Contemporary Trends in Necrotizing Soft Tissue Infections in the United States

Abstract

Introduction

Necrotizing soft-tissue infections (NSTIs) are a collection of rapidly advancing, often fatal infections of the subcutaneous tissues and fascia. Mortality for NSTI has been reported to range from 21–43% in single center studies[1, 2]. Larger studies have shown a mortality as high as 34%[3] but more recently there appears to have been a decline in mortality to 10–12%[4–6]. These epidemiological studies on NSTIs have provided some data on treatment modalities such as hyperbaric oxygen therapy (HBOT)[7, 8] and the association between a number of co-morbidities such as chronic alcoholism, obesity, diabetes, and immunocompromised status on the development of NSTIs[9]. Only two of these studies have examined a national cohort of NSTI patients[4, 5].

It appears that the basic principles of treatment for NSTI have remained unchanged over the past several decades. Early diagnosis and treatment have been the only consistently proven predictors of outcomes for NSTI[10, 11]. Aggressive antibiotic therapy with prompt, often serial, surgical debridement of all necrotic tissue is paramount to reduce morbidity and mortality[2, 12]. We undertook this study to examine recent national trends in incidence, treatment, and outcomes for NSTI from a large representative sample of the US population.

Methods

The Nationwide Inpatient Sample (NIS) (1998–2010) was queried for all patients with a primary diagnosis of NSTI (International Classification of Diagnosis ninth revision [ICD-9] codes 728.86 [necrotizing fasciitis], 040.0 [gas gangrene], or 608.83 [Fournier’s gangrene] and at least one surgical debridement (ICD-9 procedure codes defined in Appendix 1) or at least one amputation (ICD-9 procedure codes defined in Appendix 1). The NIS is a dataset from the Healthcare Cost and Utilization Project and represents the largest all-payer inpatient database in the United States. The NIS contains approximately 8 million admission records per year and represents a 20% stratified sample of all acute care hospital admissions nationally [13].

The NIS contains data on patient characteristics including demographics (gender, age, insurance status, and race) and co-morbidities (measured by the Elixhauser Index)[14], hospital characteristics (geographic region, urban vs. rural location, and teaching status), interventions (up to 15 procedures classified), and outcomes (complications recorded in up to 14 secondary diagnoses, length of stay, mortality).

For each year of data, we determined proportions in patient characteristics as well as interventions and outcomes using descriptive statistics. Interventions were classified as debridements, amputations, wound care, wound closure, and HBOT (see Appendix 1). Complications were divided into categories including cardiac, pulmonary, neurologic, thromboembolic, renal, and infectious (see Appendix 2). Complications were also measured as a dichotomous variable indicating any major complication and a continuous variable counting the total number of complications out of these six categories. These national point estimates were calculated using the NIS survey weights and sampling frame. Trends over time in these estimates were then determined using the Cochrane-Armitage trend tests for dichotomous variables and linear regression for continuous variables.

In order to determine if observed epidemiologic trends in outcomes were due to simultaneous trends in case mix or receipt of HBOT, we also conducted multivariable analyses of all patients in our cohort to determine predictors of mortality, any major complication, and LOS for NSTI. Co-variates in the model included year of treatment, age group, sex, race, insurance, Elixhauser index, and HBOT treatment. The models for mortality and LOS also included total major complications. The model for LOS included only those patients who survived hospitalization.

All analyses were performed using SAS statistical software (version 9.2; SAS Institute, Inc, Cary NC). P-value <0.05 for trend tests and co-efficients in multivariable models were considered significant. This study was deemed exempt by the University of Massachusetts Medical School Institutional Review Board.

Results

Co-Morbidities

Patients admitted with NSTIs were more likely to be obese (8.9 to 24.6%, p<0.0001), have uncomplicated diabetes (30.4 to 33.0%, p<0.0001), and chronic liver disease (1.0 to 4.0%, p<0.0001) during the study period. Overall, patients had more preexisting co-morbidities (Elixhauser index > 3 increased from 14.5 to 39.7%, p<0.0001). (see table 2 and figure 1)

Open in a separate window

Figure 1

Trends in Elixhauser > 3, incidence of any major complication, and mortality among NSTI patients 1998–2010 (N=56,527)

Table 2

Trends in specific patient co-morbidities 1998–2010 (N=56,527)

