Morbidity and mortality of Hispanic trauma patients with diabetes mellitus

Trauma and DM are leading causes of death in the Hispanic population. Despite this fact, the few studies that have evaluated the relationship between them have consisted mostly of Caucasian patients, with the Hispanic patients studied being aggregated with other minorities. Therefore, our goal was to evaluate the complications generally suffered by Hispanic DM trauma patients and the in-hospital mortality rate of this group. Our findings confirm the result of previous studies that also found that DM patients had a greater tendency than did their non-DM counterparts to develop infectious complications after trauma [10]. However, the number of infectious complications suffered by Hispanic DM trauma patients who took part in our study was significantly greater than what has been reported in previous studies, in which Hispanic DM patients were in the minority. In our study, complications, predominantly of an infectious nature, were more common among DM patients than they were among non-DM patients: 31.3% in the DM group vs. 11.6% in the non-DM group (OR 3.46; 95% CI 2.77–4.31).

Ahmad et al. performed a retrospective analysis of 12,489 patients (65% of whom were male) who were hospitalized due to injury from 1984 to 2002. They evaluated cases from the Pennsylvania Trauma Systems Foundation database and matched the DM patients with an equal number of non-DM patients. Ahmad and his team reported a compelling difference in the rate of complications between the DM and the non-DM groups: Twenty-three percent of their DM patients developed infectious complications vs. 14% of their non-DM patients, who did not. Infectious complications such as wound infections and pneumonia were less than 2 times more common in diabetics compared to non-diabetics [10]. A stark difference is seen in our results, in which Hispanic DM patients were 6 times more likely to develop wound infections than were non-DM patients. Similarly, Kao et al. evaluated 343,250 patients from the US National Trauma Data Bank who were injured from 1994 to 2003 and of which number only 2.7% were diabetic. In their analysis, they noted that insulin-dependent DM patients had a 58% excess risk of developing infectious complications compared to the risk of same sustained by non-DM patients [11].

In respect to ICU stay and need for mechanical ventilator, Ahmad et al. showed that patients with DM had extended ICU stays compared to those of the non-DM group (7.6 days in the DM group vs. 5.0 in the non-DM group). The patients in their DM group also received ventilator support for a greater period of time than did their non-DM group (10.7 days in the DM group vs. 6.7 days in the non-DM group) [10]. Even though our results are similar to those of this previous study, the difference in TICU LOS and days on MV was 2 times greater in our DM patients when compared to the results presented by Ahmad et al. [10] (Our median TICU LOS was 17 days in the DM group vs. 9 days in the non-DM group and the need for MV was 15 days in the DM group vs. 6 days in the non-DM group). This could indicate that our population of Hispanic DM patients have a significantly greater risk of developing complications than Caucasian DM patients, which is a novel finding in the trauma setting.

The presence of PECs may be responsible for the relatively higher morbidity observed in DM patients. In fact, a study conducted by Morris et al. demonstrated that a group of patients in the state of California who had been injured in 1983 and who had 1 or more PECs were 30% more likely to die than were patients without any PECs, with DM being one of the top 5 PECs in their analysis [14]. This suggests that DM affects not only a patient’s baseline health status but also his or her recovery process. Even though our study observed comorbidities similar to those noted by Morris et al. [14], there were differences in terms of mortality.

The excess morbidity observed in DM patients following trauma may be attributed to changes in the immune system that are a consequence of the inciting injury, which may be reflected by a given patient’s elevated glycemic level [10]. Trauma across all ages is associated with a metabolic response known as stress-induced hyperglycemia, in which the body releases catecholamine and stress hormones, causing an elevation in blood sugar [24]. Hyperglycemia leads to impairment of chemotaxis and adherence, oxidative bursts, and the death of polymorphonuclear cells, all of which can lead to the inability to combat infections [25‐28]. This process has been linked to increased TICU and hospital LOS [7, 8, 29], rates of infectious complications, and mortality [7‐9].

