Neonatal nosocomial bloodstream infections at a referral hospital in a middle-income country: burden, pathogens, antimicrobial resistance and mortality Angela Dramowski1, Ayanda Madide2, Adrie Bekker2 Divisions of 1Paediatric Infectious Diseases and 2Neonatology, Department of Paediatrics and Child Health, Stellenbosch University, Cape Town, South Africa Background: Data on nosocomial bloodstream infection (BSI) rates, pathogens, mortality and antimicrobial resistance in African neonates are limited. Methods: Nosocomial neonatal BSI at Tygerberg Hospital, Cape Town were retrospectively reviewed between 1 January 2009 and 31 December 2013. Laboratory and hospital data were used to determine BSI rates, pathogen profile, mortality and antimicrobial resistance in selected nosocomial pathogens. Results: Of 6521 blood cultures taken over 5 years, 1145 (17.6%) were culture-positive, and 717 (62.6%) discrete nosocomial BSI episodes were identified. Nosocomial BSI rates remained unchanged over time (overall 3.9/1000 patient days, 95% CI 3.6–4.2, x2 for trend P50.23). Contamination rates were relatively high (5.1%, 95% CI 4.6–5.7%). Among BSI pathogens, Gram-negatives predominated (65% vs 31% Gram-positives and 4% fungal); Klebsiella pneumoniae (235, 30%), Staphylococcus aureus (112, 14%) and Enterococci (88, 11%) were most prevalent. Overall crude BSI mortality was 16% (112/717); Gram-negative BSI was significantly associated with mortality (P50.007). Mortality occurred mostly in neonates of very low (33/112, 29%) or extremely low (53/112, 47%) birthweight. Deaths attributed to nosocomial BSI declined significantly over time (x2 for trend P50.01). The prevalence of antibiotic-resistant pathogens was high: methicillin-resistant Staphylococcus aureus 66%, multidrug-resistant A. baumanni 90% and extended-spectrum b-lactamase-producing K. pneumoniae 73%. Conclusion: The burden of nosocomial neonatal BSI at this middle-income country referral neonatal unit is substantial and remained unchanged over the study period, although attributable mortality declined significantly. Nosocomial BSI pathogens exhibited high levels of antimicrobial resistance. Keywords: Neonates, Bloodstream infection, Sepsis, Hospital-acquired infection, Healthcare-associated infection, Nosocomial, Antimicrobial resistance Abbreviations: BSI, bloodstream infection; CI, confidence interval; CLABSI, central line-associated bloodstream infection; CoNs, coagulase negative staphylococci; ESBL, extended-spectrum b-lactamase; HAI, healthcare-associated infection; LMIC, low- and middle-income countries; MRSA, methicillinresistant Staphylococcus aureus; MDR, multidrug-resistant; NICU, neonatal intensive care unit; PMTCT, prevention of mother-to-child transmission of HIV
Introduction The overwhelming majority of the annual three million neonatal deaths globally occur in low- and middleincome countries (LMIC)1 (with gross annual income per capita of v$12,746).2 About one-third of these deaths are caused by neonatal infections including sepsis (15%), pneumonia (10%), diarrhoea (2%) and tetanus (2%).3 The burden of deaths attributable to nosocomial infection in LMICs is uncertain, but is estimated to be 3–20 times higher than in
Correspondence to: A Dramowski, Department of Paediatrics and Child Health, Faculty of Medicine and Health Sciences, Stellenbosch University, PO Box 19063, Tygerberg, 7505, South Africa. Fax: þ 27 21 938 9138; email: [email protected]
ß W. S. Maney & Son Ltd 2015 DOI 10.1179/2046905515Y.0000000029
high-income settings.4,5 Rates of neonatal nosocomial infection (including nosocomial bloodstream infections) reported from high-income settings range from 6 to 9 infections/1000 patient days6 vs 15–62 infections/1000 patient days reported in a review of developing country data.7 Bloodstream infection (BSI) is the most prevalent nosocomial infection affecting neonates and is associated with high mortality.6 Well recognized risk factors for nosocomial BSI in this vulnerable population include prematurity (v37 wks), low birthweight (v2500 g), prolonged hospital stay and the use of indwelling devices.8,9 Additional risk factors for nosocomial infection in LMICs include neonatal HIV exposure and infection, understaffing,
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
265
Dramowski et al.
