糖心vlog

Key Points

Question听 What is the epidemiology of Staphylococcus aureus bacteremia (SAB) in children?

Findings听 In this cohort study, 30-day mortality in 1153 Australasian children was 4.7% (50 of 1073 children with complete data on mortality), and new risk groups identified for mortality were infants younger than 1 year; M膩ori/Pacific children; those with pneumonia, endocarditis, or sepsis syndrome or no focus; and those who were treated with vancomycin for methicillin-susceptible SAB. Methicillin-resistant SAB and hospital-onset infection were not associated with mortality.

Meaning听 Staphylococcus aureus bacteremia has a wide spectrum of manifestations in children, which differs from that in adults.

Importance听 Staphylococcus aureus bacteremia (SAB) in children causes significant morbidity and mortality, but the epidemiology in children is not well characterized.

Objective听 To describe the epidemiology of SAB in children and adolescents younger than 18 years from Australia and New Zealand.

Design, Setting, and Participants听 A prospective cohort study, using data from the Australian New Zealand Cooperative on Outcomes in Staphylococcal Sepsis cohort for 1153 children with SAB from birth to less than 18 years in pediatric and general hospitals across Australia and New Zealand, collected between January 1, 2007, and December 31, 2012. Multivariate analysis was performed to identify risk factors for mortality. Incidence calculations were calculated separately for Australasian children younger than 15 years using postcode population denominator data from Australian and New Zealand census data.

Main Outcomes and Measures听 Demographic data, hospital length of stay, principal diagnosis, place of SAB onset (community or hospital), antibiotic susceptibility and principal antibiotic treatment, and 7- and 30-day mortality.

Results听 Of the 1153 children with SAB, complete outcome data were available for 1073 children (93.1%); of these, males accounted for 684 episodes (63.7%) of SAB. The median age was 57 months (interquartile range, 2 months to 12 years). The annual incidence of SAB for Australian children was 8.3 per 100鈥�000 population and was higher in indigenous children (incident rate ratio, 3.0 [95% CI, 2.4-3.7]), and the incidence for New Zealand children was 14.4 per 100鈥�000 population and was higher in M膩ori children (incident rate ratio, 5.4 [95% CI, 4.1-7.0]). Community-onset SAB occurred in 761 cases (70.9%), and 142 cases (13.2%) of the infections were methicillin-resistant S aureus (MRSA). Bone or joint infection was most common with 348 cases (32.4%), and endocarditis was uncommon with 30 cases (2.8%). Seven- and 30-day mortality rates were 2.6% (n鈥�=鈥�28) and 4.7% (n鈥�=鈥�50), respectively. Risk factors for mortality were age younger than 1 year; M膩ori or Pacific ethnicity; endocarditis, pneumonia, or sepsis; and receiving no treatment or treatment with vancomycin. Mortality was 14.0% (6 of 43) in children with methicillin-susceptible S aureus (MSSA) treated with vancomycin compared with 2.6% (22 of 851) in children treated with alternative agents (OR, 6.1 [95% CI, 1.9-16.7]). MRSA infection was associated with increased length of stay but not mortality.

Conclusions and Relevance听 In this large cohort study of the epidemiology of SAB in children, death was uncommon, but the incidence was higher for infants and varied by treatment, ethnicity, and clinical presentation. This study provides important information on the epidemiology of SAB in children and risk factors for mortality.

Staphylococcus aureus is an important cause of bacteremia in children and of infectious diseases consultation¹ but is less well characterized in children than adults.^2,3 It is unclear whether vancomycin treatment is inferior to treatment with 尾-lactams in children, as it is in adults with methicillin-susceptible Staphylococcus aureus (MSSA) bacteremia,⁴ and whether there are population-level differences in outcomes based on ethnicity or antibiotic susceptibility.

The global incidence of S aureus bacteremia (SAB) varies substantially: in the Western Pacific region, which includes Australia and New Zealand, all-age annual SAB incidence rates are 11 to 65 per 100鈥�000 population,⁵ but most reported cases are in adults. Pediatric incidence data are limited, although infants younger than 1 year have been recognized as having a higher incidence⁶ and older children a lower incidence of SAB.⁶ In New Zealand, the annual incidence of SAB in children younger than 18 years was 16.9 per 100鈥�000 population overall, but 105 cases per 100鈥�000 for Pacific children younger than 1 year.⁷ In children, mortality with SAB is generally low but is reportedly higher with hospital-onset infection and methicillin-resistant S aureus (MRSA) infection.^8,9 Using data from a prospective, multicenter cohort study鈥攖he largest pediatric cohort to date鈥攚e report the epidemiology, outcomes, and risk factors for mortality from SAB in infants, children, and adolescents younger than 18 years from Australia and New Zealand.

