The Impact of Gram-Negative Bacilli in Bacteremic Skin and Soft Tissue Infections Among Patients With Diabetes

Skin and soft tissue infections (SSTIs) are common and may be more severe in patients with diabetes. In particular, the development of diabetic foot ulcers facilitates diabetic foot infections (DFIs), which are generally not included in SSTI classifications (1). In these situations, gram-negative bacilli (GNB) may have an impact on the etiology of severe SSTIs.

We aimed to analyze the GNB etiology, epidemiologic trends, and risk factors for mortality among a large cohort of adult patients with diabetes attended in our hospital (1996–2016) with bloodstream infections from an SSTI source (1). Cases with nosocomial acquisition or due to a sacral decubitus ulcer were excluded.

Out of 30,363 episodes of bacteremia, 7,202 (23.7%) occurred in patients with diabetes; this proportion increased over time from 17.7% in 1996–2000 to 28.3% in 2011–2016 (P < 0.001). Among bacteremic episodes in patients with diabetes, 247 (3.4%) were secondary to SSTIs, of which 223 were analyzed (Table 1). The most common clinical form was cellulitis (n = 184, 82.5%). DFIs represented 25% of cases (with abscesses in 23%), of which 41.8% required surgery (amputation, 43.5%).

Regarding the microbiology, 138 (61.9%) cases were caused exclusively by gram-positive cocci (GPC) (mainly Streptococcus spp. [63.8%] and Staphylococcus aureus [29%]) and 68 (30.5%) by GNB (Escherichia coli [26.5%] and Pseudomonas spp. [26.5%]). Seventeen cases (7.6%) were polymicrobial mixed infections (combining GPC and GNB); those were excluded for specific comparison of patients with GNB SSTIs vs. GPC SSTIs (Table 1). The former were older, had higher rates of diabetic complications and health care–related infections, had received prior antibiotics more often, and had infections that were mostly located on the lower extremities.

Comparison of the 1996–2000 and 2011–2016 time periods revealed an upward trend in the prevalence of GNB SSTIs (14.3% vs. 40.6%; P = 0.020), in the previous use of antibiotics (14.3% vs. 44.9%; P = 0.003), and in the proportion of GNB resistant to frequently used antibiotics (amoxicillin-clavulanate and fluoroquinolones) (0% vs. 20.3%; P = 0.100).

The 30-day mortality rate was 21.5%, ranging from 57.9% in necrotizing fasciitis to 17.2%, 14.3%, and 10.5% in cellulitis, pyomyositis, and abscesses, respectively (P < 0.001). Mortality from DFIs was 16%. Patients with GNB SSTIs had higher mortality rates (32.4%) compared with patients with GPC SSTIs (13.8%) (P = 0.002) in all types of SSTIs but necrotizing fasciitis, which had an intrinsic high mortality independent of the etiology. In multivariate analysis after adjusting for age, sex, presence of coronary disease, type of SSTI (excluding necrotizing fasciitis), and site of acquisition, infection due to GNB was the only statistically significant risk factor for mortality (adjusted odds ratio 2.73 [95% CI 1.11–6.70]; P = 0.028).

Our study shows an increasing trend in GNB SSTIs among patients with diabetes, with consequent greater mortality. The results may be inherently limited by the retrospective review and conditioned not only by diabetes but also by other comorbidities of the patients or the presence of bacteremia itself.

The most common SSTI was cellulitis, but DFIs represented 25% and had higher rates of abscesses and need for amputation. This supports the previous opinion about considering DFIs in further classifications of SSTIs because of their frequency and particular characteristics (1).

Despite the dominance of GPC etiology, we observed a greater proportion of GNB (30.5%) than is classically reported (2). This fact seems not only related to the presence of ulcers but also the location of SSTIs on the lower extremities and the specific immunosuppressive conditions that patients with diabetes may present. Also, we recognized a significant increase in the proportion of GNB bacteremia over time, as has been noted worldwide in similar settings (3). Finally, we noticed an increase in the SSTIs caused by GNB resistant to the most commonly used antibiotics but without a specific increment in those GNB defined as multidrug resistant, thus leading us only to hypothesize about their lower virulence or a possible emergence of multidrug resistance in the future.

In our study, the mortality rates varied according to the type of SSTI and were high considering overall mortality from SSTIs but comparable to other studies with the same type of patients or with bloodstream infections (4). Notably, infections due to GNB were associated with twofold increased odds of mortality, after adjusting for other causes. These differences were detected in all types of SSTIs except necrotizing fasciitis. To our knowledge, scarce reference has been made previously to GNB etiology as an independent risk factor for mortality in SSTIs (5).

In conclusion, because of the increase in GNB infections, including those resistant to the most commonly used antibiotics, and their greater mortality, appropriate empirical coverage is mandatory, whether in patients with SSTIs and severe systemic manifestations or suspicion of bacteremia.

• Received November 2, 2018.
• Accepted April 18, 2019.

• © 2019 by the American Diabetes Association.