Daily Multidisciplinary Rounds To Implement the Ventilator Bundle Decreases Ventilator-Associated Pneumonia in Trauma Patients: But Does It Affect Outcome?

Patients and Methods

With the approval of the Institutional Review Board, we conducted a quasi-experimental study where data were collected retrospectively before and after institution of the goal-focused rounds. The setting was a 12-bed SICU at an urban Level I trauma center. Our SICU admits more than 500 patients a year, of which at least 50% are trauma patients. Our SICU team is composed of a surgical critical care Board-certified attending, one physician assistant, three to four post-graduate year 1 (PGY-1) residents, and one PGY-3 general surgery resident, who serves as the chief resident.

A multidisciplinary team comprising the SICU nursing staff, a respiratory therapist, a physician assistant, and the SICU chief resident (PGY-3) make daily morning GRs on each patient seven days a week. The GRs are brief, lasting 15–20 min in total, and are completed before attending rounds. The charge nurse for the day and the nurse caring for the patient being discussed are always present. If there are no patient care issues that need immediate attention, given the brevity of GRs, usually, the entire SICU nursing staff is able to be present. A paper checklist on each patient is used to determine whether specific VAPB practices are being employed. The checklist also includes other practice goals such as central line maintenance. The VAPB review includes three questions that are directly involved in VAP prevention: (1) Is the head of bed elevated to 30°; (2) is this patient a candidate for sedation interruption; and (3) is this patient a candidate for a spontaneous breathing trial to assess readiness for ventilator liberation? Additionally, our VAPB reviews whether the patient is receiving peptic ulcer disease and deep venous thrombosis prophylaxis. The only other preventive measure added to the VAPB during the study period was the introduction of chlorhexidine mouth washes. This took place only during the last week of the study period. For the purposes of this study, compliance with GRs was defined by the percentage of daily paper checklists completed and accounted for during each patient's SICU stay.

The trauma registry was used to identify all trauma patients admitted to the SICU over a 20-month period. This time was divided into two ten-month periods: Before GRs were established (November 1, 2007, to August 30, 2008; pre-GR); and after (September 1, 2008, to June 30, 2009). The registry and patient charts were reviewed for demographics and clinical characteristics, which included age, gender, mechanism of injury, admitting Glasgow Coma Scale score (GCS), Injury Severity Score (ISS) [8,9], chest Abbreviated Injury Severity (AIS) [8,9], and the number of units of red blood cell concentrates (PRBC) transfused in the first 24 h after admission. Additionally, the patient charts provided secondary outcome data on hospital mortality rate, ventilator days, and length of stay in the SICU (SLOS) and hospital (HLOS). Patients with incomplete information were excluded from analysis. The National Nosocomial Infection Surveillance (NNIS) criteria [10] were used by our Infection Control Department to record and track VAP prospectively. All chest radiographs were reviewed by an attending radiologist. The standard definition of VAP incidence rate was used, defined as the number of VAPs per 1,000 ventilator days [1].

The primary goal of the analysis was to assess the effectiveness of our GR intervention on VAP incidence rate. Additionally, we compared the secondary outcomes of ventilator days, SLOS, HLOS, and death in the pre-GR and GR cohorts. In the second part of our analysis, we compared ventilator days, SLOS, HLOS, and mortality rate for the VAP and non-VAP cohorts. Death was defined simply as death before hospital discharge.

All categorical data were compared via chi-square test or, where required, by small sample size, the Fisher exact test. Continuous data for two groups were compared via independent unpaired t-tests when observed to follow a Gaussian distribution. A Welch correction for unequal variance was applied as necessary. Continuous data not observed to follow Gaussian distribution were compared via the Mann-Whitney U test. Statistical significance was defined as p<0.5. All analyses were performed with GraphPad Prism, Version 5.0b (GraphPad Software, Inc., La Jolla, CA).

Results

There were a total of 25 VAPs recorded in the SICU for the 20-month study period. Twenty-one, or 84%, of these infections occurred in the trauma population. A total of 366 trauma patients were admitted to the SICU in this same period; accordingly, the incidence of VAP in our trauma population was 5.7% (21/366). There were 182 of 366 trauma patients (49.7%) who required intubation. Of this group, eight patients—four in the pre-GR group and four in the GR group—had incomplete information and were excluded from analysis. One excluded patient in the pre-GR group had a recorded VAP. The remaining 20 VAPs in 174 intubated trauma patients were analyzed.

Of the 174 intubated patients, 85 were admitted in pre-GR period and 89 in the GR period. The pre-GR and GR groups were similar in baseline demographics and clinical characteristics (Table 1). Sex and mechanism of injury—in the categories of blunt and penetrating—were distributed almost evenly between the groups. There was no difference between the two groups with regard to age, chest AIS, admitting GCS, and number of units of PRBC transfused in the first 24 h after admission.