Consecutive episodes of SAB were prospectively identified from institutions participating in the Australian New Zealand Cooperative on Outcomes in Staphylococcal Sepsis (ANZCOSS) study between January 1, 2007, and December 31, 2012.¹⁰ Sites included 33 pediatric, general, and adult hospitals and represented all states and territories of Australia and the North Island of New Zealand. Approval to prospectively collect data and waiver of consent for individual data collection were provided by Southern Health Ethics Committee and additionally by ethics committees of participating institutions, as required.

All data were entered into a web-based database by site investigators. Each anonymized entry was uniquely identified at the participating hospital to allow for follow-up and correction of discrepant results through regular audits.¹⁰ We report on the pediatric patients in the cohort. Incidence rate ratios for pediatric SAB were calculated using numerator data from the ANZCOSS data set for children from birth to 15 years, matched by postcode and indigenous status to denominator data from 2011 Australian census data in Australia.¹¹ Incidence rate ratios for New Zealand children were calculated using data from the 2013 New Zealand Census¹² (details in eTable 1A and B in the Supplement). All other data were collected for each SAB episode from infants and children younger than 18 years. Length of stay (LOS), principal diagnosis, place of SAB onset (community or hospital), antibiotic susceptibility and principal antibiotic treatment, and 7- and 30-day mortality were recorded. Mortality refers to 30-day mortality unless otherwise stated. Infections were considered device related if a device was considered likely to be involved by investigators; these devices included central and peripheral catheters and cardiac, orthopedic, or other localized devices. The pediatric cohort younger than 18 years for which complete data were available was used for all further analyses (eFigure in the Supplement).

An episode of SAB was defined as a single positive blood culture specimen for S aureus in a patient with signs consistent with infection. A new episode in the same patient was recorded if the bacteremia had initially cleared but an additional blood culture performed more than 14 days after the initial positive culture was positive.¹⁰ SAB episodes occurring 48 hours or more and less than 48 hours after hospital admission were considered hospital-onset and community-onset bacteremia, respectively. Community-onset episodes were deemed health care associated if patients had 1 or more of the following risk factors in the previous 12 months: hospitalization, surgery, residency in a long-term care facility, or receipt of renal dialysis.¹⁰ Australian Indigenous children are those reported as Australian Aboriginal and/or Torres Strait Islander children. Neonates were defined as infants from birth to younger than 28 days. Methicillin-susceptible S aureus includes penicillin-susceptible isolates except where otherwise specified. Nonmultiresistant MRSA (nmMRSA) refers to MRSA resistant to 2 or fewer non鈥撐�-lactam antimicrobial classes. Multiresistant MRSA (mrMRSA) refers to MRSA resistant to more than 2 non鈥撐�-lactam antimicrobial classes. Staphylococcus aureus susceptibilities were reported according to standard practice at participating laboratories; methods used were principally Clinical and Laboratory Standards Institute¹³ and Calibrated Dichotomous Sensitivity.¹⁴ Treatment refers to the primary antibiotic used as definitive treatment since only 1 antibiotic was recorded. Combination regimens or initial empirical therapy were not recorded. Endocarditis was determined by site investigators on the basis of available echocardiography.

Descriptive statistics are reported as frequencies and percentages for categorical variables and median and interquartile ranges (IQRs) for continuous variables. Univariate analysis was conducted using 蠂² and Fisher exact tests. Factors considered in univariate analysis were initially included in the multivariate model, and nonsignificant variables were removed in stepwise fashion. Multivariable logistic regression analysis was performed using Stata, version 12.1 (StataCorp LP).

The overall incidence of SAB for Australian children younger than 15 years was 8.3 per 100鈥�000 population. The indigenous to nonindigenous incidence rate ratio was 3.0 (95% CI, 2.4-3.7). The overall incidence for New Zealand children younger than 15 years was 14.4 per 100鈥�000 population. The M膩ori/Pacific to non-M膩ori/Pacific incidence rate ratio was 5.4 (95% CI, 4.1-7.0). Incidence rates and denominator details are reported in eTable 1A and B in the Supplement.

