (11)/6 (8)‡ | — | 2.84 (1.25–6.48), in-hospital mortality; 2.06 (0.89–4.80), 28-day mortality | 2.80 (1.10–7.12), in-hospital mortality; 2.10 (0.80–5.60), 28-day mortality | Age;APACHE; positive CT findings;mannitol; yr of admission; hypotension;hypoxia; craniotomy |
| Haddad et al., 2011 | 477 | 11/52 (in-hospital mortality); 7/52 (ICU mortality) | 65/425 (in-hospital mortality); 49/425 (ICU mortality) | 17.0 (8.8)/11.6 (8.3)‡ | 86.7 (81.9)/78.4 (332.2)‡ | 1.19 (0.51–2.79) ICU mortality; 1.49 (0.73–3.04) in-hospital mortality | 1.01 (0.41–2.45), ICU mortality; 1.71 (0.79–3.70), in-hospital mortality | Age; APACHE II; admission GCS; isolated TBI; postop status |
| Farahvar et al., 2012 | 1307 | 212/1084 (2-wk mortality) | 74/223 (2-wk mortality) | — | — | 0.49 (0.36–0.67), 2-wk mortality | 0.64 (0.41–1.00), 2-wk mortality | Age; initial GCS score; hypotension present on Day 1; CT scanning findings; pupil abnormalities on Day 1 |
| Biersteker et al., 2012 | 265 | 59/123 (6-mo mortality) | 52/142 (6-mo mortality) | 10.8 (4.2–21)/2.65 (1.00–6.9)† | 22.0 (8.3–44)/7.48 (1.9–20)† | 1.60 (0.98–2.61), 6-mo mortality | 0.93 (0.47–1.85), 6-mo mortality | Propensity score (the probability of a patient receiving an ICP monitor given a range of baseline demographic & clinical variables) |
| Talving et al., 2013 | 216 | 33/101 (in-hospital mortality) | 62/115 (in-hospital mortality) | 16.8 (1.3)/8.2 (1.0)‡ | 19.4 (1.6)/10.1 (1.2)‡ | 0.41 (0.24–0.72), in-hospital mortality | 0.15 (0.03–0.74), in-hospital mortality | Head AIS; age; early nutrition; best GCS score w/in 24 hrs of admission; CT scan findings; sex; hypotension on admission; tachycardia on admission; extremity AIS score; chest AIS score; ISS; admission PT, PTT, & INR values; alcohol intoxication; GCS 3 at admission; pupils on admission |
| Alali et al., 2013 | 10,628 | 592/1874 (in-hospital mortality) | 3177/8754 (in-hospital mortality) | — | — | 0.81 (0.73–0.90), in-hospital mortality | 0.44 (0.31–0.63), in-hospital mortality | Age; sex; mechanism of injury; GCS in ED; head AIS;CT scanning findings; hypotension on admission; comorbid illnesses; type of insurance; hospital-level characteristics |
| Gao et al., 2013 | 36 | 5/18 (mortality)§ | 4/18 (mortality)§ | — | — | 1.35 (0.30–6.15), ICU/in-hospital mortality | — | — |
| Zeng et al., 2013 | 168 | 2/77 (6-mo mortality) | 10/91 (6-mo mortality) | — | — | 0.22 (0.05–0.99), 6-mo mortality | — | — |
| Chesnut et al., 2012 | 324 | 33/157 (2-wk mortality); 56/144 (6-mo mortality) | 50/167 (2-wk mortality); 67/153 (6-mo mortality) | 12 (6–17)/9 (6–16)† | — | 0.62 (0.37–1.03), 2-wk mortality; 0.82 (0.51–1.30), 6-mo mortality | — | — |
APACHE = Acute Physiology and Chronic Health Evaluation; ED = emergency department; GCS-M = GCS motor response; INR = international normalized ratio; ISS = injury severity score; PT = prothrombin time; PTT = partial thromboplastin time; RTS = revision of trauma score; SAPS II= Simplified Acute Physiology Score II; — = not reported.
Weighted number.
Presented as the median (interquartile range).
Presented as the mean (SD).
In-hospital and ICU mortality combined.
Methodological Quality of Studies Included
In the 13 observational cohort studies, 4 presented a flow diagram for participants, 13 studies adequately described their study population (including missing data and patients lost during follow-up), and 5 presented their funding sources. Table 3 provides a more complete evaluation of methodological quality.
