Epidemiology and Clinical Relevance of Toxic Shock Syndrome in US Children

Abstract

Introduction: 

It is important for clinicians to recognize the contribution of toxic shock syndrome (TSS) to the overall burden of pediatric septic shock because the clinical features, optimal therapy and prognosis differ from non-TSS septic shock.

Methods: 

We analyzed cases of pediatric septic shock reported to the Pediatric Health Information Systems database between 2009 and 2013 to define the clinical and demographic characteristics of pediatric TSS in the United States. Using a validated International Classification of Diseases, 9th revision, coding strategy, we identified patients with infectious shock among inpatients age 1–18 years and classified cases of staphylococcal and streptococcal TSS for comparison with non-TSS cases.

Results: 

Of 8,226 cases of pediatric septic shock, 909 (11.1%) were classified as TSS and 562 (6.8%) were possible TSS cases. Staphylococcal TSS represented the majority (83%) of TSS cases and occurred more commonly in females and at an older age. Compared with non-TSS septic shock, TSS had significantly lower fatality rates, disease severity and length of hospital stay and was present more often at the time of admission (P < 0.001 for each). Streptococcal TSS was associated with poorer outcomes than staphylococcal TSS. Treatment for TSS differed from non-TSS septic shock in use of more clindamycin, vancomycin and Intravenous Immunoglobulin and less need for vasopressors.

Conclusions: 

Results demonstrate a significant contribution of TSS to the burden of pediatric septic shock in the United States. The findings emphasize the importance of inclusion of TSS diagnostic and therapeutic considerations in sepsis treatment protocols for children.

Septic shock is a major cause of morbidity and mortality among US children.¹ Because rapid recognition and treatment of pediatric septic shock are essential to ensuring good outcomes, recent initiatives in the United States and elsewhere have urged hospitals caring for children to develop and implement protocols that standardize and systematize clinical management of patients presenting with shock.² Treatment protocols emphasize the need for early recognition, rapid diagnostic testing, aggressive maintenance of circulation and early initiation of empiric antimicrobial treatment, consisting of bactericidal agents which cover a broad spectrum of both Gram-positive and Gram-negative pathogens.²

While such protocols may be effective for management of septic shock because of many bacterial infections, alternative therapeutic approaches may be more effective for disease mediated by bacterial toxins, including both staphylococcal and streptococcal toxic shock syndromes (TSS).³ Several recent epidemiologic studies have examined pediatric septic shock and severe sepsis but do not characterize TSS as a distinct entity.^1,4–6 In a recent analysis of the epidemiology of septic shock cases in Colorado, TSS represented 14.4% of all cases, with a wide age distribution, peak incidence in adolescent females and low mortality relative to non-TSS cases.⁷ In this study, we extend this analysis to a national pediatric population using the Pediatric Health Information Systems (PHIS) database, to describe the relative epidemiologic burden and examine clinical and demographic characteristics of pediatric TSS in the United States.

MATERIALS AND METHODS

Data for this study were obtained from the PHIS, an administrative database affiliated with the Children’s Hospital Association (Overland Park, KS) that contains inpatient, emergency department, ambulatory surgery and observational encounter-level data from over 45 not-for-profit, tertiary care pediatric hospitals representing more than 6 million pediatric subjects with a wide distribution within the United States.⁸ Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and participating hospitals. Data are deidentified at the time of data submission, and data are subjected to a number of reliability and validity checks before being included in the database.

The database was queried for patients with International Classification of Diseases, 9th revision, Clinical Modification codes that identified diagnoses consistent with septic shock, including TSS definitions previously validated in our institution.⁷ The study population was comprised of inpatients 1–18 years of age, discharged during 2009–2013 from the 34 PHIS hospitals with complete information available for that time period. Patients included had at least one of the following codes: 040.82—TSS, 036.3—Waterhouse-Friderichsen syndrome, 785.52—septic shock, 728.86—necrotizing fasciitis or 995.92—severe sepsis. Because the frequency, causes and outcomes of sepsis in infants, particularly in the context of extreme prematurity, may differ substantively from those in older children, we excluded patients less than 1 year of age from this analysis.

For both staphylococcal and streptococcal TSS, we elected to identify two categories of TSS, cases and possible cases, to allow for the possibility that some cases of TSS would be characterized by combined diagnosis codes for septic shock and the specific pathogens without explicit identification of TSS (Supplemental Table, https://links.lww.com/INF/D115). Because MRSA strains in the United States are typically not associated with Toxic Shock Syndrome Toxin-1 (TSST-1) production, they were included in the staphylococcal TSS definition only when TSS was explicitly coded.⁹ Cases with severe sepsis without TSS or septic shock codes as well as those with necrotizing fasciitis without TSS or septic shock were excluded. The remainder of septic shock cases were defined as any diagnosis of septic shock or Waterhouse-Friderichsen syndrome, not included in the aforementioned sets or exclusions (septic shock—not TSS).