In all, 1153 episodes of pediatric SAB were identified during the 72 months of the study. Complete mortality data were available for 1073 episodes (93.1%); 815 children were from Australia and 258 were from New Zealand (eFigure in the Supplement). There was a mean (SD) of 14.7 (4.8) cases/mo, and little seasonal variation was seen. A total of 1041 children (97.0%) were admitted to the hospital.

Males accounted for 684 of 1073 cases (63.7%) and had more bone or joint infections and fewer presentations with sepsis or no focus (eTable 2 in the Supplement). The origin of SAB was community onset in 761 cases (70.9%); 550 of the total cases (51.3%) were community acquired and 211 cases (19.7%) were health care associated. Median age was 57 months (IQR, 2 months to 12 years). A total of 368 infants (34.3%) younger than 1 year were included in the cohort: 176 (16.4%) were neonates (ie, <28 days) and 192 (17.9%) were older infants (Table 1). Of the 1073 cases of SAB, bone or joint infection was the most frequent clinical manifestation (348 [32.4%] cases) and was seen most often in community-acquired infection (eTable 3 in the Supplement); endocarditis was uncommon (30 [2.8%] cases) (Table 1). Compared with the older cohort, children younger than 1 year were more likely to have device-related infection (164 [44.6%] vs 119 [16.9%]; risk ratio [RR], 2.6; 95% CI, 2.2-3.2; P鈥�<鈥�.001) and hospital-onset infection (221 [60.1%] vs 91 [12.9%]; RR, 4.7; 95% CI, 3.8-5.7; P鈥�<鈥�.001) but were similar overall with respect to MRSA infection rates, antibiotic treatment, and ethnicity.

The cohort included 285 Australian Indigenous children and M膩ori and Pacific children (26.5%); 673 children were white (62.7%) (Table 1). Neonates constituted 15 of 285 (5.3%) Australian Indigenous, M膩ori, and Pacific children, but 161 of 788 (20.4%) other children in the cohort (P鈥�<鈥�.001); age distribution was otherwise similar. There were 127 Australian Indigenous children (14.9%) in the 851 Australian cases and 133 M膩ori and Pacific children (51.6%) in the 258 New Zealand cases. Primary clinical manifestations in these children were more likely to be bone or joint infection (107 of 285 [37.5%] vs 241 of 788 [30.6%]) or skin or soft-tissue infection (55 of 285 [19.3%] vs 106 of 788 [13.5%]) and less likely to be sepsis syndrome or no focus (38 of 285 [13.3%] vs 183 of 788 [23.2%]) or device-related infection (32 of 285 [11.2%] vs 137 of 788 [17.4%]) than in other children (P鈥�&濒迟;鈥�.001).

There were 142 cases (13.2%) of MRSA; susceptibility varied by institution and region (Figure 1) but not by age group or year of study. Overall, 931 of the isolates (86.8%) were MSSA, and 114 MRSA isolates were nmMRSA (80.3%) (eTable 4 in the Supplement). Median LOS was 17 days with MRSA (IQR, 8-38 days) compared with 14 days (IQR, 7-33 days) for MSSA (P鈥�=鈥�.04). Children with MRSA bacteremia were not significantly more likely to be admitted to an intensive care unit (ICU) (odds ratio [OR], 1.4 [95% CI, 0.9-2.2]). Sixty-nine percent of MRSA cases were 鈥渃ommunity onset鈥� (as opposed to 鈥渃ommunity acquired,鈥� which were 69 of 142 cases) Sixty-one percent of MRSA cases were community acquired. The percentages of MRSA infections that were hospital acquired and community acquired were similar to those of MSSA infections (Table 1). Of the Australian Indigenous children, 41 of 127 (32.3%) had MRSA compared with 101 of 946 (10.7%) of all other children (P鈥�<鈥�.005). Of the total cohort, 931 (86.8%) had MSSA infection; 114 (10.6%), nmMRSA infection; and 28 (2.6%), mrMRSA infection.