266
Nosocomial bloodstream infections
overcrowding, lack of isolation rooms, lack of provision for hand hygiene and low levels of awareness and training of healthcare workers in infection prevention and control (IPC).4,10 Despite efforts to establish the nosocomial BSI burden in LMICs, published data are likely to underestimate rates as many LMICs lack the resources for laboratory confirmation of infection, and treat for suspected neonatal sepsis (BSI and/or meningitis) on clinical grounds alone. In addition, most LMIC have limited or no resources to implement rigorous infection surveillance, reporting and post-discharge monitoring for nosocomial infection.4 In contrast, most high-income settings use standardized, computerized reporting systems to track infection rates at individual institutions, and at regional and national level.11 These reports are used to inform development of targeted interventions to reduce neonatal infection. It is likely that the lack of organized surveillance and reporting of nosocomial BSI in LMIC neonates downplays their important and significant contribution to overall neonatal mortality. In addition to accurate estimates of the burden of nosocomial BSI, recent data on trends in antimicrobial resistance in neonatal BSI pathogens in LMICs is required. Of note is the increasing prevalence of enterobacteriaceae producing extended-spectrum b-lactamases (ESBLs) in Africa. Given the limited availability of appropriate antibiotics for drug-resistant infection in these settings, this trend is worrying.12,13 A better understanding of the epidemiology of nosocomial BSI is needed, especially in Africa. This study describes the burden, causative pathogens, mortality and antimicrobial resistance of nosocomial BSI over 5 years at a South African referral neonatal service.
two combined special-care and low-dependency care wards (with bed occupancy rates of 83–138%). Critically ill neonates are nursed in the neonatal intensive care unit (NICU) or high-care wards, with facilities available for respiratory support [nasal continuous positive airway pressure (nCPAP), conventional ventilation and oscillation], inotropic support and nitric oxide therapy. However, given the extreme shortage of NICU beds, nCPAP and central lines are used in both special-care baby wards. In the Western Cape Province, antenatal HIV prevalence increased between 2009 and 2013, from 16.1% to 16.9% (vs 29.5% nationally).15 Combination antiretroviral therapy (cART) has been available since 2004; however, universal cART in pregnancy (irrespective of CD4 count) was introduced only in 2013.16 Between 2009 and 2011, a national prevention of mother-tochild HIV infection transmission (PMTCT) programme achieved reductions in transmission rates in the Western Cape Province, from 3.6% to 1.9%.17 The hospital has an on-site Unit for Infection Prevention and Control (IPC). Routine microbiology laboratory data are used for the surveillance of selected nosocomial infection types (including BSI) and specified bacterial ‘alert’ pathogens. Four infection prevention nurse practitioners are employed (ratio 1:350 patients), with one dedicated to the neonatal and paediatric wards. Institutional changes which might have affected neonatal nosocomial BSI rates were the installation of automated alcohol hand-rub dispensers on all wards and the commencement of an NICU central line-associated BSI (CLABSI) programme in mid-2012. Other than increasing admission volumes, there were no significant changes in patient profile or physician practices during the study period.