Methodological quality of included studies (STROBE criteria)
| Variable | Authors & Year | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lane et al., 2000 | Mauritz et al., 2007 | Mauritz et al., 2008 | Thompson et al., 2008 | Shafi et al., 2008 | Griesdale et al., 2010 | Haddad et al., 2011 | Farahvar et al., 2012 | Biersteker et al., 2012 | Talving et al., 2013 | Alali et al., 2013 | Gao et al., 2013 | Zeng et al., 2013 | |
| Title/abstract | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Introduction | |||||||||||||
| Background/rationale | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Objectives | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Methods | |||||||||||||
| Study design | • | • | • | • | • | • | • | • | • | • | • | ||
| Setting | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Participants | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Variables | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Data sources/measurement | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Bias | • | • | • | ||||||||||
| Study size | |||||||||||||
| Quantitative variables | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Statistical methods | • | • | • | • | • | • | • | • | • | • | • | • | |
| Results | |||||||||||||
| Participants | • | • | • | • | |||||||||
| Descriptive data | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Outcome data | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Main results | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Other analyses | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Discussion | |||||||||||||
| Key results | • | • | • | • | • | • | • | • | • | • | • | • | • |
| Limitations | • | • | • | • | • | • | • | • | • | ||||
| Interpretation | • | • | • | • | • | • | • | • | • | ||||
| Generalizability | • | • | • | • | • | • | • | • | • | ||||
| Funding information | • | • | • | • | • | ||||||||
Outcome Measures
The studies were clearly heterogeneous (I2 = 82%, p < 0.001). Heterogeneity was not eliminated by considering only observational cohort studies, only studies reporting in-hospital mortality, or only studies including severe TBI patients (I2 = 83%, df 12, p < 0.001; I2 = 84%, df 8, p < 0.001, and I2 = 82%, df 10, p < 0.001, respectively). Because of this heterogeneity, a random-effects model was used.
Fourteen studies compared mortality due to TBI in the ICP monitoring group and the no–ICP monitoring group, with ORs ranging from 0.15 to 2.80. The meta-analysis provides no evidence that ICP monitoring decreased the risk of death (pooled OR 0.93 [95% CI 0.77–1.11], p = 0.40). After excluding the randomized controlled trial, the 13 observational cohort studies evaluated the association between ICP monitoring and mortality in TBI patients and also showed no evidence for decreased risk of death with ICP monitoring (pooled OR 0.95 [95% CI 0.79–1.15], p = 0.61; Fig. 2).
Association between ICP monitoring and mortality in patients with TBI stratified by randomization or not (shown as log[Odds Ratio] with SE). IV = inverse variance. Figure is available in color online only.
However, 7 of the studies including 12,944 patients were published after 2012 (January 2012 to October 2013). These studies revealed that ICP monitoring was significantly associated with a greater decrease in mortality than no ICP monitoring (pooled OR 0.56 [95% CI 0.41–0.78], p = 0.0006; Fig. 3), although the other 7 studies, published before 2012, showed that ICP monitoring was not significantly associated with a greater decrease in mortality (pooled OR 1.15 [95% CI 0.94–1.40], p = 0.17; Fig. 3). In addition, 7 of the studies conducted in North America showed no evidence that ICP monitoring decreased the risk of death, similar to the studies conducted in other regions (Fig. 4).
Association between ICP monitoring and mortality in patients with TBI stratified by publication date (shown as log[Odds Ratio] with SE). Figure is available in color online only.
Association between ICP monitoring and mortality in patients with TBI stratified by country (shown as log[Odds Ratio] with SE). Figure is available in color online only.
Regarding the 11 studies in which only severe TBI patients (GCS score < 9) were evaluated, a meta-analysis revealed that ICP monitoring was not significantly associated with a greater decrease in mortality than no ICP monitoring at the end of follow-up (pooled OR 0.91 [95% CI 0.71–1.17], p = 0.47; Fig. 5). The following subgroup analysis showed that ICP monitoring was significantly associated with a greater decrease in mortality than no ICP monitoring in the studies published after 2012 (pooled OR 0.59 [95% CI 0.42–0.81], p = 0.001; Fig. 5), but was not associated with a greater decrease in those published before 2012 (pooled OR 1.29 [95% CI 0.93–1.79], p = 0.13; Fig. 5).
