Contact Isolation Precautions in Trauma Patients: An Analysis of Infectious Complications

Patients and Methods

The institutional trauma database of our state-verified Level I trauma center was queried for all patients admitted between January 1, 2011 and December 31, 2012. Patient information including age, injury severity score (ISS), co-morbidities, and complications including nosocomial pneumonia and UTI was acquired. Pneumonia and UTI were categorized based on documentation in the medical record during data extraction by trained trauma registrars. The sources for this documentation included physician notes, operative reports, progress notes, radiology reports, respiratory notes, laboratory reports, nursing notes, and discharge summaries as dictated by the American College of Surgeons National Trauma Data Standard. The use of CI precautions was determined from our institution's infection control database.

A patient was deemed to be positive for CI precautions if he or she were isolated for contact, droplet, or aerosol precautions at any time during the admission. Nasal and rectal screening cultures were obtained from every patient admitted who was deemed high risk for colonization with a resistant organism, including patients with known previous colonization or infection with resistant organisms, as well as patients admitted from institutional settings such as skilled nursing or long-term care facilities. Chi-square was then used to analyze the association between CI and the development of pneumonia and UTI.

The International Classification of Diseases, 9th edition, codes documenting the injuries sustained by the patients were mapped to the Barell injury matrix [37] to allow for analysis of the potential effect of specific injury patterns. Injury types collected included traumatic brain injuries (TBI), facial injuries, spinal cord injuries, vertebral column injuries, chest and thorax injuries, abdomen injuries, pelvic and urogenital injuries, upper extremity and lower extremity injuries.

Using the trauma database, data pertaining to patient demographics, injury severity, hospital length of stay, use of mechanical ventilation, and pre-existing co-morbid conditions were collected. Age, gender, ISS, and major pre-existing medical conditions were documented. Because of the unclear relationship between the co-morbid conditions and the development of infections, all conditions were considered.

In addition, a limited chart review was performed to document the reason for which the contact precautions were initiated, as well as the specific organisms responsible for the pneumonia or UTI in patients on CI.

For categoric variables, the chi-square test was used to determine the equivalence of the contact precaution and control groups. For continuous variables, the Mann-Whitney U test was used to compare the groups. When indicated by low group numbers, the Fisher exact test was used in the comparison.

Stepwise logistic regression models were then used to examine the relationship between the variables collected and the two outcomes of interest: Development of nosocomial pneumonia and nosocomial UTI. The Wald tests were performed to examine the significance of each co-variable's effect by comparing their p values with α = 0.05. The final models consisted of only independent variables that had a statistically significant association with either pneumonia or UTI. The odds ratios were reported to quantify the effects of each independent variable of the development of pneumonia and UTI.

All statistics were calculated using JMP 11 (SAS Institute, Cary, NC). Descriptive statistics reported include sample size and median (with interquartile range reported in parentheses) for all data that were not distributed normally. This study was evaluated and approved by our Institutional Review Board for human subjects research.

Results

A total of 4,423 trauma patients were evaluated during the study period. Of these, 4,318 (97.6%) had complete trauma records and were included in subsequent analysis. The majority of the patients were male (2,822, 65.4%), and the median age was 46 years (37). The median ISS for all patients was 9 (13). The CI was in place in 249 patients (5.8%) at some time during their hospital stay. Pneumonia was documented in 190 patients (4.4%)—54 on CI (21.7%) vs. 136 not on isolation (3.3%) with a gross odds ratio of 8.01. Urinary tract infection occurred in 166 patients overall (3.8%)—48 (19.3%) on isolation versus 118 (2.9%) not isolated, with an odds ratio of 8.00.

Bivariable comparisons of age, ISS, and hospital length of stay using the Mann-Whitney U test revealed statistically significant differences between the CI and control groups with isolated patients being older, more severely injured, and with a longer overall length of stay (Table 1). In addition, when injury patterns and co-morbidities were examined, there were several significant differences between the isolation and control groups, with the isolation group having a greater chance of TBI, spinal cord injury, vertebral column injury, chest and thoracic trauma, and trunk injury (Table 2), as well as a significantly higher number of co-morbidities (Table 3).

To more fully examine the effect of CI instituted at hospital admission, patients who were placed on CI secondary to the development of a resistant nosocomial infection during their admission (n = 53) were excluded from further analysis. This left a total of 170 cases of pneumonia. Of these, 34 (20%) were isolated during their admission and 136 (80%) were not. Similarly, 153 cases of nosocomial UTI remained for analysis—35 (22.9%) isolated versus 118 (77.1%) not isolated. These groups underwent further analysis.

