Mortality in severely injured elderly patients: a retrospective analysis of a German level 1 trauma center (2002–2011)

This study analyzed the data of a level 1 university-based trauma center in Germany. Serving the catchment area of the Ruhr district with approximately 5.1 million habitants, it is one of the largest trauma centers in Germany. There are four level 1 trauma centers in the Ruhr district and the emergency medical system is mostly ground based, though air transportation via helicopter is available as well. For our hospital, the rate of helicopter ambulance is approximately 11%.

The data used in this study was collected prospectively for the national trauma registry, called the Trauma Registry of the German Society for Trauma Surgery (DGU). The data from the Trauma Registry of the DGU has received full approval from the Ethics Committee of the University of Witten/Herdecke in Cologne, Germany. Because the trauma registry of the DGU is an anonymous register, the Institution Review Board waived the need for patient consent. Additionally, the patients’ clinical records were analyzed. For this analysis we received full approval from the Ethics Committee of the medical faculty of the University Duisburg-Essen in Essen, Germany.

Inclusion and exclusion criteria are shown in Table 1.

Scales, general patient information, laboratory test values, and intervention data were collected for each patient as follows. Scales: Injury Severity Score (ISS) [12]; Abbreviated Injury Scale (AIS); New ISS [13]; Glasgow Coma Scale (GCS) [14]; Revised Trauma Score (RTS) [15]; Revised Injury Severity Classification (RISC) [16]; and Trauma and Injury Severity Score (TRISS).

General patient information: Age; sex; ASA score; systolic blood pressure at the accident scene and at admission; heart rate at the accident scene and at admission; oxygen saturation at the accident scene and at admission; length of ICU stay; length of hospital stay; count of performed surgeries; administered fluid volume; proportion of multi-organ failure (MOF); proportion of sepsis; and type of injury (penetrating vs. blunt).

Laboratory test values: First hemoglobin (Hb) value; initial number of platelets; partial Thromboplastin time (PTT); Prothrombin time; base excess; and lactate.

Length of analyzed periods: Time from admission to cranial computed tomography (CCT); time from admission to whole-body CT; time in trauma room; time from admission to operating room; and preclinical rescue time (time from arriving at the accident scene to admission in hospital).

Interventions: Intubation, resuscitation, and thoracic drainage by emergency physician at the accident scene and intubation, cardio-pulmonary resuscitation, and thoracic drainage in trauma room.

To allow age-adjusted analysis, the patients were divided into age groups as follows: 76–80 years, 81–85, 86–90, and > 90 years of age.

Similarly, patients were grouped by ISS: ISS 16–25, 26–35, 36–45, and > 45.

To determine the difference between decedents and survivors, odds ratios and relative risks with 95% confidence intervals (CI) were used to determine predictors for mortality. The cutoffs for laboratory tests were set according standard values (Table 2).

Data were analyzed using the Statistical Package for the Social Sciences (SPSS21; IBM Company; Chicago, IL, USA). Incidences are represented as percentages. Measured values are represented as means and 95% CI for continuous variables, and for categorical variables as medians and interquartile ranges (IQR). Differences were evaluated using the Chi-squared test for categorical variables and the t-test for continuous variables. When performing the t-test, Levene’s test was also performed. In cases of variance heterogeneity, the Welch-test was used instead of the t-test. Normal distribution was tested using the Kolmogorov-Smirnov-test. When an obvious deviation from normality was detected, continuous variables were tested with a non-parametric rank test (Mann–Whitney test). Differences were considered statistically significant when p < 0.05.

In the observation period, 2,304 patients were admitted to the trauma room. Of these, 258 patients were aged more than 75 years. A total of 108 patients met the inclusion criteria, of which 38.3% were male.

The median GCS was 6 (3–13), the median ISS was 25 (20–29), the median AIS head was 5 (4–5), and the mean age was 82.2 years. 62 patients died after trauma, resulting in a mortality proportion of 57.4%. The expected mortality proportion, demonstrated by the RISC, was 54.6%. Most of the patients (29; 46.8%) died within the first 24 hours (Figure 1). In addition, 15 (24.2%) died between days 2 and 5, 8 (12.9%) between days 6 and 10, and 10 (16.1%) after day 10. The in-hospital mortality was analyzed.

The primary cause of death was severe head injury, claiming 37 (59.7%) patients by traumatic brain injury. In addition, 12 (29.4%) patients died as a result of hemorrhage, 7 (11.3%) as a result of sepsis, and 6 (9.7%) as a result of MOF (Figure 2).

