Differences in Characteristics and Outcome of Patients with Penetrating Injuries in the USA and the Netherlands: A Multi-institutional Comparison

This multi-institutional retrospective cohort study was performed at five Level-1 trauma centers in the Netherlands (Netherlands trauma center (NLTC): Academic Medical Center, Erasmus Medical Center, Vrije Universiteit Medical Center, Haaglanden Medical Center and Leiden University Medical Center) and three Level-1 trauma centers in Boston, USA (US trauma center (USTC): Boston Medical Center, Brigham and Women’s Hospital, and Massachusetts General Hospital). These NLTC and USTC are all located in urban areas with comparable population densities (4200/km² in Boston versus 5000/km² in the Amsterdam–Leiden–Rotterdam region) [16, 17] and comparable violent crime rates (390 and 360/100.000 in Massachusetts and the Dutch region, respectively) [18, 19].

Eligible patients were identified in the trauma registries of the participating centers. All patients over 15 years of age who had been admitted to the NLTC or USTC with truncal penetrating injuries, i.e., penetrating injuries to the neck, thorax, abdomen, back or inguinal area, between July 1, 2011, and December 31, 2014, were included. Patients with isolated penetrating injuries to the head or the extremities (i.e., without truncal penetrating injuries) were excluded. Patients who were managed at another hospital before arriving at the participating hospital or were transferred to another hospital after initial treatment in participating hospital were excluded. Also, patients who died before arrival or arrived more than 48 h after trauma at the emergency department were excluded. Institutional review board permission was obtained from all participating centers.

Demographic data and injury data, defined according to the Abbreviated Injury Score (AIS, updated 1998) [20], Injury Severity Score (ISS) [21], vital signs and Revised Trauma Score (RTS) [22] on admission were extracted from the trauma registries. Data on comorbidity, scored using the age-adjusted Charlson comorbidity index [23, 24], and complications were collected from the medical records.

The primary outcome was in-hospital mortality. Secondary outcomes included hospital length of stay (HOS-LOS), intensive care unit admission and length of stay (ICU-LOS) ventilator-free days [25], readmission rates, complications (pneumonia, urinary tract infection (UTI), deep venous thrombosis (DVT), sepsis and wound infection) and discharge disposition.

During the study period, 470 patients with truncal penetrating injuries were admitted in the NLTC and 861 in the USTC. The number of included patients per trauma center in each country is presented in Fig. 1. In general, more patients with penetrating trauma per trauma center were admitted in the USTC compared to the NLTC.

Table 1 summarizes the demographics and clinical characteristics in both centers. USTC patients were younger, slightly more often male and had a somewhat higher ISS than NLTC patients (median ISS 9 in both groups, p = 0.01), but no difference in RTS was seen. In USTC, significantly more patients with gunshot wounds were admitted (36.1 vs. 17.4%, p < 0.0001), which were more often the result of assault compared to NLTC. In both centers, the ISS of gunshot wound patients (NLTC median ISS 16 [interquartile range IQR 7.5–25] vs. USTC median ISS 16 [IQR 9–20], p = 0.82) was significantly higher compared to the ISS of patients with stab wounds (NLTC median ISS 9 [IQR 2–11] vs. USTC median ISS 6 [IQR 2–11], p = 0.64).

Figure 2 shows the distribution of severe penetrating injuries (AIS > 2) per body region in both centers. NLTC patients had more often severe injuries to the spine (2.3 vs. 0.1%, p < 0.0001), while in USTC patients more penetrating injuries to the abdomen (24.2 vs. 18.9%, p = 0.03), extremities (9.4 vs. 4.9%, p = 0.0003) and multiple body regions (30.2 vs. 18.4%, p = 0.001) were seen.

The in-hospital mortality rate in NLTC was 3.6% (17/470) compared to 5.0% (43/861) in USTC (p = 0.25) (Table 2). The unadjusted OR for mortality in the USTC compared to the NLTC was 1.40 (95% CI 0.75–2.49, p = 0.25). After correction for clinically relevant confounders, the adjusted OR for in-hospital mortality in the USTC compared to the NLTC was 0.95 (95% CI 0.35–2.54, p = 0.91). Higher ISS, RTS < 10 and gunshot wounds were statistically significant predictors of in-hospital mortality (Table 3). There was no difference in mortality in patients with gunshot wounds (NLTC 11.3% vs. USTC 11.0%, p = 0.48) and patients with stab wounds (NLTC 2.1% vs. USTC 1.5%, p = 0.94).

HOS-LOS was similar in both centers (Table 2). After correction for differences in case mix, USTC HOS-LOS was on average 0.17 days shorter than in NLTC (95% CI −0.29 to −0.05, p = 0.005). A higher age, gunshot wounds, higher ISS and low RTS were statistically significant predictors of a longer LOS. The ICU admission rate in USTC appeared higher compared to NLTC (33.8 vs. 28.6%, p = 0.05), but this association was not statistically significant after adjustment for differences in case mix (OR 0.96, 95% CI 0.71–1.31, p = 0.80). ICU-LOS was significantly longer in USTC compared to NLTC (median 2 [IQR 1–5] days vs. 1 [IQR 1–2] days, p < 0.0001). This association remained statistically significant after correction for differences in case mix (difference of 0.39 days, 95% CI 0.18–0.60, p < 0.0001). A higher ISS and gunshot wounds were statistically significant predictors of a longer ICU-LOS. More ICU admitted USTC patients received mechanical ventilation than NLTC patients (47.1 vs. 58.1%, p = 0.04), after correction for clinically relevant parameters this difference was no longer statistically significant (OR 1.59, 95% CI 0.99–2.57, p = 0.06). DVT was more often diagnosed in USTC, and the incidence of other complications was similar in both countries (Table 2). This difference in DVT incidence ceased to exist after adjustment for differences in case mix (OR 3.0, 95% CI 0.36–35.1, p = 0.31).

A statistically significant difference in discharge disposition was seen (p < 0.0001), with more USTC patients being discharged to a rehabilitation center (5.0 vs. 1.1%), while more NLTC patients were discharged to a mental health facility (9.3 vs. 5.6%) or nursing home (2.4 vs. 0.7%). Readmission rates were similar, even after correction for differences in case mix (OR 1.4, 95% CI 0.75–2.71, p = 0.28) (Table 2).

The detailed data collection and the large cohort are strengths of our study. Data collected from the trauma registries were complemented by data from electronic medical records collected by one researcher (AH), limiting the amount of missing data. A limitation of our study was that no information on morbidity and mortality was available after hospital discharge. A second limitation is that we excluded specific patient groups from the study such as patients with isolated penetrating injuries to the brain and extremities. They are considered a different group, and the involvement of trauma surgery is often limited after the initial resuscitation phase. Additionally, all patients who were first managed in another hospital before being admitted to one of the participating centers were excluded. Although studies have shown that mortality is similar between transferred and non-transferred patients, differences in complication rates do exist [40]. By excluding these patients, we may have caused a selection bias in the study groups, since transfer rates were higher in USTC. However, it was not feasible to collect primary data for these patients, so we felt compelled to exclude them. A third limitation is that the study was performed in a limited number of trauma centers in both countries. Although we feel that the participating USTC and NLTC are representative for the Level-1 trauma centers in the densely populated urban areas in the USA and Netherlands, the results of this study may not allow for a comparison of care for patients with penetrating injuries in the two countries as a whole.