All case definitions were based on International Classification of Diseases, 9th revision, Clinical Modification codes assigned by the discharging hospital. Demographic and clinical information was collected from the database for each subject. Clinical data included length of stay, severity score, antibiotic, vasopressor, corticosteroid and Intravenous Immunoglobulin use and whether antibiotics were initiated on day 1 or 2 of the hospital encounter (defined in the database as any period of time up to 11:59 PM on the first or second day of admission), which served as a proxy measure of the presence of infection at time of admission. The severity score is a proprietary component of the 3M All Patient Refined Diagnosis Related Group (DRG) Classification System.¹⁰

All study procedures were approved by the Colorado Multiple Institutional Review Board and were performed in compliance with PHIS External Data Release Guidelines.

Statistical Analysis

For comparison of clinical and demographic characteristics, we used t tests, χ² tests and Mann-Whitney U tests where appropriate. Statistical calculations were performed using Stata v.13 (College Station, TX).

RESULTS

The study population consisted of 8,226 cases of septic shock (Fig. 1). Of the total cases, 11.1% (909) were categorized as TSS, the majority of which (83.1%; 755) were staphylococcal TSS. An additional 6.8% (562) of cases met criteria for possible TSS. MRSA infection was coded explicitly in 50 of the staphylococcal TSS cases. The vast majority (152/154) of streptococcal TSS were explicitly related to discharge codes of TSS plus group A streptococcal infection; two cases were TSS plus necrotizing fasciitis. No secular trend in numbers of TSS or non-TSS septic shock cases was noted for the 5-year study period.

FIGURE 1.: Categorization of pediatric septic shock, TSS, and possible TSS, PHIS database 2009–2013.

Significant differences in demographic and clinical features were noted between TSS subtypes and between TSS and non-TSS septic shock cases (Tables 2 and 3). Cases of staphylococcal TSS were more likely to be older (median, 14 yr) and occurred more often in females (69%) than both streptococcal TSS (median, 8 yr; 54.6% female; P < 0.001 for both) and non-TSS septic shock (median, 9 yr, 49.3% female; P < 0.001 for both). The mean severity of illness score was lower in both forms of TSS than non-TSS septic shock, corresponding to a notable difference in fatality rate (0.55% vs. 13.0%; P < 0.001). Streptococcal TSS had measures of severity that approached those of non-TSS septic shock, with higher severity score, length of stay and increased need for epinephrine compared with staphylococcal TSS (P < 0.001 for all).

TABLE 1.: Case and Possible Case Definitions: Staphylococcal and Streptococcal TSS: Diagnosis (ICD-9-CM Code)

TABLE 2.: Comparison of Demographic and Clinical Features, TSS Versus Non-TSS Septic Shock

TABLE 3.: Comparison of Demographic and Clinical Features: Staphylococcal Versus Streptococcal TSS

While the majority of all septic shock episodes were associated with antibiotic therapy on admission, nearly all TSS cases (97.7%) received antibiotics at admission, compared with non-TSS septic shock cases (90.1%) (Although the data did not allow a direct assessment of the location of onset of infection, this observation served as a proxy measure for cases of TSS cases that were symptomatic upon presentation to the hospital and thus most likely community—as opposed to hospital—acquired.).

Analysis of the possible TSS cases demonstrated that the demographic and clinical characteristics were more consistent with septic shock cases than TSS-specific cases, with younger age (median age, 10 yr), severity score (mean, 3.89) and mortality (9.3%). We also examined the characteristics of staphylococcal TSS associated with diagnosis codes for MRSA, which is not typically associated with TSST-1 toxin production in the United States. In this case, the age (median, 12.5 yr), severity (mean, 2.78) and mortality (0%) were more consistent with cases of staphylococcal TSS than non-TSS septic shock.

DISCUSSION

In this study of more than 8,000 noninfant pediatric patients in US hospitals with septic shock, we identify a significant burden of TSS, which represented an estimated 11.1% of all septic shock cases. Although the overall mortality from TSS was low, reported severity scores and need for intensive cardiovascular therapies suggest that TSS is a potential cause of severe pediatric illness and that early recognition and directed treatment are important to reduce morbidity and assure effective use of medical resources. Staphylococcus was the most common etiology of TSS and occurred more frequently in older, female patients suggesting that menstrual-associated TSS is a key contributor to the overall burden of staphylococcal TSS. Conversely, that 30% of staphylococcal TSS occurred in males, and at a younger age than females confirms that the menstruating female patient is not the sole demographic in which this diagnosis should be considered as was demonstrated in the original description of TSS, in which 3 of the 7 children were males.¹¹ Streptococcal TSS was less common but more severe than staphylococcal TSS and occurred more commonly in younger children with no differences in gender distribution.