There were two exceptions to inter-group similarity. In the pre-GR group, there were only five (5.9%) falls from standing height, whereas there were 19 (21.3%) in the GR group (p=0.01). The second exception involved a higher ISS mean in the pre-GR group than in the GR group—17.2±9.1 vs. 13.9±8.1 (p=0.01), respectively.

Analysis of the primary outcome of number and incidence rate of VAPs revealed a significant difference between the pre-GR and GR groups (Table 2). There was a 67% decrease in the number of VAPs after GRs were established—15 (17.6%) in the pre-GR group and 5 (5.6%) in the GR group (p=0.02). This corresponded to a VAP incidence rate decrease from 26.8 to 7.0 VAPs per 1,000 ventilator days (p=001; relative risk 0.26; 95% confidence interval 0.07–0.75) or a 74% decrease in VAP incidence rate.

A total of 33 patients died, yielding an overall all-cause mortality rate of 19% (33/174). Despite the significant decrease in VAP incidence rate associated with GRs, there was no significant difference in the mortality rate for the pre-GR and GR groups—16.5% (14/85) vs. 21.3% (19/89), respectively (p=0.41). Likewise, there was no significant difference in mean ventilator days in the pre-GR and GR groups (Table 2). There were decreases in mean SLOS and HLOS in the GR group compared with the pre-GR group; however, none of these differences reach statistical significance (Table 2).

In the second part of the analysis, the intubated trauma population was divided into non-VAP and VAP groups. There were 154 (88.5%) in the non-VAP group and 20 (11.5%) in the VAP group. The two groups were similar in demographic and clinical characteristics except the VAP patients had a higher mean ISS—19.6±7.5 vs. 15.0±8.8 (p=0.01) (Table 3).

There was no significant difference in the mortality rate in the non-VAP and VAP groups. There were only two deaths in the VAP group and 31 in the non-VAP group. The VAP group had statistically significantly greater number of mean ventilator days and SLOS and HLOS than the non-VAP group (Table 4).

Compliance in completing the VAPB checklist during GRs remained high throughout the study period. Except for the month of January, in which compliance was 85.1%, compliance ranged from 90.5–99.8% (Fig. 1).

OPEN IN VIEWER

FIG. 1.

Daily goal rounds compliance by month.

Discussion

This study demonstrates that daily multidisciplinary goal rounds focused on the VAPB can reduce the VAP incidence rate among trauma patients. Moreover, this can be accomplished with a small resource outlay—a paper checklist and daily investment of 15–20 min. Thus, with a relatively simple intervention, we were able to obtain a dramatic reduction (74%) in the VAP rate in our SICU trauma population. Our resultant VAP incidence rate of 7.0 VAPs per 1,000 ventilator days (previously 26.8) placed us in the lowest quartile of VAP rates for trauma patients among hospitals participating in the NNIS monitoring system [1].

The VAPB has already been shown to be effective in reducing VAP in the trauma population [3,4], so it is not completely clear why it was not effective in our SICU trauma population before GRs were instituted. Although the VAPB philosophy had been adopted in our SICU, its actual implementation was informal and inconsistent, and this likely contributed to its failure before the GR intervention. The reduction in the VAP incidence rate after the introduction of GRs is most likely multifactorial; however, several reasons are easily discernible. First, the GRs provided a formal vehicle that ensures all VAPB items will be covered consistently and completely on a daily basis. Second, enhanced communication has been touted as one of the benefits of multidisciplinary approaches to patient management [5]. The GRs allow all the disciplines—respiratory therapy, nursing, physician—an opportunity for augmented communication at the same time. This approach not only provides a more efficient way to communicate, but undoubtedly lessens the chance of miscommunication between disciplines, which is likely when information is exchanged between several parties at different times. Finally, the sheer simplicity of GRs—a paper checklist—and the fact that all involved in the GRs multidisciplinary team are always immediately available in the SICU probably facilitate the daily implementation of GRs. This, in turn, helps ensure that all the elements of the VAPB are applied consistently to all intubated patients.

Compliance with any intervention is essential for its success. Compliance may become more difficult to achieve when a particular intervention involves several disciplines as opposed to one person. The ease and simplicity of the GRs, as mentioned above, undoubtedly played a large role in the consistently high compliance rates for conducting the GRs. Notably, the GRs were well received by the nursing staff and gradually began to be initiated by the nursing staff rather than the SICU physician over the ten-month study period. Eighteen months after the completion of this study, GR compliance remained at or above 98%.

