Increasing Adherence to Brain Trauma Foundation Guidelines for Hospital Care of Patients With Traumatic Brain Injury

This study is a secondary analysis of data from hospitals that participated in the AWI program between 2005 and 2013. In brief, the AWI program was initiated in 2004 at Mission Hospital, Mission Viejo, California; physicians and nurses from 37 hospitals designated as level I or level II trauma centers received training, support, and equipment from AWI. As part of the agreement between the hospitals and AWI, participating sites were responsible for entering data into the online TBI-trac system for at least 2 years after the initial site training. Information entered into the TBI-trac by the AWI-trained hospital staff included baseline records and a daily hospital record for up to 10 days and a limited amount of outcome data on blood pressure and oxygenation, ICP monitoring, hypothermia, Glasgow Coma Scale (GCS) scores, infection prophylaxis, ventilation, deep venous thrombosis (DVT) prophylaxis, cerebral perfusion pressure (CPP), sedation, nutrition, antiseizure prophylaxis, and use of steroids.

Our study was confined to data from AWI hospitals with at least 10 cases of TBI documented in the TBI-trac database and at least 12 months between the first and last case. Eligibility criteria for the TBI-trac database includes overall GCS score less than 9 for at least 6 hours after injury, with a GCS motor score less than 6 and a GCS eye score of 1, and a mechanism of injury consistent with TBI. Patients were excluded if they died in the emergency department, had do-not-resuscitate or do-not-intubate orders, or were transferred to the study hospital more than 24 hours after injury.

We analyzed data from 16 of the 37 AWI participating sites. Hospital records were included if the associated hospital provided a minimum of 10 patient records in the BTF database, with at least 1 daily hospital record and outcome data. Total data analyzed included information on 1301 patients, with 8083 records from 16 hospitals. Data were summarized as frequency and percentage or mean and standard deviation. SAS, version 9.4, software (SAS Inc) was used for all analyses.

Table 1 gives demographic data for the sample. Most patients were non-Hispanic (93.9%), white (88.5%), and male (74.6%). Computed tomography (CT) scans were performed on 99% of the patients; 88% had abnormal findings. The GCS scores on day 1 (hospital admission) for 89% of patients were in the severe range (score 3-8). Data from daily hospital records were entered up to 10 days; a total of 8083 daily hospital records were analyzed in our study. The mechanism of injury of TBI was mostly accidents involving an automobile or a motorcycle or falls. Specifically, 34.9% of the study patients were the driver or a passenger in a motor vehicle accident, 14.0% were in a motorcycle accident, 5.1% were a pedestrian in a motor vehicle accident, and 26% had TBI caused by a fall (55% fell < 3 m; 45% fell > 3 m).

Data on the key 15 topics from the BTF guidelines⁸  are presented in Table 2.

Centers affiliated with AWI recorded systolic blood pressure less than the threshold 90 mm Hg for a total mean duration of about 30 minutes (0.56 hours) in 337 patients in the first 24 hours of admission, and an increase in mean duration to 1.05 hours in 398 patients during the first 48 hours of admission.

A total of 68.9% of all patients had ICP monitoring. Approximately half of all patients who received hyperosmolar therapy (50.3%) had a record of meeting ICP and CPP treatment thresholds (ICP > 20 mm Hg; CPP < 50 mm Hg). Furthermore, mannitol was given more commonly (49.2%) than was hypertonic saline (26.6%).

A total of 11.8% of TBI patients were treated with prophylactic hypothermia in the first 3 days of their hospital stay; the value increased to 13.3% for the first 10 days. Among the 532 patients who had a GCS score less than 6, only 11.3% had hypothermia prophylaxis in the first 3 days.

Tracheostomies were placed in 138 patients (10.6%) in the first 10 days of admission; among these patients, 97.8% began treatment with mechanical ventilation on day 1. Furthermore, 18.9% of patients had placement of ventricular catheters; among these, 17.1% received a one-time dose of periprocedural antibiotics, and 15.9% received continuous antibiotics. Only 1.46% of patients had pneumonia, and none had records indicating infections associated with ICP monitoring or sepsis.

A total of 68.8% of all patients had DVT prophylaxis (either mechanical or chemical) during the first 2 days of hospitalization, 0.3% of patients between day 3 and day 10, and 30.9% had no DVT prevention or intervention noted. Forty-one (3.2%) patients were diagnosed with DVT, of whom 97.6% had already undergone DVT prophylaxis by day 2.

Among patients with GCS scores of 3 to 8 (severe), 72.1% had a record of ICP monitoring. A total of 88.4% of all patients had abnormal findings on CT scans (see Table 2 for criteria). Among patients who had normal CT findings but were older than 40 years and had a GCS motor score less than 4, 28.2% had ICP monitoring.