When limiting the examination of injury patterns to patients in whom pneumonia (Table 4) and UTI (Table 5) developed, no significant differences were found between the groups. When co-morbidities were examined in a similar fashion, patients with pneumonia with more likely to have pre-existing Parkinson disease on admission (Table 6), while patients in whom UTI developed were more likely to have chronic obstructive pulmonary disease (COPD) at admission (Table 7). All other co-morbidities were not significantly different between the sub-groups.

Our stepwise logistic regression model for pneumonia revealed male gender, increasing ISS, hospital length of stay, COPD, mechanical ventilation during admission and contact precautions to have a significant association (Table 8). Mechanical ventilation was, as expected, the most significant factor in the development of pneumonia. The CI was the second most significant factor with an odds ratio of 2.57. The area under the receiver operating characteristic curve (AUC) for the pneumonia regression was 0.868, indicating good accuracy for this model.

When UTI was examined, the model shifted slightly. Increasing age, female gender, increasing ISS, mechanical ventilation, spinal cord injury, vertebral column injury, pelvic-urogenital injury, and CI precautions were all found to be significant (Table 9). Most significantly associated with UTI were the use of contact precautions (odds ratio 4.61) and spinal cord injury (odds ratio 4.22). The AUC for the UTI regression was 0.806, also indicating good accuracy for this second model. Co-variables considered during the stepwise model development but that did not achieve significance and were ultimately not included in the final models included all other co-morbidities identified on univariable analysis (Table 3) as well as all anatomic injury classifications previously described (Table 2), which did not reach our threshold for significance (p < 0.05).

We next examined specifically the effect of isolation and mechanical ventilation on pneumonia rates. Several interesting relationships were noted. Those who did not require mechanical ventilation but had a diagnosis of pneumonia were significantly more likely to be on CI—48 (1.3%) of non-isolated patients versus 24 (15.5%) of isolated patients (odds ratio 13.82, p < 0.0001). For those patients in whom clinical pneumonia developed and who were ventilated mechanically, however, the relationship was not significant—20 (58.8%) isolated patients versus 88 (64.7) non-isolated patients (p = 0.5239).

In those patients undergoing mechanical ventilation, an interesting anomaly was noted. Isolated patients with pneumonia had no significant difference in the median duration of mechanical ventilation—median ventilator days of 13 in isolated patients versus 11 in non-isolated patients (p = 0.1405). Patients without pneumonia who were ventilated mechanically, however, experienced a significant increase in their median ventilator days—eight days in isolated patients versus two days in non-isolated patients (p < 0.0001).

Finally, we examined the indications for CI precautions based on the data gleaned from our chart review (Table 10). A total of 173 patients (69.5%) had contact precautions initiated on admission because of the results of a nasal swab positive for methicillin-resistant Staphylococcus aureus (MRSA). Of these, 87.9% remained asymptomatic throughout their admission. Fifty-three (21.2%) patients were placed on isolation during their hospital course because of a newly acquired infection with a resistant organism. There was no discernible reason for the initiation of contact precautions in 22 patients (8.9%). Combined with the asymptomatic carriers, 78.4% of the total patients isolated during the study period were asymptomatic and isolated either inappropriately or out of abundance of caution.

Discussion

The current study represents a continuation of our previous work in which we established initially the association between CI and VTE in both surgical patients [29] and patients with traumatic injuries [30]. While our initial study in surgical patients had several methodologic issues because of a lack of comprehensive analysis of potential co-founders, in the current project, we have accounted for a wide variety of factors that may contribute to the effects of contact precautions on the trauma patient population.

Our findings demonstrate clearly a strong association between the use of contact precautions and the development of both nosocomial pneumonia and UTI in the trauma patient population. While gender, injury severity, use of mechanical ventilation, and hospital length of stay all play a role, the use of CI precautions remains a powerful predictor of the development of both pneumonia and UTI. In addition, our data demonstrate that almost 70% of the patients placed on contact precautions were on CI because of colonization at the time of admission, largely via MRSA nasal swabs. While a small proportion of these patients did go on to develop active MRSA infections during their admission, the vast majority did not. This is consistent with previous literature suggesting that placing patients who are asymptomatic carriers on CI does not change the rate of MRSA infection [38].

Another interesting group is the nearly 9% of isolated patients who had no definitive reason for the isolation on chart review. This stems likely from “precautionary” contact precautions initiated by either the treating team or nursing staff because of suspicion of either colonization or infection with a resistant organism. Of specific concern is that the precautions were never removed after negative cultures were established. With the growing body of literature suggesting that contact precautions are not a benign intervention, this practice should be avoided because it places the patient potentially at risk without any corresponding potential benefit.