In the same 18-mo period after the study was completed, the VAP rate was maintained at 7.6/1,000 similar to the GR ventilator-period rate of 7.0 days. These similar VAP rates over more than two years suggest strongly that the GR intervention was not only successful but sustainable. The introduction of chlorhexidine mouthwash at the end of the study period likely did not affect the results of our study. However, it may have contributed to the sustained prevention of VAP. It is difficult to know how much this addition to the VAPB helped maintain the VAP rate near its previous level during the study period. Notwithstanding, sustainability remains an important, indeed essential, component of any PI project or intervention. Incidentally, this also argues against the success of the GRs being attributed to a Hawthorne effect, as, by definition, this phenomenon is temporary.

Even though the GRs were successful in reducing the incidence of VAP, there was no corresponding reduction in the mortality rate in the GR group. One must be careful about drawing any definitive conclusions from this result. Note this comparison between the pre-GR and GR cohorts was for an all-cause mortality rate, not deaths specifically associated with VAP. Given that there were only two deaths in the VAP cohort—in the pre-GR and GR periods combined—there is insufficient statistical power to elucidate the true impact of the GR intervention or VAP on the mortality rate. Beyond the insufficient statistical power, one can hypothesize other possible explanations for the small number of deaths in the VAP cohort. There is a growing body of evidence that questions whether VAP is associated with the same risk of death in trauma patients as in non-trauma patients [11–14]. In fact, these studies suggest that death caused by VAP may be significantly less common in trauma patients. If this were indeed true for our trauma population, it would explain why a decrease in VAP incidence would have little to no effect on the mortality rate attributable to VAP.

Although there was a significant decrease in the VAP rate in the GR cohort, there was no significant difference between the groups in mean ventilator days, SLOS, or HLOS. Given that two of the VAPB prevention measures, sedation interruption and spontaneous breathing trials, involve liberating the patient from the ventilator—in other words, decreasing the overall time on the ventilator—this is an interesting and unexpected result. A type II error resulting from our study's relatively small sample size may account for this phenomenon; however, there are other plausible explanations. Keeping the head of the patient's bed at 30° was also part of our GR VAPB checklist. Drakulic et al. showed a semirecumbent body position reduces the incidence of VAP [15]; however, this VAPB intervention is not directly associated with reduced ventilator days, SLOS, or HLOS. Alternatively, there are characteristics unique to the trauma patient that are associated with prolonged ventilation and ICU stay, irrespective of the presence or absence of VAP. Cervical spine and traumatic brain injury, along with chest trauma, such as rib fractures and pulmonary contusions, all have been associated independently with prolonged ventilation and consequently longer ICU stay [16–18].

Not surprisingly, when we compared patients without VAP with those with VAP, we found a significant increase in ventilator days, SLOS, and HLOS. These results add to the established body of evidence that VAP is associated with more ventilator days and longer SLOS and HLOS [19–21]. Although we could not show a statistically significant decrease in mean SLOS and HLOS for the GR cohort, there was a trend in that direction. Along these lines, any intervention that reduces VAP has great merit, in that it may reduce HLOS and thus hospital costs. A financial analysis was beyond the scope of this study, but VAP has been associated with an added hospital cost as high as $57,000 secondary to extended ventilator days and longer SLOS and HLOS [20,21]. It is safe to assume that our simple GR intervention with its minimal resource outlay would result in an excellent cost:benefit ratio and could save significant amounts of money related to extended hospital stay. This becomes especially important in a climate of soaring healthcare costs, where healthcare providers are being asked to do more with fewer resources.

Our results need to be interpreted in the light of several study limitations that deserve emphasis. First, as noted previously, the small number of deaths in our VAP group does not provide sufficient power to draw any definitive conclusions about the impact of our GR intervention or VAP on the mortality rate. Second, we were unable to show definitely that our GR intervention had an effect on the secondary outcomes of ventilator days, SLOS, and HLOS, possibly because of the small sample size. Lastly, the retrospective nature of the study makes it impossible to control for differences between study groups. Our patient demographic and characteristic analysis revealed a significantly lower mean ISS in the GR group than in the pre-GR group. This may have created a bias favoring more VAPs in the pre-GR group, which had a higher mean ISS than the GR group. However, in this context, Magnotti et al. [12] did not show ISS to be an independent risk factor for VAP. These limitations warrant further study and, ideally, should be addressed with a larger data set studied prospectively.

Conclusion

This study demonstrates that daily multidisciplinary GR focused on the implementation of the VAP bundle can significantly lower the incidence of VAP in trauma patients. This intervention can be sustained with high compliance and, additionally, with a minimal resource outlay. Ventilator-associated pneumonia is associated with extended durations of mechanical ventilation and longer ICU and hospital stays. Our study did not have sufficient statistical power to determine whether this GR intervention to prevent VAP can have a positive effect on the mortality rate and secondary outcomes such as ventilator days and ICU stay. Answers to these questions warrant further study.