Differences between survivors and decedents are shown in Table 3.

The decedents were more severely injured, having a higher ISS (26 vs. 20, p < 0.001) (Figure 3) compared to survivors. Decedents suffered more severe head trauma, evidenced by a lower GCS (4 vs. 12, p < 0.001), and the higher AIS head score (5 vs. 4, p = 0.006). No differences were found in other the AIS scores.

There was no statistical difference in the ASA score, suggesting the two groups were similar regarding comorbid conditions. There were also no differences in the physiologic parameters, such as systolic blood pressure, heart rate, or oxygen saturation, whether at the accident scene or in the trauma room.

The first hemoglobin count was different between the groups, with lower values in the decedents (10.5 vs. 11.6 g/dl, p = 0.004). The coagulation values were different as well. The PTT was extended to 41.1 seconds in decedents compared to 27.6 seconds in survivors (p = 0.008). Prothrombin ratio was lower in the decedents (66.5% vs. 82.8%, p = 0.016). No difference was found in platelet counts.

Although the base excess was not different between the two groups, the lactate value was higher in the decedents than in survivors (2.9 vs. 1.6 mmol/l, p = 0.043).

The lengths of stay in the ICU and in hospital were shorter in the decedents. All other investigated time periods were not different.

No differences were found in the fluid volumes the patients received. In total, decedents received 2439.7 ml and the survivors received 2215.5 ml (p = 0.321). 52 patients received more than 1500 ml fluid volume. Of these, 23 presented normal physiologic parameters at the accident scene and in the trauma room (systolic blood pressure ≥ 120 mm Hg, pulse rate < 100 bpm). The blood pressure cutoff of 120 mm Hg was used, because no emergency physician might consider such a blood pressure as hemorrhage-inducted hypotension. The distribution of infused fluid volumes is presented in Figure 4.

Decedents suffered from MOF more often (51% vs. 23.3%). Analysis by organ system resulted in a higher proportion of failure in decedents in almost every system, respiratory (56.1% vs. 23.3%; p = 0.001), central nervous system (86.4% vs. 43.5%; p < 0.001), heart and circulatory system (71.2% vs. 18.6%; p < 0.001), and renal system (22% vs. 4.8%; p = 0.018). Failure rate in only the hepatic system was not different between the groups.

The results of determining odds ratios and relative risks are shown in Table 4. Six predictors for mortality in elderly trauma patients were identified: ISS > 25, GCS < 9, PTT > 32.4 seconds, prothrombin ratio < 70%, AIS head > 3, and Hb < 12 g/dl. The strongest predictor for mortality was ISS > 25, with a 7-fold higher risk for a fatal course.

The most common cause of trauma was fall from a height less than 3 m (for decedents and survivors), followed by accidents as a pedestrian. No differences occurred between decedents and survivors according the cause of trauma (Table 5).

The mortality proportion was similar in all age-adjusted groups with no great difference to the prognostic mortality proportion, as reported by the RISC-score (Figure 5). In all age-groups, the observed and expected mortality rate was 50% or greater.

The preclinical intubation proportion was different between survivors and decedents. Most decedents were intubated at the accident scene. This difference was found in all age-groups.

Looking at the group of patients aged less than 86 years, survivors had a significantly higher GCS. Only in the youngest group (76–80 years of age) did the decedents suffer more MOF than the survivors. In older patients, no difference in MOF and sepsis proportion occurred between decedents and survivors. There were no differences in AIS scores between groups (Table 6).

The ISS-adjusted analysis, presented in Table 7, showed differences between decedents and survivors only in those scoring 16–25. Only 10 patients presented an ISS > 45 (Table 7).

The surviving-proportion was only 17.1% among patients presenting an ISS > 25. The ISS-adjusted mortality proportion and the expected mortality proportion are shown in Figure 6.

In the ISS 16–25 group, decedents suffered more severe head injury than survivors, evidenced by lower GCS and significantly higher AIS score head. Also, decedents had deranged coagulation with significantly lower prothrombin ratio and a prolonged PTT, compared to survivors.

This study is limited because it is a retrospective analysis. Also, as a single-center study, there might be selection for patients treated in our hospital. To confirm these results, a trauma registry analysis with multicenter data should be performed. No statement can be made about patients who died at the accident scene or on the way into hospital because no documentation is available for these patients. This might introduce a bias, although this study investigated only the in-hospital mortality.