A total of 66.9% of patients had some form of ICP monitoring. Among these, 27.5% were monitored via ventriculostomy devices and 74.1% via parenchymal devices (not mutually exclusive events).

In a subset of 429 patients who had monitoring of the partial pressure of oxygen in the extracellular fluid of the brain (partial pressure of brain tissue oxygen; Pbto₂), 75% had a record of measures taken to decrease ICP when thresholds (ICP > 20 mm Hg) were crossed.

Among patients with ICP monitoring, 68.2% were within target CPP ranges (50–70 mm Hg) during day 1 of hospitalization for some amount of time. The same percentage (68.2%) of patients with ICP monitoring had recorded CPP values exceeding the upper threshold (70 mm Hg) for any time during day 1; of these, 48.5% exceeded this threshold for 1 to 8 hours.

Among all patients, 33.0% had Pbto₂ monitoring, and 0.2% had monitoring of jugular venous oxygen saturation. Of the 429 patients who had Pbto₂ monitoring, 81.6% had at least 1 episode of Pbto₂ less than 20 mm Hg. Of patients who had Pbto₂ values less than 20 mm Hg, values were 16 to 20 mm Hg in 80.0%, 11 to 15 mm Hg in 55.5%, 6 to 10 mm Hg in 45.2%, and 0 to 5 mm Hg in 31.0%.

A total of 11.3% of patients were treated with high-dose barbiturates. In patients receiving mechanical ventilation, 84.8% were given sedatives or propofol at least once, and 29.5% of these patients were given high-dose propofol at least once.

By day 3, 59% of patients began some form of nutritional therapy; of these, 70.6% had enteral feeding. Among patients who received supplemental feeding (ie, to meet caloric needs), 72.9% received it before day 10.

Among all patients, 48.6% received prophylaxis with antiepileptic drugs; of these patients, 99.0% had the prophylaxis started before day 7.

Among the patients in the sample, 6.2% had at least 1 record of Paco₂ less than 25 mm Hg. Among the 1274 patients who had Paco₂ monitoring on day 1, Paco₂ values were less than 25 mm Hg in 4.1%, 25 to 29 mm Hg in 5.0%, 30 to 34 mm Hg in 19.4%, and within the reference range (ie, ≥ 35 mm Hg) in 61.3%.

During the first 10 days of hospitalization, 96.6% of all patients received no steroids.

Hospitals that participated in the AWI had generally high adherence to BTF guidelines, supporting the hypothesis that a dedicated and structured educational program is associated with high adherence to BTF recommendations. Gupta et al¹¹  found that higher adherence to BTF guidelines is associated with lower mortality. Our results extend those findings¹¹  by revealing an association between a focused educational effort and guideline adherence. By encouraging education and training, AWI helps nurses and physicians focus on key evidence-based metrics proven to improve outcomes in TBI patients. However, numerous factors affect implementation of and compliance to guidelines, including characteristics of the guidelines, the implementation strategies, the professionals involved, the patients, and the environment.¹⁴  Additionally, measuring efficacy of guidelines without involvement of patient outcomes measures is difficult.

In lieu of using a single variable to measure adherence, we looked grossly at patient data and compared the data with individual recommendations in the BTF guidelines based on 15 key areas of hospital care of TBI patients. Overall, our sample matched the general population¹⁵  of TBI patients: mostly male and with similar mechanisms of injury (primarily falls and moving vehicle accidents). We made no adjustments for age or race, because compared with adults, fewer infants and children have TBI, and race-based differentiation factors are not available.¹⁶  The study population was skewed toward whites; 88.5% of the sample were white. The AWI selected sites on the basis of a formal application process. Annual review of applications included an evaluation of patient volume, characteristics of the organization, and commitment to performance improvement. Perhaps the hospitals that reached out to the foundation had additional resources that allowed them to ask for the assistance, but this notion cannot be confirmed. Furthermore, the degree of injury in our population of patients was restricted to patients who required hospitalization, resulting in increased need for nursing care and increased subsequent morbidity and mortality.

Blood pressure and oxygenation control in the intensive care unit contribute greatly to improved patient outcomes. Uncontrolled and untreated hypotension or hypertension is associated with increased mortality in TBI patients.¹⁷  Most centers in our study met blood pressure goals for almost the entire first 24 to 48 hours of admission. The availability of numerous monitoring techniques, along with vasoactive medicines, makes adherence to blood pressure guidelines a key in management of patients with TBI. However, better definition within our data set is needed to understand the adherence to oxygenation guidelines. We did not include numerous unreliable Pao₂ values and erroneous or skewed values of oxygen saturation measured by using pulse oximetry (Spo₂) in our observations. Indeed, changes in Spo₂ are not reliably predictive of equivalent changes in arterial oxygen saturation in critically ill patients,¹⁸  and large differences between Spo₂ and arterial oxygen saturation may occur in critically ill patients with poor reproducibility of Spo₂.¹⁹ 