Our study determined that diabetic Hispanic trauma patients had a twofold increase in the rate of complications, compared to the rate observed in Caucasian diabetic trauma patients. Strategies to identify these patients upon arrival at the trauma bay may decrease in-hospital complications and LOS. For example, the HbA1C levels of all the diabetic patients could be measured automatically as part of that patient’s initial work-up; in this way, health care personnel could determine whether the patient’s DM was controlled or uncontrolled. Strict measurement of glucose levels in the early stages post-injury could also be implemented. This could lead to an earlier and more aggressive management of that individual’s DM and, in turn, fewer complications. A prospective randomized trial performed by Van den Berghe demonstrated a relationship between intense insulin therapy (keeping glucose levels < 200 mg/dL) and a reduction in the morbidity and mortality of diabetic surgical ICU patients [30]. This has yet to be proven in the diabetic trauma patient; therefore, our study would serve as the initial step in establishing an insulin regimen for diabetic trauma patients, especially in the Hispanic patient population, which is disproportionately affected by diabetes.

As to mortality, our results indicate that there is no statistically significant difference in mortality when DM patients are compared to non-DM patients. The scientific literature is inconsistent regarding the role of DM in trauma mortality. For instance, Tebby et al. found an excess mortality risk of 64% for diabetics, after adjusting for age, sex, ISS, GCS score, and PECs [12]. Meanwhile, Liou et al. determined that IDDM patients have a 46% excess mortality risk compared to patients without DM [15]. However, this relationship was not observed in the NIDDM group. This fact is of particular importance because when these two categories are collapsed (IDDM and NIDDM), the effect that DM has on trauma mortality is diluted.

Other researchers, however, have reported findings similar to those documented by us. In Ahmad and colleagues’ study, the mortality analysis was performed matching DM patients with non-DM patients by sex, age, and ISS to allow comparisons of similar groups [10]. With a total of 24,978 participants for this sub-analysis, the authors did not find statistically significant differences for mortality between the compared groups. Furthermore, when Kao et al. evaluated the hypothesis that DM results in a higher mortality in trauma patients, they did not find a statistically significant association, after adjusting for age, ISS, and mechanism of injury [11]. The authors also entered the variables IDDM and NIDDM into the mortality prediction model as separate terms, and neither was statistically significant.

These four studies confirm the heterogeneity that exists in the evidence produced on diabetes and trauma mortality. All these investigations included sophisticated analyses that, in some way, considered potential confounding variables. In our case, we decided to control the potential confounding variables in our design phase using a propensity score matching technique to match patients based on strong mortality predictors such as sex, age, mechanism of injury, GCS score, and ISS. The study by Ahmad et al. [10], who also used the matching technique, found no difference in mortality. Therefore, the method selected to control confounding variables, as well as the populations in which the studies were conducted, might explain the differences found in the studies on this topic.

Our study has several limitations, including its retrospective nature, which may have led to misclassification bias. Because we used ICD-9 codes to identify the DM patients, there exists the possibility that we might have missed patients who did not have an established diagnosis of DM on admission. As a result, undiagnosed patients may have been included in our non-DM group.

An important limitation within our study was the inability to include the glycemic and HbgA1C levels of DM patients, as this clinical information was not consistently available in the records of all the patients selected for our study. Despite differences in the pathophysiology of DM types 1 and 2, DM status also could not be further subdivided, and therefore, differences in the outcomes of such patients could not be analyzed. Another limitation of our database was that it lacked certain variables, such as admission glucose levels, body mass index, medication compliance, and time since diagnosis of DM.

These limitations are balanced by the key strengths of our study, the primary being that it focused on Hispanic trauma patients, a population that has not been previously studied on its own. Considering the lack of knowledge in this area, there was a need to determine if there are any differences in terms of morbidity and mortality in this patient population, as it is the largest minority group in the US. Our study also analyzed data collected over an extended period of time, which adds consistency to our findings.