A 尾-lactam antibiotic, mainly flucloxacillin (a penicillinase-resistant penicillin), was administered in 827 children (77.1%) (eTable 4 in the Supplement). Most children with MRSA were treated with vancomycin (109 of 142 [76.8%]). Forty-three children were not treated; 12 of these children (27.9%) died. Few children were treated with oral agents, and none of this cohort died. Twenty-two of 1073 children (2.1%) received clindamycin or lincomycin (MRSA, 11; MSSA, 11) and none died (eTable 4 in the Supplement).

Intensive care unit admission was needed for 154 children (14.4%), principally for SAB with no focus, sepsis syndrome, or device infection, and 22 of these children (14.3%) died. Median LOS in these children was 44 days (IQR, 20-79 days). Neonatal and pediatric ICU admissions were not differentiated.

Overall 7- and 30-day mortality rates were 2.6% and 4.7% in 28 and 50 children, respectively (Table 1). Risk factors for mortality on univariate analysis were age, antibiotic susceptibility, primary clinical manifestation, and treatment type (Table 2). Risk factors for mortality on multivariate analysis were age younger than 1 year; primary clinical manifestation of endocarditis, pneumonia, or sepsis; receiving no treatment or treatment with vancomycin; and M膩ori/Pacific ethnicity (Table 3). The most common clinical manifestations in children receiving no treatment who died were sepsis or no focus. Vancomycin treatment remained significantly associated with mortality if children receiving no treatment or 尾-lactam treatment for MRSA infection were excluded (OR 2.9; 95% CI, 1.4-6.3).

Six of 43 children with MSSA bacteremia who were treated with vancomycin died, resulting in a 30-day mortality rate of 14.0% compared with a rate of 2.6% in 22 of 851 children with MSSA treated with alternative agents (OR, 6.1 [95% CI, 1.9-16.7]); this group comprised mainly young infants (median age, 0 months [IQR, 0-2 months]). Four children with MRSA bacteremia were treated with flucloxacillin; 1 of these children died. Mortality was highest in 43 children recorded as not treated (12 [27.9%]) for whom the median LOS was 2 days (IQR, 0-6 days); none were admitted to the ICU. There were no deaths in 8 children with neurologic infections. The OR for mortality in children with MRSA bacteremia was 2.2 (95% CI, 1.1-4.3) on univariate analysis (MRSA, 12 of 142 [8.5%]; MSSA, 38 of 931 [4.1%]) but was not significant on multivariate analysis. Nineteen of 312 children (6.1%) with hospital-onset infection and 31 of 761 children (4.1%) with community-onset infection died (P鈥�=鈥�.15). Thirty-five of 50 (70.0%) deaths occurred in children younger than 1 year (Figure 2).

In Australia, 6 of 127 indigenous children died (4.7%) compared with 33 of 688 nonindigenous Australian children (4.8%) (OR, 1.0 [95% CI, 0.3-2.4]). In New Zealand, 6 of 65 M膩ori children died (9.2%) and 4 of 68 Pacific children died (5.9%) compared with 1 of 125 non-M膩ori, non-Pacific children (0.8%). Overall, 10 of 133 M膩ori/Pacific children (7.5%) from New Zealand died compared with 1 of 125 other children (0.8%) from New Zealand (OR, 10.1 [95% CI, 1.4-440.9]).

We report on what we believe to be the largest prospective pediatric cohort of SAB. To our knowledge, this is the first description of the epidemiology of SAB in Australasian children and includes a detailed analysis of risk factors for mortality. Key findings of this study are higher mortality in children younger than 1 year, M膩ori and Pacific children, those with endocarditis and pneumonia, and those treated with vancomycin. The 30-day mortality rate of 4.7% is lower than the 8% to 25% reported for adults^10,15 but comparable to the rates determined in smaller pediatric studies from other regions^8,16,17; MRSA infection was not associated with increased mortality on multivariate analysis. Bone and joint infections were the most common manifestation but were associated with low mortality. The incidence was higher in Australian Indigenous and New Zealand M膩ori children.