Association between ICP monitoring and mortality in patients with severe TBI (GCS score < 9) stratified by publication date (shown as log[Odds Ratio] with SE). Figure is available in color online only.
Nine of the studies including 22,313 patients reported data on in-hospital mortality at the end of follow-up and revealed that ICP monitoring did not decrease the risk of in-hospital mortality (pooled OR 1.06 [95% CI 0.80–1.42], p = 0.68). Four studies contained information on ICU mortality, and 3 studies contained information on 6-month mortality. Meta-analyses showed that ICP monitoring was not significantly associated with a greater decrease in either ICU or 6-month mortality. However, only 2 studies contained information on 2-week mortality; ICP monitoring did decrease the risk of 2-week mortality (pooled OR 0.63 [95% CI 0.45–0.88], p = 0.007; Fig. 6).
Association between ICP monitoring and ICU mortality, hospital mortality, 2-week mortality, and 6-month mortality in patients with TBI (shown as log[Odds Ratio] with SE). Figure is available in color online only.
Six of the studies including 3309 patients in the metaanalysis reported data on ICU LOS. ICU LOSs were significantly longer for the group subjected to ICP monitoring (MD 0.29 [95% CI 0.21–0.37], p < 0.00001; Fig. 7). Four of the studies including 2604 patients in the metaanalysis reported data on hospital LOS. In the pooled data, the hospital LOS with ICP monitoring was significantly longer than with no ICP monitoring (MD 0.21 [95% CI 0.04–0.37], p = 0.01; Fig. 8).
Association between ICP monitoring and ICU LOS (shown as mean with SD in transformed days) in patients with TBI. Figure is available in color online only.
Association between ICP monitoring and hospital LOS (shown as mean with SD in transformed days) in patients with TBI. Figure is available in color online only.
To investigate potential sources of heterogeneity, several analyses were performed. The leave-one-out procedure revealed that no single study was responsible for the heterogeneity (heterogeneity tests were always significant when one study at a time was excluded).
In performing the meta-regression, we considered the following covariates: year of publication, country of origin, sample size, severity of TBI, and type of mortality. Meta-regression showed that only the year of publication was significantly associated with OR estimates: in particular, publication after 2012 correlated with lower OR (slope point estimate: 0.689; p = 0.037).
Publication Bias
Visual evaluations of funnel plots indicated no publication bias (Figs. 9–12). Also, no publication bias was detected by Egger regression (bias 0.433, SE 0.712, p = 0.554), or by Begg and Mazumdar rank correlation (Kendall's z = −0.27, p = 0.784).
Funnel plot for all studies included in the meta-analysis. Figure is available in color online only.
Funnel plot for the studies including only severe TBI patients. Figure is available in color online only.
Funnel plot for the studies of ICU LOS. Figure is available in color online only.
Funnel plot for the studies of hospital LOS. Figure is available in color online only.
Discussion
To our knowledge, this is the first reported meta-analysis to evaluate the association between ICP monitoring and mortality in patients with TBI. Stein et al. conducted a meta-analysis of clinical series on treatment for severe TBI since 1970 and found a 12% lower mortality rate and 6% more favorable outcome for patients who underwent aggressive ICP monitoring with intensive treatment;28 however, that meta-analysis concerned intensive treatment rather than ICP monitoring and not all patients who underwent intensive treatment underwent ICP monitoring. Also, the lack of a uniform definition of “intensive” or “aggressive” treatment is unfortunate.
This meta-analysis examined the association between ICP monitoring and outcome in patients with TBI. Fourteen studies compared mortality in TBI in an ICP monitoring group and a no–ICP monitoring group and reported no evidence that ICP monitoring decreased the odds of death. After excluding the only randomized controlled trial, the other 13 observational cohort studies evaluated the association between ICP monitoring and mortality in TBI patients and reported no evidence for a decreased risk of death with ICP monitoring. Although subgroup analysis revealed that ICP monitoring was significantly associated with a decrease in mortality versus no ICP monitoring in studies published after 2012, the isolated benefit of ICP monitoring in severe TBI is still not clearly established, because there is relatively little clinical evidence as to its efficacy. Additionally, pooled data analyses showed that both ICU LOS and hospital LOS with ICP monitoring were significantly longer than with no ICP monitoring.