Collectively, these data support the need to incorporate TSS in treatment protocols for pediatric septic shock and maintain awareness of this diagnosis among pediatric providers, particularly in ill children with erythroderma, conjunctival hyperemia, strawberry tongue, supportive laboratory findings (elevated blood urea nitrogen and creatinine, elevated liver enzymes and thrombocytopenia) and menstruation/tampon use or other nidus of possible staphylococcal or streptococcal infection.^3,12 Although many of the treatment principles overlap between TSS and non-TSS septic shock, optimal treatment of TSS has substantive differences from non-TSS septic shock, including potentially more aggressive volume resuscitation; localization, drainage and culture of the principal site of infection; removal of infected foreign bodies (tampons, nasal packing, etc.); reduction of toxin production through antibiotics that cause bacterial protein synthesis inhibition and intravenous immune globulin.^3,13–16 Of these, elimination of any nidus of toxin-producing staphylococcal or streptococcal infection and inclusion of clindamycin for protein synthesis–inhibiting antibiotic coverage may be the most important for inclusion in initial septic shock management protocols where TSS is suspected. Drainage of focal staphylococcal infections may reduce the microenvironment conditions necessary for TSST-1 toxin production.¹⁷ Clindamycin has an antimicrobial effect regardless of bacterial growth phase (whereas a large inoculum of bacteria in a stationary growth phase may be less susceptible to cell wall–inhibiting antibiotics) and has been demonstrated in vitro to reduce bacterial toxin production.^18–21 Although definitive clinical data are lacking, reports and case series indicate that inclusion of clindamycin reduces mortality for both staphylococcal and streptococcal TSS.^14,22,23 Our data confirm that clindamycin is commonly, although not universally, utilized (86.7% of all TSS cases) for treatment of pediatric TSS in the United States.

These data also contribute significantly to our understanding of the epidemiology and clinical burden of TSS in children. The PHIS database allows access to a significant proportion of US children, and this study is the largest to explicitly focus on TSS. Ruth et al¹ used the PHIS database to characterize the burden of severe sepsis in US children. Staphylococcus aureus and Streptococcus pyogenes were the two most commonly identified etiologies of severe sepsis, but the contribution of TSS within these groups was not defined. The relative contribution of TSS in our study is consistent with a previous report of adults and children in Colorado using similar methodology, in which definite cases of all ages of TSS represented 14.4% of all sepsis episodes.⁷ Our findings demonstrate a low overall fatality rate of TSS, particularly in staphylococcal TSS. A 1-year survey of reported pediatric TSS in the United Kingdom and Ireland identified 49 cases with a predominance of streptococcal etiology and 16% mortality.²⁴ The high fatality rate in this series is inconsistent with recently reported rates in TSS in the United States and much higher than in our clinical experience with TSS.²⁵ In comparison, the larger numbers, higher median age of staphylococcal TSS (14 vs. 9.5 yr) and lower mortality in our survey likely reflect better capture of less severe cases, particularly menstrual-associated staphylococcal TSS, and suggest that the true incidence of TSS in US children is higher than the 0.38/100,000 estimate from the UK/Ireland study.

It is notable that the PHIS database contained 50 cases of discharge-coded staphylococcal TSS associated with other codes for MRSA. The common strains of MRSA circulating in the United States (USA 300 and to a lesser extent USA 400) are not known to carry genes encoding TSS toxin, although TSS cases attributed to MRSA have been recently reported in burn and postsurgical patients in the United States and elsewhere.^9,26–28 As our data rely on diagnostic coding, it is likely that in some of these cases, one of the diagnoses (MRSA or TSS) was incorrectly coded. However, the similarities in severity and outcome between the MRSA-associated TSS cases and typical TSS suggest that many of the MRSA cases were characterized by an illness clinically consistent with TSS, potentially through action of non-TSST MRSA toxins.²⁹ It is important for practitioners to be aware of this clinical phenomenon, so appropriate antimicrobial therapy for MRSA is included initially until the causative organism can be identified.

Certain limitations of this study must be acknowledged. Most fundamentally, the precision of our estimate of the relative TSS burden is limited by reliance on diagnosis coding. We are also unable to calculate an accurate incidence rate for pediatric TSS, as the underlying population denominator of pediatric patients is not precise for PHIS hospitals. Secular changes in discharge diagnostic coding over the study period (such as have been documented for septic shock cases in the United States) may also lead to artificial changes in disease incidence.³⁰ Although the PHIS database represents a very large pediatric population in the United States, potential bias toward sicker children or those living in proximity to a tertiary care facility may exist, which might exclude milder or unrecognized cases in smaller community hospitals. Furthermore, because the data rely on discharge diagnoses in deidentified patients, we are unable to verify if included patients were correctly classified or documented by treating physicians, nor could we examine the clinical impact of varied management strategies. Such errors are likely to be more common in possible rather than confirmed cases of TSS.

Recent international guidelines emanating from the “surviving sepsis campaign” emphasize the importance of rapid application of evidence-based diagnostic and treatment strategies for critically ill children.² To this end, our findings highlight the significance of staphylococcal and streptococcal TSS within the wider context of pediatric septic shock and emphasize the need for clinicians to recognize specific signs and symptoms of TSS in the pediatric patient presenting with critical illness and to optimize therapeutic interventions when TSS is present.