Infants younger than 1 year made up 34.3% of the SAB cases, with 47.8% neonates. Staphylococcus aureus is an important contributor to bacteremia in infants, with higher mortality rates than in older children.⁶ Current national antibiotic guidelines do not target neonatal SAB.¹⁸Staphylococcus aureus bacteriemia should be considered in empirical guidelines for treatment of febrile infants, particularly as vaccine-preventable causes of bacteremia (eg, pneumococcal disease) decline.¹⁹

The incidence of SAB in Australian Indigenous children and New Zealand M膩ori children was disproportionately high, and mortality was also greater in M膩ori and Pacific children. Australian Indigenous and M膩ori or Pacific children represented 26.5% of our cohort, and M膩ori or Pacific children made up 51.6% of the cases from New Zealand. Only 4.9% of children younger than 15 years were identified as indigenous in the 2011 Australian census²⁰ compared with 15.6% in our study. Sixteen percent of children younger than 15 years were identified as M膩ori/Pacific peoples in the 2013 New Zealand census compared with 51.6% of those from New Zealand in our study.¹² Australian Indigenous children had significantly more MRSA bacteremia but not increased mortality. Conversely, M膩ori/Pacific ethnicity was significantly associated with mortality, albeit with wide 95% CIs. M膩ori/Pacific children have been reported²¹ to have a disproportionate incidence of S aureus skin infections and hospital admissions. Australian Indigenous (38 of 127 [29.9%]) and M膩ori/Pacific children (24 of 158 [15.2%]) were significantly more likely to have SAB caused by nmMRSA than were other children. Strains of nmMRSA were first described in indigenous Australians in northern Australia²² and are commonly reported in M膩ori and Pacific children.^7,23 Reasons for SAB disproportionately affecting these children include household overcrowding and delayed or inadequate access to health care.⁵ Skin infections were also more common in our study and are well reported²¹ as a likely portal of entry.

Despite MRSA¹⁶ and hospital-onset infection⁸ being reported as risk factors for mortality in smaller cohorts, our large study did not confirm these diagnoses as risk factors. This lack of confirmation may be due to a lower burden of MRSA across our cohort (13.2%)¹⁶ or reflect a different epidemiology of MRSA bacteremia in Australia and New Zealand.¹⁰ MRSA bacteremia was associated with a 3-day increased LOS, representing significant health care costs and separation from family and community for the patient. Reasons for this increased LOS without increased mortality might include delays in starting and thus completing MRSA-targeted therapy or services being less able to manage MRSA in community settings (eg, with hospital in the home [delivery of acute and postacute care in the patient鈥檚 home as a substitute for hospitalization]). The prevalence of MRSA in our study was lower than that in comparable adult populations.^10,24

Most children were treated with 尾-lactams, which is consistent with national guidelines.^18,25 In the Northern Territory (MRSA rate, 34.2%) and Western Australia (20.8%), guidelines now recommend empirical treatment with both a 尾-lactam and vancomycin until susceptibilities are known due to a high prevalence of community-associated MRSA (written communication, S. Tong, MBBS[Hons], PhD, and A. Bowen, MBBS, PhD, February 13, 2016).

In adults, vancomycin treatment has been associated¹⁰ with higher mortality than 尾-lactam therapy. We provide evidence that the same principle holds true in children: those with MSSA bacteremia treated with vancomycin had higher mortality. Although this group was younger, vancomycin therapy was independently associated with mortality when accounting for age in multivariate analysis. Reasons for this increased rate may include poorer efficacy of glycopeptides, suboptimal dosing, or selection bias (eg, sicker patients or those with allergies).^26,27 We recommend that empirical combination of a 尾-lactam and vancomycin should be considered, especially in populations in whom MRSA is prevalent. Combination empirical therapy means that patients with MSSA bacteremia do not receive substandard therapy with vancomycin and that those with MRSA receive early treatment with vancomycin and possibly additional benefit from the 尾-lactam.²⁸ As for adults,¹⁸ definitive therapy for MSSA bacteremia in infants and children should be a 尾-lactam鈥攏ot vancomycin monotherapy.