There are several limitations to our meta-analysis. First, the major limitation was the relative dearth of studies that met our inclusion criteria. Second, all studies included in the meta-analysis were observational in nature, with the exception of one randomized controlled trial. Although most observational studies used multivariate analyses to determine an adjusted odds ratio after adjusting for many confounding factors, residual or unmeasured confounding remains a possible alternative explanation for our findings. Third, there was considerable heterogeneity for all the outcomes of interest. Heterogeneity among studies that assessed mortality was found by publication year and by the time period over which outcome was evaluated. Sensitivity analyses, including country of origin, study design, sample size, and the timing of outcome evaluation after TBI, however, did not fully explain the observed heterogeneity for mortality. Finally, although we carried out our systematic review according to high methodological standards, the results of any meta-analysis are limited by the quality of the studies included. For example, only 4 studies presented a flow diagram for participants, indicating a risk of bias.
Our systematic review also had some important strengths. We conducted a thorough systematic search, included various databases, and used a comprehensive analytical method that allowed the inclusion of studies presenting not only crude odds ratios, but also adjusted odds ratios, thus enhancing the exhaustiveness of the results. Our rigorous methods were also based on established guidelines for conducting and reporting systematic reviews.
There are several potential reasons for the negative results regarding ICP monitoring. First, because these studies were observational, with the exception of 1 randomized controlled trial, the decision to insert an ICP monitor had been made by the neurocritical care clinicians prior to the onset of the study. Thus, this decision may have been influenced by factors relating to the premorbid conditions of patients and the hospitals, the severity of the current TBI, and specific practices or preferences. Consequently, even with multivariate adjustment, determining an unbiased outcome estimate for ICP monitoring in these studies is problematic. This is exacerbated by the small sample sizes of some of the published studies, which limit adjustment for confounders.
Second, although the only randomized controlled trial included in the meta-analysis also reported that maintaining monitored ICP at 20 mm Hg or lower in severe TBI patients was not superior to care based on imaging and clinical examination fidings, there was a limitation in the trial design. Specifically, the trial was conducted in South America, where practices in ICUs and aftercare may not be comparable with those in North America and Europe, and therefore lacks generalizability. Also, the intraparenchymal monitors used in the trial do not have the advantage of the external ventricular drains used in many ICUs, which allow drainage of spinal fluid to reduce pressure, and the primary outcome was not common mortality but a composite end point.7,25
Third, ICP monitoring would likely benefit only a subtype of TBI patients. Stratifying patients according to a multidimensional classification system of TBI is likely to be necessary to link specific patterns of brain injury with ICP monitoring and other types of evaluation that are most likely to be of benefit.
Fourth, the important difference between the information obtained from the ICP monitor and management based on this information must be highlighted. For example, a change in treatment can be instituted based on ICP or cerebral perfusion pressure findings.16 Moreover, ideal ICP and cerebral perfusion pressure target values have not been determined. Thus, clinicians may be able to gain indications and benefits not currently evident from use of ICP monitors.
Finally, patients who undergo ICP monitoring are subject to increased interventions based on information, such as brain tissue oxygen monitoring and central venous pressure, similar to other monitoring devices/technologies.20 We hope that interventions based on the information gained from these monitoring devices will improve outcomes. However, many of our current therapies for increased ICP, such as sedation and barbiturates, may be associated with worse outcomes in critically ill patients.11 Thus, lack of knowledge remains a confounding factor that cannot be ignored.
Subgroup analysis of the association between ICP monitoring and mortality in studies published after 2012 indicated a significant association between a decreased risk of death and ICP monitoring. This may be attributable to selection bias in recent years, but the more standardized TBI management and more rigorous compliance with BTF ICP monitoring guidelines in recent years, resulting in imexplanation. Several studies have assessed ICP monitoring utilization outcomes in TBI. Alali et al. stratified hospitals into quartiles based on ICP monitoring utilization and found that hospitals with higher rates of ICP monitoring use had lower mortality rates (OR 0.52 [95% CI 0.35–0.78]).1,30 Thus, ICP monitoring may influence outcomes independently, and thus further investigation is warranted.