Methods Setting
Study population
Tygerberg Hospital is a 1384-bed public teaching hospital in Cape Town, South Africa (an uppermiddle-income country). The hospital’s obstetric service is extremely busy, and annual deliveries increased by 24% between 2009 and 2013, from 5975 to 7855. The neonatal service provides neonatal medical and surgical care for sick and/or low-birthweight (v2500 g) neonates (36.7% low birthweight rate in 2013),14 with prematurity, perinatal asphyxia and neonatal sepsis being the most common reasons for admission. Between 2009 and 2013, annual neonatal admissions increased by 22%, from 4470 to 5756 [including inborn and outborn neonates (neonates from home at v10 days of life or from surrounding clinics and hospitals)]. A 128-bed neonatal unit comprises an intensive care unit, high-care unit, two special-care baby wards and
Nosocomial (hospital-acquired) neonatal BSIs at Tygerberg Hospital were retrospectively reviewed for the period 1 January 2009 to 31 December 2013 (excluding the paediatric wards). Requests for neonatal ward blood-culture during the 5-year study period were reviewed and data on positive cultures were extracted from the electronic laboratory records; additional demographic data on the neonates who died were obtained from the departmental mortality registry (clinical records are routinely reviewed and discussed by the neonatal consultants responsible for monitoring and reporting neonatal deaths, in order to determine cause of death and identify remedial factors). As this study was retrospective and laboratory data-based, additional clinical information regarding the probable source of BSIs, risk factors for BSIs and co-morbidities, e.g. HIV exposure status, was not available.
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
Nosocomial bloodstream infections
Data analysis
Blood culture sampling and analysis
Nosocomial BSI rates, blood culture contamination, pathogen profile, antimicrobial resistance and BSIassociated mortality were determined. Positive blood cultures obtained on or after day 3 of hospitalization (i.e. w48 hours of life or hospitalization) were classified as nosocomial BSIs. Organisms were categorised using the United States Centers for Disease Control’s (US CDC) list of pathogens and contaminants.18 Coagulase-negative staphylococci (CoNs) were classified as pathogens if a repeat blood culture (at least two separate blood draws collected on the same or consecutive calendar days) isolated the same species. A discrete BSI episode was defined as a blood culture that yielded a pathogen/s but excluded any cultures isolating the same pathogen within 14 days of the original specimen. The 14-day cut-off is used to avoid duplicate inclusion of pathogens obtained on subsequent blood cultures prior to appropriate therapy or pathogens that persisted owing to an untreated source of infection. Fluconazole-resistant Candida species, methicillin-resistant Staphylococcus aureus (MRSA), multi-drug resistant Acinetobacter baumannii (resistant to at least three classes of antimicrobials) and ESBL-producing Enterobacteriaceae were classified as antimicrobialresistant pathogens using proposed standard definitions.19 For the Enterobacteriaceae, resistance to cefotaxime was used as a marker of ESBL phenotype. Mortality associated with BSI was determined by consensus opinion of consultant neonatology staff participating in routine mortality review discussions, and had to occur within 7 days of the BSI episode. The following standard definitions were used to stratify the mortality cohort: low birthweight (v2500 g), very low birthweight (v1500 g) and extremely low birthweight (v1000 g).
In most neonates with suspected BSI, a single blood culture sample (one bottle) is submitted at the discretion of attending clinicians. Blood culture specimens may be obtained by peripheral blood draw (the majority) or through a central or arterial line for infants in the NICU (site of blood draw was not available from laboratory records). Local guidelines recommend inoculation of at least 1–2 ml of blood into a paediatric blood culture bottle (paediatric bottles contain specialized media for small-volume samples and charcoal for antibiotic neutralization). Before April 2011, the Bactec system (Becton Dickinson, New Jersey, USA) and BACTEC Peds Plus/F were used, and then replaced according to changes in laboratory protocol with the BacT/Alert system (BioMe´rieux, Marcy l’Etoile, France) and the BacT/ALERTH PF bottle. Specimens are transferred to the on-site National Health Laboratory Service (NHLS) microbiology laboratory for processing; if bacterial growth is detected, a Gram stain is performed and the sample sub-cultured onto appropriate media and incubated overnight. Further identification and antimicrobial susceptibility testing of clinically significant isolates is performed with the automated Vitek II system (BioMe´rieux, Marcy l’Etoile, France) using breakpoints annually published by the Clinical and Laboratory Standards Institute (CLSI).20
Investigation and management of BSIs Blood cultures to identify possible nosocomial BSI pathogens are obtained (iday 3 of life) when there are any clinical, radiological and/or laboratory features which suggest infection. These may include lethargy, apnoea, need for increased respiratory support, poor feeding, temperature instability, abdominal distention, raised white cell count or C-reactive protein, among others. When meningitis is considered a possibility and the clinical condition allows, a cerebrospinal fluid specimen is submitted in addition to blood culture. Given the high prevalence of ESBL-producing Enterobacteriaceae in our institution, empirical treatment of hospital-acquired infection usually includes meropenem. Vancomycin is added if MRSA is considered a likely pathogen, e.g. with suspected central line or soft tissue infection. Fluconazole prophylaxis is not routinely used.
