Overall, the REACT project was a success. While there was negligible improvement in guideline compliance among prehospital providers, the hospital EDs improved their compliance over time. There was also a reduction in the preventable mortality rate during the 1997/98 REACT intervention year compared to the 1992 rate for the same region. These results, their implications and project limitations are discussed below. In addition, we have provided a discussion of how this study fits with recent perspectives on the utility of peer review panels for assessing the effectiveness of trauma care.
There was no significant improvement over time in the prehospital compliance data, which may be due to the large numbers of prehospital providers in the study region. In order to reach the prehospital providers, we conducted as many prehospital education sessions as possible throughout the region. Seventy-six education sessions were conducted and approximately 700 prehospital providers attended these sessions. However, many of the prehospital providers are volunteers and their other commitments make it difficult for them to attend education sessions.
A paucity of experience may also explain the lack of improvement in the prehospital data. Many of the volunteers may only care for one severely injured patient in a year. On the one hand, this lack of exposure is good in that it means the trauma problem remains relatively rare. But it keeps the rural prehospital providers from developing and practicing their trauma care skills.
The feedback to prehospital providers on their compliance with the guidelines was limited. This was because there was only one full-time staff member for the project and, because of all her other project duties, she had inadequate time to devote to providing feedback. Prehospital compliance results perhaps could have been better if the prehospital providers had received more direct feedback.
The guideline compliance results suggest that there were improvements over time in the care delivered in regional emergency departments. The improvement was likely related to the education sessions as well as the feedback on guideline compliance sent to all EDs participating in the project. In the majority of the emergency departments participating, the feedback was shared with the ED staff. In some EDs, the feedback information was used as a part of their internal quality improvement programs.
Compared to 1992, there was a significant (p<0.01) decrease in the overall preventable mortality rate as well as a significant (p<0.01) decrease in the related episodes of inappropriate care given the preventable mortality study patients. The results suggest that implementing the STAF model decreased the preventable mortality rate.
The overall preventable mortality rate of nearly 15% is closer to that found in comparable rural-based trauma preventable mortality studies. A Montana study conducted in 1994 found a preventable death rate of 13% (Esposito et al., 1995). A similar study conducted in Michigan about the same time found an overall preventable death rate of 12.9% (Maio et al., 1996). These rates are higher than those of urban systems where sophisticated trauma systems have been implemented. For one urban system, the preventable mortality rate went from 13.6% to 2.7% after trauma system implementation (Shackford SR, 1986).
Despite significant decreases in episodes of inappropriate care, compared to the 1992 study, the results suggest a need for continued emphasis in training in the following areas: airway control; oxygenation/ ventilation management; injury recognition; chest trauma management; and the management of shock. Since physicians are the primary decision-makers and treatment managers, an intervention directed specifically at them would be helpful. The REACT project intervention was restricted in this regard as it was aimed primarily at prehospital providers and ED staffs.
There are many reasons that might explain the higher rural preventable mortality rates. These include different patient populations and injury patterns; and longer discovery and transport times (Rogers, 1997). Rural trauma is one of the remaining major challenges in trauma care. Rogers and colleagues have recently (1999) illuminated the many important issues that need to be addressed in order to improve the prospects for rural trauma patients. These include: better definition of “rural” trauma patients; combined federal and professional efforts; a national rural trauma database; and increased public awareness.
The major contributor to rural trauma deaths rates is the motor vehicle crash. This work was no exception as nearly 50% of all deaths studied were related to events on the highway. The improved results suggest that the STAF model had an effect in reducing motor vehicle crash deaths. The implication is that prehospital emergency medical care should have a definite role in any highway safety plan.
Other factors may have contributed to the overall reduction in the preventable mortality rate found in this study. These are discussed briefly below.
Differences in the populations of the 1992 and 1997/98 preventable mortality studies could be a factor. However, analysis of the composition of the study samples with regard to demographic variables and mechanisms of injury failed to reveal any statistically significant differences.
The significant decrease in the overall preventable death rate could be related to a decrease in the number of mechanical trauma deaths in the region. Injury prevention efforts such as seatbelt and child restraint laws and advances in vehicle driver and passenger safety. However, as illustrated by Table 24, the trauma death rate has not declined over the intervening years in the study area.
|Year||Total Population|| Mechanical Trauma Deaths*
*Mechanical trauma deaths are defined by e-codes
used for the 1992 PMS and 1997-98 PMS
Another factor which may have played a part in reducing the preventable mortality death rate is the general increase in knowledge about trauma patients and trauma care. Advances in trauma care have been made over the years between the two studies. As health providers at all levels are made aware of changes and advance through continuing education or word of mouth, these may be incorporated into practice. Therefore, the preventable death rate may have decreased without the implementation of the STAF model. However, since a part of STAF is training and education, it can also be argued that REACT was perhaps responsible for making new information available to practitioners.