Despite the fact that the BTF guidelines have some recommendations for ICP monitoring in patients with TBI, significant variability remains in the practice of ICP monitoring. Management plans based on ICP are controversial due to a lack of agreement in precise surgical and medical managements. This may be one reason for the significant heterogeneity in our study. Therefore, we first have to determine the optimal medical and surgical management plan and then randomize to ICP monitoring versus no ICP monitoring within a single management plan.
Conclusions
In this systematic review and meta-analysis of ICP monitoring studies including 24,792 patients, we found that the current clinical evidence does not indicate that ICP monitoring overall is significantly superior to no ICP monitoring in terms of the mortality of TBI patients. However, studies published after 2012 indicated a lower mortality in patients who underwent ICP monitoring. Further effort should focus on standardizing ICP monitors and monitoring and on identifying optimal threshold values for ICP-directed care and distinct subgroups of severe TBI patients who may benefit from monitoring. This will improve the efficacy of ICP monitoring and help guide neurocritical care decisions in a shared decision-making process.
Author Contributions
Conception and design: Hu. Acquisition of data: Yuan, Wu. Analysis and interpretation of data: Yuan, Wu, Sun. Drafting the article: Yuan, Wu, Sun. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Hu. Statistical analysis: Yuan, Wu. Study supervision: Hu.
References
- 1↑
Alali AS, , Fowler RA, , Mainprize TG, , Scales DC, , Kiss A, & de Mestral C, et al.: Intracranial pressure monitoring in severe traumatic brain injury: results from the American College of Surgeons Trauma Quality Improvement Program. J Neurotrauma 30:1737–1746, 2013
- 2
Biersteker HA, , Andriessen TM, , Horn J, , Franschman G, , van der Naalt J, & Hoedemaekers CW, et al.: Factors influencing intracranial pressure monitoring guideline compliance and outcome after severe traumatic brain injury. Crit Care Med 40:1914–1922, 2012
- 4↑
Bratton SL, , Chestnut RM, , Ghajar J, , McConnell Hammond FF, , Harris OA, & Hartl R, et al.: Guidelines for the management of severe traumatic brain injury. VIII. Intracranial pressure thresholds. J Neurotrauma 24:Suppl 1 S55–S58, 2007. (Erratum in J Neurotrauma 25: 276–278, 2008)
- 5↑
Bulger EM, , Nathens AB, , Rivara FP, , Moore M, , MacKenzie EJ, & Jurkovich GJ: Management of severe head injury: institutional variations in care and effect on outcome. Crit Care Med 30:1870–1876, 2002
- 6↑
Bullock MR, & Povlishock JT: Guidelines for the management of severe traumatic brain injury. Editor's commentary. J Neurotrauma 24 Suppl 1:vii–viii, 2007. (Erratum in J Neurotrauma 25: 276–278, 2008)
- 7↑
Chesnut RM, , Temkin N, , Carney N, , Dikmen S, , Rondina C, & Videtta W, et al.: A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med 367:2471–2481, 2012
- 8
Farahvar A, , Gerber LM, , Chiu YL, , Carney N, , Härtl R, & Ghajar J: Increased mortality in patients with severe traumatic brain injury treated without intracranial pressure monitoring. J Neurosurg 117:729–734, 2012
- 9↑
Forsyth RJ, , Wolny S, & Rodrigues B: Routine intracranial pressure monitoring in acute coma. Cochrane Database Syst Rev 2 CD002043, 2010
- 10
Gao L, , Wu X, , Hu J, , Jin Y, , Han X, & Wu X, et al.: Intensive management and prognosis of 127 cases with traumatic bilateral frontal contusions. World Neurosurg 80:879–888, 2013
- 11↑
Girard TD, , Kress JP, , Fuchs BD, , Thomason JW, , Schweickert WD, & Pun BT, et al.: Efficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (Awakening and Breathing Controlled trial): a randomised controlled trial. Lancet 371:126–134, 2008
- 12