Variable
	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	p-value
	Individual Co-morbidities*
Obese (%)	8.9	11.6	11.2	12.4	16.4	14.5	12.7	13.7	15.7	16.5	20.4	30.0	24.6	<.0001
Diabetes Type 2 (%)
w/Chronic Complication	17.6	17.3	13.7	14.8	14.0	13.8	13.2	14.5	14.1	15.9	14.4	14.3	14.0	<.0001
w/o Chronic Complication	30.4	32.4	34.2	32.4	36.8	32.4	31.6	35.3	34.0	32.5	33.8	39.7	33.0	<.0001
Alcohol abuse (%)	5.9	4.8	4.4	4.5	5.1	5.9	5.1	4.8	5.9	6.1	4.3	6.5	5.7	0.02
Liver disease (%)	1.4	2.3	3.3	3.3	3.9	4.3	4.1	4.0	5.6	3.7	4.5	4.2	4.0	<.0001
Peripheral Vascular Disease (%)	6.9	6.3	5.3	5.0	6.0	6.8	5.4	4.8	4.5	5.8	5.9	7.4	5.7	0.47
	Aggregate Co-morbidity Measure†
Elixhauser Index (%)
0	12.5	13.0	10.9	11.2	11.1	10.7	12.3	10.6	10.1	7.9	8.9	5.7	8.8	<.0001
1	27.2	22.3	25.9	24.4	20.4	19.8	19.6	19.2	17.4	17.7	15.4	12.9	14.4	<.0001
2	28.2	28.5	28.3	27.1	25.7	26.5	26.6	23.3	23.4	23.1	23.2	20.1	20.1	<.0001
3	17.6	20.6	21.2	20.1	21.7	23.5	21.8	23.4	22.4	23.2	22.3	20.8	17.0	0.08
> 3	14.5	15.7	13.7	17.1	21.2	19.4	19.7	23.5	26.6	28.2	30.3	40.6	39.7	<.0001

Open in a separate window

^*These individual co-morbidities have been shown in the literature to be associated with NSTI. See Appendix 3 for a list of all co-morbidities measured by the Elixhauser Index.

^†See Elixhauser, Steiner, Harris, and Coffey, Med Care, 1998;36(1):8–27.

Discussion

NSTI, though decreasing in mortality, is a highly morbid disease. Our national, thirteen year analysis elucidates recent trends in the incidence, treatment, and outcomes for this high acuity surgical illness. We have found that there have been significant national trends in patient population, treatment patterns, and outcomes. Patients have been increasingly from Hispanic and publicly insured groups and overall co-morbidities have risen sharply along with complication rates. Despite worsening acuity, length of stay has been relatively stable and mortality has decreased by nearly 50%.

Our study found demographic changes in the NSTI population, with Hispanics accounting for a greater proportion of patients. A likely explanation for this finding is the rapid expansion of the United States Hispanic population. According to the United States Census Bureau greater than half of the total population growth within the United States during the 2000 to 2010 decade was due to increases in the Hispanic population. This segment grew 43% during this time, seeing an increase of 15.2 million persons[13]. We also found that the proportion of Medicaid patients also increased during our study period. The fact that Hispanics and Medicaid patients are being affected by this morbid infection at such increasing rates raises further questions concerning etiologies, access to care, and prevalence of risk factors that warrants further investigation. Given our findings it is important for clinicians caring for patients of diverse backgrounds to consider NSTI in their differential diagnosis for patients who present with signs and symptoms of soft tissue infection in order to ensure that all patients benefit equally from early diagnosis, aggressive debridement, and improved wound therapy.

Our results suggest, however, that even after diagnosis disparities not explained by factors such as age and co-morbidities may persist. When controlling for these possible confounders, we found that both Hispanics and Medicaid patients have an odds of mortality 1.6 times higher than white patients and privately insured patients, respectively. Meanwhile, Medicaid patients have higher odds of complications and increased lengths of stay compared to patients with private insurance. Black patients also suffer more complications than their white counterparts with NSTI. There is a growing body of evidence that patients in the United States get differential treatment based on race/ethnicity and socioeconomic status. The causes of these healthcare disparities are multiple and the source of much debate[14–17]. One explanation for increased lengths of stay with Medicaid patients may be that their insurance is not favorable for discharge to rehabilitation or skill nursing facilities[18]. Hospital factors may also be the source of differences in complication rates as a number of studies have shown that hospitals that treat more minority and/or indigent patients have worse outcomes[19–21]. When resources and access are otherwise equitable across groups, differences in outcomes by race, ethnicity, or socioeconomic status may be due to unconscious bias[22].

The percentage of NSTI patients who were obese tripled during the study period. This follows the trend of increasing obesity among the general United States population over the past several decades. The current lifetime risk of an American becoming overweight or obese exceeds 50% and 25%, respectively[23]. Given the alarming rates of obesity in our population, the overall importance of obesity on the NSTI population maybe becoming more significant. While obesity prevention initiatives have traditionally focused on reducing risk of cardiovascular disease and certain types of cancer, the association between obesity and NSTIs is an important public health concern. Meanwhile, clinicians faced with an obese patient with seemingly subtle cellulitis with pain out of proportion to exam should have a high index of suspicion for a NSTI. Given the difficulty in diagnosis, this patient population in particular may benefit from contrast enhanced computed tomography to aid in diagnosis[24].