Further development and acceptance of the trauma center and trauma systems concept in eastern North Carolina can also be credited with contributing to the decline in the preventable mortality rate. Transfer rates from regional hospitals to the trauma center have increased steadily over the past few years. In 1992, 45% of the trauma admissions to our trauma center were transfer patients. In 1998, transfers constituted 52% of the admissions. There has also been an increase in the number of out-of-county scene runs for the aeromedical transport service for the trauma center. Research validates that implementation of a trauma system results in a 15 to 20% improved survival rate among seriously injured patients (Mullins and Mann, 1999).
The use of peer review panels to evaluate improvement in trauma care defined as a decrease in the preventable death rate has been reviewed recently. MacKenzie (1999) identified four major issues with panel reviews. The first of these involves the definition of a preventable or possibly preventable death and the separation of judgments regarding appropriate versus inappropriate care. Criteria for the determination of preventable death rate should be clearly defined. Although it is logical to assume that acts of inappropriate care are directly related to preventable death rate, determinations regarding the appropriateness of care should be made independently of preventable death rate. Not all acts of inappropriate care result in a preventable death. The second issue deals with the study population. Often, prehospital deaths are excluded from these studies since information regarding their care is limited. If these deaths are excluded, the only real care that can be evaluated fully is hospital care as opposed to trauma system care. Therefore, to fully evaluate the effectiveness of an intervention or trauma system, all patients should be included. A third issue relates to the composition of the review panel. A panel should be multidisciplinary so that all aspects of care can be reviewed fully. The fourth issue is the review process itself. Independent review with a unanimous decision regarding preventable death rate provides the most reliable results, according to MacKenzie.
In this study, the issues identified by Mackenzie as affecting the validity of the panel findings were addressed. Judgments as to the preventable death rate were made independently of determinations of appropriateness of care. Criteria for determining preventable death were the same as those used in the 1992 study and were well defined. Appropriateness of care decisions were based on ATLS, TNCC, and BTLS guidelines. A multidisciplinary review panel was used, composed of trauma surgeons, emergency physicians, an anesthesiologist, and a forensic pathologist. Decisions as to preventable death and appropriateness of care were made after an independent review of the records and a discussion by the panel. Unanimous decisions were required for a determination of preventable death. Patients dying as a result of mechanical trauma were randomly selected for inclusion. Only patients pronounced dead at the scene and receiving no prehospital care were excluded. All others would have been transported to an emergency department in the current EMS system.
The methods used for data collection, which were based on available resources, explains the limitations related to the guideline compliance data. The major limitation is that data collection was done by numerous people in many different settings. Also, data collectors were of different education and experience levels. In many cases, the data collection forms were completed at a later date rather than while the patient was present in the ED. Therefore, the information entered would have been based on documentation rather than actual observation.
Because of the limited resources, we were unable to hire project staff for every location or even for the regional trauma center to be available 24 hours a day to monitor and collect guideline compliance data on every single trauma patient in the 29-county region. We were relying on the voluntary efforts of the personal in the hospitals and trauma center to complete the forms. The rates of data collection were relatively high, however, because we maintained frequent contact with the personnel at the hospitals. Still, since this is a convenience sample which fails to have data on every single patient during the study period, the reader is cautioned to avoid comparing this data with that on inappropriate care from thePreventable Mortality Study.
To allow comparison of the 1992 and 1997/98 studies, we used the same peer review panels. While these panelists were from North Carolina they were not from the region that was being studied. They may have been biased, however, by not being blinded to the purposes of the project.
While the same panelists were used for both studies and in both studies they used the same review form, there was one aspect of the 1997/98 review that was different from 1992. That is, in 1992 the two panelist assigned to a case worked together to review their assigned cases rather than reviewing and reporting individually as they did in the 97/98 REACT PMS. We do not believe this had any significant bearing on the different results in the two studies.
One possible significant limitation with the 97/98 PMS related to the list of names from the Office of the Chief Medical Examiner for the second half of the study. This caused us to have a smaller list than in the 1992 study and, therefore, a smaller number of deaths. We believe there was no selection bias, though, since the project investigators did not control which deaths were available for study. Other factors that may have affected the results other than the intervention are discussed in the Contributing Factors section above.
Continue to Recommendations | Return to Contents