Hyperosmolar therapy is used to manage increased ICP. Although we could not determine whether an incident of increased ICP was followed by hyperosmolar therapy, our finding that half of the patients who did receive hyperosmolar therapy met ICP/CPP threshold criteria for such treatment leads us to think that the other half (with no record of hyperosmolar treatment) most likely would have also met those criteria but were not having ICP monitoring at the time. Furthermore, according to hospital records, mannitol was being given more commonly as hyperosmolar therapy than was hypertonic saline, a difference that is probably influenced by the BTF guidelines. Even though a meta-analysis²⁰  of randomized clinical trials of comparisons of mannitol and hypertonic saline indicated that hypertonic saline is more effective than mannitol for elevated ICP treatment, both solutions are still commonly used. We think that the frequent use of either form of hyperosmolar therapy indicates the usefulness of the therapy and good adherence to BTF guidelines by nurses and clinicians.

Although the BTF guidelines from 2007 do not recommend prophylactic hypothermia because the intervention is not associated with any significant decrease in mortality, scores on the Glasgow Outcome Scale reportedly improved significantly when prophylactic hypothermia was used in TBI patients.⁸  We suspect this reported improvement is the reason approximately 11% of patients in our study received the intervention. Although evidence supports the use of early prophylactic mild-to-moderate hypothermia in patients with severe TBI (GCS score < 7) to decrease mortality and improve neurological recovery,^21,22  recent findings²³  indicate no better outcomes in TBI patients when hypothermia is used along with standard care. These studies have contributed to complicated adherence to the BTF guidelines on hyperthermia.^21–23 

Although infection prophylaxis to reduce pneumonia is recommended for patients when tracheostomies are created, we found no strong adherence to such prophylaxis. The low incidence of pneumonia might be attributed to use of other periprocedural or continuous antibiotics. The BTF guidelines make no recommendation for continuous antibiotics; however, other factors (preexisting infectious conditions, iatrogenic infections) and local hospital policy on antibiotic usage might have resulted in inconsistent adherence to BTF guidelines. Furthermore, additional information on tracheostomy placement in our data set is needed.

Only 68.8% of the patients in our study had DVT prophylaxis; the recommendations for TBI patients are universal mechanical or chemical prophylaxis (with exception for chemical prophylaxis in patients at risk for intracerebral hemorrhage).²⁴  Of our patients diagnosed with DVT, 97.6% already had received early DVT prophylaxis; most of the prophylaxis was probably mechanical, not chemical. The BTF guidelines have no recommendations on the timing, dosing, and preferred agents in DVT prophylaxis. Indeed, no national standard of care exists to guide TBI caregivers in the initiation and use of prophylactic anticoagulation, specifically chemoprophylaxis, despite the frequency and morbidity of DVT events after TBI.²⁵  This variance in adherence to the guidelines may be attributed to the fear of an increase in intracerebral hemorrhage after chemoprophylaxis in TBI patients; however preliminary trials²⁶  have indicated no significant worsening of TBI after chemoprophylaxis. We hope future TBI guidelines will use new evidence to encourage stronger adherence to the BTF recommendations.

Currently, ICP monitoring is ubiquitous in TBI management, with guidelines recommending that all patients with severe GCS scores (3–8) and abnormal CT findings be monitored. We defined abnormal CT scans by using the 5 measures listed in Table 2. Although we were unable to determine how many patients meeting those measures were placed on ICP monitoring, we think adherence most likely was high because the emphasis on numerical or objective ICP criteria has become more important. Indeed, although guidelines do not recommend specific ICP monitoring technology, they encourage low-cost methods (ie, ventricular catheters), a practice that has increased adherence, as our results indicate.

Our observations indicate that ICP and CPP goals (ICP < 20 mm Hg and CPP 50–70 mm Hg) were fairly closely adhered to; 75.3% of the patients in our study had measures taken to decrease ICP when thresholds were crossed. Although we could not obtain data indicating that direct action was taken whenever ICP or CPP thresholds were crossed, crossing the ICP and CPP thresholds was common. More direction is needed in the guidelines about prevention of increased ICP and its efficacy.

In our data set, because CPP values were recorded in intervals of time, further refinement in the collecting process might help clarify the usefulness of measuring CPP. However, of note, CPP exceeded the upper threshold (70 mm Hg) many times; exceeding the threshold is not recommended because of the association with higher rates of acute respiratory distress syndrome.²⁷  Also, BTF guidelines have had different CPP goals over the years. In 2003, the goal was greater than 60 mm Hg, and in 2007 it was 50 to 70 mm Hg. The centers that reported CPP higher than 70 mm Hg might have been in the appropriate range according to which guideline recommendations they followed.⁸ 

Other methods for monitoring brain oxygen levels include Pbto₂ and jugular venous saturation. Because these methods have not been universally adopted (most likely because of high costs of placement and use of the devices required) and because of poor data quality, we could not make further conclusions about adherence.