Early and aggressive debridement remains the mainstay of surgical management for NSTIs. During our study period there were simultaneous increases in the proportion of patients requiring only a single debridement (nearly half in 2010) and the proportion of patients requiring multiple debridements (approximately 16% in 2010). Thus, while a certain percentage of patients appear to be benefiting from early diagnosis and aggressive single debridement, the number of patients with severe, progressive disease has also increased although the proportion of patients who need amputation is very small. Notably, the decrease in non-primary skin closure along with increases in wound care procedures suggests that more recent advances in wound therapy such as negative pressure wound therapy may be benefiting NSTI patients in the modern era. More research is needed to understand the role of different wound care options on the management of NSTI patients after all necrotic tissue has been debrided. However, clinicians should consider employing such methods prior to subjecting patients to an additional donor site wounds as would be needed for complex reconstructions.

Despite the many reports regarding the utility of HBOT in the treatment of NSTI[7, 8, 25, 26], including several that report a survival advantage[8, 26], the use of HBOT in the United States is rare. This is likely because HBOT is a specialized treatment not readily available at all hospitals. Patients who did receive HBOT saw no improvements in complication rates or mortality but their hospital LOS was nearly 3.5 days higher than patients who did not undergo HBOT. Thus, we cannot recommend HBOT as an adjunct treatment for NSTI even when readily available.

Overall, the mortality rate of NSTI has declined to 4.9% nationally over the past decade. This is considerably lower than similar national studies using both NIS and NSQIP data (10.9% and 12% respectively)[4, 5]. We found that increasing age was associated with more complications and higher mortality. And, while higher co-morbidity burden increased complications and LOS, this had no effect on mortality. Importantly, however, each added complication increased LOS by almost 9.5 days and resulted in a 2.7 increased odds of mortality when controlling for confounders. This suggests that patients who may have previously died from NSTI now surviving with some measure of morbidity. Possible explanations for the stable LOS and reduced mortality despite the trend toward increased baseline co-morbidities and greater complications during treatment course include increased awareness and early diagnosis by local physicians, improvements in intensive care and resuscitative efforts, better would care options, and improved antibiotic coverage. Our findings have implications for surgeon patient discussions at the time of diagnosis of NSTI. What in the past may have been a discussion about the very high likelihood of death despite aggressive interventions is today more appropriately a detailed discussion of a reasonable chance of survival that comes after a long hospital course involving multiple trips to the operating room for removal of affected tissue and wound care and many potential complications.

This study has a number of important limitations. The NIS is an administrative dataset reliant upon UB-40 billing sheets generated by sampled hospitals. Relying on administrative data may result in underestimation of true NSTIs. However, we believe our selection strategy, combining at least one surgical debridement with a diagnosis code for NSTI, would minimize this effect. Previous research on NSTI has detailed changes in microbiology with single organism infections more prevalent in the past and polymicrobial infection more prevalent in the modern era[10]. Unfortunately, the NIS does not provide any clinical data on wound microbiology or choice of anti-microbial therapy. Without access to clinical data, administrative databases are also subject misclassification/underestimation errors for both measures of co-morbidity and complication severity. The Elixhauser Index, though validated and commonly used in studies using administrative data[27–34], simply measures presence or absence of co-morbid conditions without gauging the individual severity of each co-morbidity with the possible of exception of diabetes (see Appendix 3). Furthermore, there has been recent criticism of Elixhauser’s method but this newer application has yet to be applied widely to the literature due to a delay in implementation of ICD-10 codes in US administrative data[27]. Similarly, we are also unable to measure complication severity using common, validated metrics such as the Accordion Severity Grading System of Surgical Complications, Acute Physiology and Chronic Health Evaluation (APACHE) score or the Sequential Organ Failure Assessment (SOFA) score that rely upon actual clinical data points that are not available in the NIS. Additionally, factors not captured by the administrative database may lead to an underestimation complications rates due both to limitations on the total number of fields available for coding complications and lack of readmission data (NIS treats each admission as a stand-alone event). Our finding of increased co-morbidity burden over the study period may have been due to up-coding by billing specialists or improved documentation by providers rather than a true increase in pre-existing conditions among NSTI patients. However, based on other studies on the overall health of the US population we believe that our findings represent a true change in patient characteristics[35, 36]. It is also possible that increases in complications were attributable similar “improvements” in coding. It is well known that race is coded poorly in administrative data, thus our demographic findings may also be at risk for misclassifications; however, we know of no other large national dataset which would provide more precise race data for NSTI patients. Finally, the overall use of HBOT and amputations was very low and our ability to judge trends in these treatments was thus limited.

Despite these limitations, this nationwide analysis of NSTI is, to our knowledge, the largest population-based sample of NSTI in the US population and has shown that demographics and co-morbidity shifts have occurred and complications have increased. Meanwhile, however, LOS has been relatively stable and mortality has decreased. This suggests improvements of our modern health care system during this study period may be at play. Further research is needed to determine which aspects of anti-microbial treatment, critical care, surgical debridement, and wound management provide these improvements in the overall outcomes of NSTI patients. Future studies looking at NSTI risk factors including socioeconomic status, access to health care, and underlying disease are also necessary to better understand this disease and increasing our ability to prevent, diagnose, and treat it.

Acknowledgments

Footnotes

References