Griesdale DE, , McEwen J, , Kurth T, & Chittock DR: External ventricular drains and mortality in patients with severe traumatic brain injury. Can J Neurol Sci 37:43–48, 2010
- 13
Haddad S, , Aldawood AS, , Alferayan A, , Russell NA, , Tamim HM, & Arabi YM: Relationship between intracranial pressure monitoring and outcomes in severe traumatic brain injury patients. Anaesth Intensive Care 39:1043–1050, 2011
- 15↑
Higgins JP, , Thompson SG, , Deeks JJ, & Altman DG: Measuring inconsistency in meta-analyses. BMJ 327:557–560, 2003
- 16↑
Huang SJ, , Hong WC, , Han YY, , Chen YS, , Wen CS, & Tsan YS, et al.: Clinical outcome of severe head injury in different protocol-driven therapies. J Clin Neurosci 14:449–454, 2007
- 17↑
Hyder AA, , Wunderlich CA, , Puvanachandra P, , Gururaj G, & Kobusingye OC: The impact of traumatic brain injuries: a global perspective. NeuroRehabilitation 22:341–353, 2007
- 18↑
Juul N, , Morris GF, , Marshall SB, & The Executive Committee of the International Selfotel Trial, Marshall LF: Intracranial hypertension and cerebral perfusion pressure: influence on neurological deterioration and outcome in severe head injury. J Neurosurg 92:1–6, 2000
- 19↑
Lane PL, , Skoretz TG, , Doig G, & Girotti MJ: Intracranial pressure monitoring and outcomes after traumatic brain injury. Can J Surg 43:442–448, 2000
- 20↑
Martini RP, , Deem S, , Yanez ND, , Chesnut RM, , Weiss NS, & Daniel S, et al.: Management guided by brain tissue oxygen monitoring and outcome following severe traumatic brain injury. J Neurosurg 111:644–649, 2009
- 21
Mauritz W, , Janciak I, , Wilbacher I, & Rusnak M: Severe traumatic brain injury in Austria IV: intensive care management. Wien Klin Wochenschr 119:46–55, 2007
- 22↑
Mauritz W, , Steltzer H, , Bauer P, , Dolanski-Aghamanoukjan L, & Metnitz P: Monitoring of intracranial pressure in patients with severe traumatic brain injury: an Austrian prospective multicenter study. Intensive Care Med 34:1208–1215, 2008
- 23↑
Murray GD, , Butcher I, , McHugh GS, , Lu J, , Mushkudiani NA, & Maas AI, et al.: Multivariable prognostic analysis in traumatic brain injury: results from the IMPACT study. J Neurotrauma 24:329–337, 2007
- 24↑
Roozenbeek B, , Maas AI, & Menon DK: Changing patterns in the epidemiology of traumatic brain injury. Nat Rev Neurol 9:231–236, 2013
- 26↑
Sahjpaul R, & Girotti M: Intracranial pressure monitoring in severe traumatic brain injury—results of a Canadian survey. Can J Neurol Sci 27:143–147, 2000
- 27
Shafi S, , Diaz-Arrastia R, , Madden C, & Gentilello L: Intracranial pressure monitoring in brain-injured patients is associated with worsening of survival. J Trauma 64:335–340, 2008
- 28↑
Stein SC, , Georgoff P, , Meghan S, , Mirza KL, & El Falaky OM: Relationship of aggressive monitoring and treatment to improved outcomes in severe traumatic brain injury. J Neurosurg 112:1105–1112, 2010
- 29↑
Stocchetti N, , Longhi L, , Magnoni S, , Roncati Zanier E, & Canavesi K: Head injury, subarachnoid hemorrhage and intracranial pressure monitoring in Italy. Acta Neurochir (Wien) 145:761–765, 2003
- 30↑
Talving P, , Karamanos E, , Teixeira PG, , Skiada D, , Lam L, & Belzberg H, et al.: Intracranial pressure monitoring in severe head injury: compliance with Brain Trauma Foundation guidelines and effect on outcomes: a prospective study. J Neurosurg 119:1248–1254, 2013
- 31
Thompson HJ, , Rivara FP, , Jurkovich GJ, , Wang J, , Nathens AB, & MacKenzie EJ: Evaluation of the effect of intensity of care on mortality after traumatic brain injury. Crit Care Med 36:282–290, 2008
- 32↑
von Elm E, , Altman DG, , Egger M, , Pocock SJ, , Gøtzsche PC, & Vandenbroucke JP: The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 370:1453–1457, 2007
- 33
Zeng J, , Tong W, & Zheng P: Decreased risk of acute kidney injury with intracranial pressure monitoring in patients with moderate or severe brain injury. J Neurosurg 119:1228–1232, 2013