Endocarditis and pneumonia were significantly associated with mortality in both univariate and multivariate analyses. Endocarditis in children with SAB is uncommon compared with adults,²⁹ but odds for mortality were 19-fold higher (4 of 30 vs 2 of 348) than for bone and joint infection. Staphylococcus aureus bacteremia with pneumonia had 16-fold higher odds (10 of 54) for mortality compared with bone or joint infection. Pneumonia complicating SAB has been identified¹⁰ as a mortality risk factor in a cohort of mainly adults but has not been identified as a risk factor for mortality in children. Endocarditis and pneumonia should be suspected in all children with SAB, investigated with imaging when indicated, and treated aggressively when found. Devices contribute to hospital-acquired S aureus infections in adult settings.³⁰ Device infections have been considered less important in pediatric SAB but made up 26.4% of our cases, including peripheral cannulae (3.4%), central venous access devices (15.5%), and other devices (7.5%). Central venous access devices are an important component of pediatric care, with increasing use globally.³¹Staphylococcus aureus bacteremia should remain a target organism in empirical regimens for line-associated bacteremia. Admission to the ICU was necessary in 14.4% of our cohort, and 48.7% of the cases were associated with health care. Staphylococcus aureus bacteremia has been identified as the leading bacterial cause of sepsis in Australasian pediatric ICUs, highlighting the significance of this condition.³²

To our knowledge, this is the largest reported pediatric cohort of SAB. Despite this size, there are several limitations to the study, including lack of data on whether patients had multiple clinical manifestations; comorbidities, including prematurity; duration of intravenous or oral treatment; use of multiple antibiotics; details on the initial empirical antibiotic regimen; timing of death; attributable mortality; echocardiography; other imaging; relapse; or readmission. Children recorded as having SAB with sepsis or no focus may represent an overestimate in our data set. On close review, Ligon and colleagues³³ found that only 5.7% of children with SAB had no focus compared with 37.1% on initial coding. We were not able to perform secondary medical record reviews on cases with no focus for this purpose. A small proportion of our cohort was listed as not treated (43 [4.0%]) or not admitted (32 [3.0%]), which is highly unusual for children with SAB. It is possible that some isolates from children who were not admitted or not treated but survived were contaminants; however, criteria for inclusion in the study required symptoms and signs consistent with an infection, making this scenario unlikely. Of 12 children who died and were listed as not treated, 9 (75.0%) were admitted to the hospital.

To our knowledge, this study represents the first nationally representative SAB cohort in children from Australia and New Zealand and the largest published series on SAB in children globally. The pattern of SAB in children is different from that in adults, with lower mortality and a distinct spectrum of clinical manifestations. Important risk groups for death from SAB include infants, M膩ori and Pacific children, and those with endocarditis or pneumonia as well as those treated with vancomycin. Methicillin-resistant S aureus itself was not associated with mortality. Further studies are needed to elucidate risk factors more clearly and define effective treatment regimens for severe SAB to reduce morbidity, LOS, and mortality.

Correction: This article was corrected on November 7, 2016, to fix a figure title.

Corresponding Author: Brendan J. McMullan, BMed(Hons), Department of Immunology and Infectious Diseases, Sydney Children鈥檚 Hospital, Randwick, NSW, Australia (b.mcmullan@unsw.edu.au).

Accepted for Publication: May 12, 2016.

Published Online: August 15, 2016. doi:10.1001/jamapediatrics.2016.1477

Author Contributions: Dr Tong and Prof Turnidge contributed equally to the study. Dr McMullan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: McMullan, Blyth, Van Hal, Korman, Tong, Turnidge.

Acquisition, analysis, or interpretation of data: McMullan, Bowen, Blyth, Korman, Buttery, Voss, Roberts, Cooper, Tong, Turnidge.

Drafting of the manuscript: McMullan, Bowen, Blyth, Tong.

Critical revision of the manuscript for important intellectual content: McMullan, Bowen, Van Hal, Korman, Buttery, Voss, Roberts, Cooper, Tong, Turnidge.

Statistical analysis: McMullan, Tong.

Obtained funding: Turnidge.

Administrative, technical, or material support: Blyth, Van Hal, Korman, Roberts, Cooper, Turnidge.

Study supervision: Bowen, Blyth, Voss, Tong.

Conflict of Interest Disclosures: None reported.

Funding/Support: Funding was provided by the Australian Society for Antimicrobials for data management of the original Australian New Zealand Cooperative on Outcomes in Staphylococcal Sepsis (ANZCOSS) study data.

Role of the Funder/Sponsor: The Australian Society for Antimicrobials had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Despina Kotsanas, BSc(Hons), MClinEpi (Monash Infectious Diseases, Monash Health), provided data management of the original ANZCOSS data and cohort data for this study. Kylie-Ann Mallitt, BSc (Hons) (Centre for Big Data Research, University of New South Wales), assisted with statistical analysis. No financial compensation was given for their contributions to this study.