The main concern with use of anesthetics, analgesics, and sedatives in management of TBI patients is avoiding high-dose propofol (notable for causing marked mortality) and high-dose barbiturates.⁸  Nevertheless, among our sample, 29.5% of patients receiving mechanical ventilation were given high-dose propofol. In addition, although the BTF guidelines allowed for regular doses of barbiturates, recent evidence²⁸  indicates minimal benefit in outcomes for TBI patients who receive regular doses. Future clarification and suggestions for alternative therapy might help centers improve care and increase adherence to this guideline.

Adequate nutrition by day 7 is recommended. We were unable to determine the exact percentage of patients who received nutritional feedings by day 7, but we conclude that TBI patients are indeed receiving appropriate nutritional feedings in a timely manner, and are given supplemental nutritional feedings as needed to ensure that full caloric needs are met.

Our findings indicate that antiseizure prophylaxis as specified by the BTF guidelines was strictly followed. Almost all (99.0%) of patients requiring prophylaxis for early posttraumatic seizures received antiepileptic medication before day 7 of hospitalization. This high level of adherence may be due to the clarity and specificity in the recommendation itself.

Hyperventilation therapy is not recommended for TBI patients. Lowering Paco₂ to less than 30 to 35 mm Hg is in the first 24 hours after injury is specifically contraindicated. Yet a small percentage of patients (4.1%) had low Paco₂ values (an indication of hyperventilation) on day 1. However, because of the secondary use of hyperventilation in lowering increased ICP, clinical discretion with hyperventilation therapy leads to a small amount of non-adherence to the BTF recommendations.

Although steroids are not recommended for TBI patients, a small percentage of patients in our group did receive steroids. The risk of death increases in TBI patients when they receive corticosteroids,²⁹  resulting in the level 1 recommendation by the BTF. Despite the increased risk, we assume that comorbid conditions and clinical decisions contributed to our findings.

In our study, we noted certain trends in adherence to the BTF guidelines. First, the more specific the guidelines, the better adherence was. As done in the AWI program,¹⁰  the BTF guidelines might be revised or illustrated with specific algorithms. Second, we think that several guidelines from the BTF require revision. The third edition was published in 2007, and the AWI program ended several years afterward, possibly compromising adherence. This lack of updating is an inherent pitfall that occurs with all disease-based guidelines.³⁰  Undoubtedly, AWI is encouraging proactive TBI care by giving nurses and clinicians the background to evaluate whether guidelines are still valid; an acquisition of this knowledge further encourages adherence. Third, AWI-affiliated centers most likely are encouraging adherence by forcing nurses and staff to enter data—indeed, this Hawthorne-like effect is a benefit of initiatives such as AWI and may prove to be valuable.

The question of deontological versus teleological ethics is worth discussion. In short, deontology is the branch of ethics that addresses “doing the right thing” without external motivation. As it relates to the AWI, deontology is adhering to guidelines because guideline adherence is good.³¹  Teleology (sometimes called consequentialism) addresses the concern that hospitals may have been (at least partially) motivated to more closely adhere to the BTF guidelines under the assumption that higher adherence would result in continued support from the BTF.³¹  Of note, in the hospitals involved in our study, the AWI primarily provided educational resources, but also capital equipment, and covered the fees associated with access to the BTF database.¹⁰ 

Ultimately, the effectiveness of dissemination of the BTF guidelines to health care providers by AWI has had a markedly positive impact. Indeed, discrepancy often exists between dissemination and implementation of guidelines.³²  Translating guidelines into practice requires a multidisciplinary approach, with both nursing staff and clinicians taking a proactive role in guideline research, as well as treatment teams taking an active role in implementing these guidelines in the teams’ local institutions. Implementation of the BTF guidelines relies on astute and timely nursing care; the crux of these guidelines is parameters initially observed by immediate nursing staff and then afterward by physicians. Ultimate outcomes may improve over time because of all these factors leading to improved adherence to the BTF guidelines for hospital care for TBI patients due to the work of the AWI.

The BTF Guidelines are important because they provide a platform for developing and standardizing best practice. The work of the AWI, to help promote compliance to the guidelines for hospital care of patients with TBI, is associated with high adherence to evidence-based clinical guidelines. This high adherence may subsequently improve patient care and outcomes for TBI patients. The hands-on learning, equipment support, and clinical support and teaching are important elements in promoting adherence to guidelines.

The authors acknowledge Adam Williams and Nancy Williams for their continued support and encouragement. The Adams Williams Initiative was funded by the Adam Williams Foundation, which provided support for this study. The authors thank all the nurses, therapists, and physicians who have collaborated with and contributed to the Adam Williams Initiative.