The reason for this increased nosocomial infection burden is likely because of fewer contacts with clinical staff throughout an isolated patient's intensive care unit (ICU) and hospital stay compared with non-isolated patients in the control group. This relationship has been established previously in surgical patients [38]. Fewer visits by nursing and physician staff is a fairly well established consequence of CI, and we believe that these effects may be pronounced particularly among trauma patients and especially those in the ICU, who benefit from an aggressive and team-based approach to inpatient care [39]. Delayed mobilization and less attention from respiratory therapists, nursing staff, occupational and physical therapists, dietitians, and other team members likely manifests as greater complications overall with nosocomial infections only one of a variety of such adverse events that may be associated with CI.

This may also be a reason for the disparity noted in the mechanical ventilation results. Unsurprisingly our data suggest that prolonged mechanical ventilation has an association with pneumonia. The fact that patients with CI in place who did not develop pneumonia spent almost four times as long on the ventilator, however, suggests that some degree of inattention may be taking place. This would be consistent with previously reported literature as detailed above.

Given that CI precautions have multiple potential adverse effects, what other options are available to protect all hospitalized patients from the hazards of drug-resistant organisms while minimizing the effect on patients either colonized or actively infected with drug-resistant organisms? One option would be to require active surveillance of all isolated patients, especially if efforts are made to treat active infections or clear colonization. This has been demonstrated to significantly increase the rate at which contact precautions are discontinued in hospitalized patients [39]. As the number of colonized and therefore isolated patients grows, this may be an important step given that evidence demonstrates that as the number of isolated patients increases, the compliance with isolation protocols by healthcare workers decreases significantly [7].

Because a recent review questions even the effectiveness of contact precautions in the prevention of nosocomial transmission of resistant organisms [40], what other options might be available? One option that has been studied suggests that precautions should only be used for documented infectious or toxin-mediated diarrheal disease, draining incisions whose drainage cannot be controlled, or uncontrolled respiratory secretions. In a trauma patient population, switching to these transmission-based precautions led to a decrease in contact precaution by half, without any increase in hospital-acquired multi-drug–resistant infections [41].

Another group suggests that universal decontamination practices should be performed in high-risk patient populations, which led to a significant decrease both in MRSA isolates as well as blood stream infection from any organisms in large multi-center study [42]. Others suggest that more emphasis on excellent hand hygiene is the key to preventing transmission of extended spectrum beta-lactamase producing organisms, even in the absence of CI [43]. Some experts are even calling for the re-consideration of the use of all contact precautions, except in the most unusual or extreme circumstances [44,45].

This study does have several important limitations that need to be considered. First and foremost, it is a retrospective review that relies solely on the accuracy of two institutional databases. Any errors introduced in the data entry could influence the results. Use of urinary catheters, which are established risk factors for UTI, were not captured in this database. While this alone certainly makes the interpretation of the data more challenging, the lack of data regarding indwelling urinary catheter use is consistent with the reporting scheme used by other large surgical databases including the National Trauma Databank. Even with the absence of catheter use data, our UTI rates should be comparable to those of other large publically reported data sources. Further, given the known risk of indwelling urinary catheters, efforts are made at our institution to discontinue the use of Foley catheters as soon as clinically indicated.

Another limitation is that temporal data regarding the imposition of CI and development of infections were not available, necessitating the censure of patients who had both an infection and CI from the study as mentioned previously. An additional issue is that our definition of clinical infection relies on the documentation available in the medical record rather than strict criteria. This introduces the potential for error in our infection rates.

When looking at large institutional quality databases, however, the extraction of complication information follows a predefined pattern. This uniformity should allow for valid comparisons between institutions. Even in our limited chart review, a number of cases of suspected pneumonia were present, based on the retrospective review of the medical record, that were not documented clinically. It is unclear whether this was the result of poor documentation or the feeling that these cases represented colonization rather than clinical infection. Given the large effect size encountered, however, it is unlikely that these small errors would affect the outcomes in a major fashion. In addition, because admission screening was not performed on the entire population, it is possible that the effect of isolation may be underestimated.

Finally is the issue of provider compliance with isolation practices. While we do not have specific data at our institution, we can only speculate that our compliance is average. Significant increased or decreased compliance may alter our findings; however, it also represents that reality of the situation, and significantly increased compliance may only serve to amplify the potential worrisome effects of CI.

Conclusions

Patients with trauma are placed typically on contact precautions because of chronic colonization with MRSA. Doing so may place them at risk for the development of both nosocomial pneumonia and UTI, however, as well as an increased duration of mechanical ventilation. Growing evidence demonstrates that the initiation of contact precautions, however well intentioned, places the patient at risk for a wide variety of complications. To date, this is the first published study that has demonstrated an increase in infectious complications associated with the use of CI precautions. This is ironic, because the overall intent of the precautions is to prevent the spread of infection. The use of contact precautions in this and other populations at risk needs to be re-evaluated carefully. Alternatives to the use of CI in these populations should be explored more fully unless definitive interventions can be developed to mitigate this risk.