Statistical analysis The nosocomial BSI rate was calculated by dividing the total number of BSI episodes (on or after day 3 of life) by the total inpatient days accumulated during the 5-year period. The pathogen and contamination rates were calculated by dividing the number of blood cultures yielding pathogens and contaminants, respectively, by the total number of blood cultures requested. Continuous and categorical variables were compared using Student’s t-tests and the x2 test, respectively. A x2 test for linear trend was used to assess change in rates over the study period. To determine factors associated with mortality from BSI and antimicrobial resistance, binary logistic regression analyses were performed. Pv0.05 was considered statistically significant. Stata Statistical Software version 13.0 IC (College Station, TX: StataCorp LP) was used.
Ethics approval Ethical approval and waiver of individual informed consent was obtained from the Human Health Research Ethics committee of Stellenbosch University (N10/10/341).
Results A total of 6521 blood cultures specimens were submitted over 5 years for investigation of suspected
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
267
Dramowski et al.
Nosocomial bloodstream infections
10 days (IQR 7–19). The 717 BSI episodes yielded 796 individual pathogens (650 single pathogens, 57 cultures with 2 pathogens, 8 with 3 pathogens and 2 with 4 pathogens). Gram-negative pathogens predominated (519, 65%), followed by Gram-positives (244, 31%) and fungi (33, 4%) (Table 1). K. pneumoniae (235, 30%), S. aureus (112, 14%) and enterococci (88, 11%) were most prevalent (Table 1). Crude nosocomial BSI mortality was 16% (112/717, 95% CI 13–18.5%) (Table 2). Mortality rates varied by type of pathogen, with Gram-negative BSI resulting in significantly greater mortality than Gram-positive and fungal BSI (Pv0.007). Of all monomicrobial BSI pathogens, P. aeruginosa, A. baumannii and S. marcescens were associated with the highest case fatality rates (Table 2). A total of 969 neonates died from all causes over the 5 years, including 139 (14%) neonates with at least one documented episode of BSI. In 112 of these 139 neonates (81%), death was directly attributed to nosocomial BSI. Among the 27 (19%) neonates whose demise was not thought to be BSI-related, death occurred at a median of 51 days after the BSI episode. In addition, many of these neonates had alternative causes of death, e.g. necrotizing enterocolitis (n57), congenital abnormalities (n55), cot death (n52), cardiac tamponade, TPN-related pleural effusion, pneumothorax, massive intraventricular haemorrhage (n51 each) and presumed but
nosocomial BSI from 23,920 hospitalised neonates (one blood culture per 3.7 neonatal admissions). Of 1145 (17.6%) culture-positive specimens, 717 discrete nosocomial BSI episodes were identified [650 (91%) monomicrobial BSI, 67 (9%) polymicrobial, Table 1]. Twenty-three neonates had multiple nosocomial BSI episodes (4 patients with 3 BSI episodes and 19 patients with 2 BSI episodes). The overall nosocomial BSI rate was 3.9/1000 patient days (95% CI 3.6–4.2), and remained unchanged over time (x2 for trend P50.23, Fig. 1). Pathogen yield declined significantly over time from a peak of 12.9 in 2010 to 10.5 in 2013 (x2 for trend P50.03, Fig. 1). Blood culture contamination rates were high, and remained unchanged over time [333/6251 (5.1%), 95% CI 4.6–5.7%, x2 for trend P50.18, Fig. 1). Coagulase-negative staphylococci (CoNS) were the most commonly isolated contaminants (289/333, 87%), followed by non-pathogenic streptococci (15/333, 5%), bacillus species (10/333, 3%), diptheroids (8/333, 2%) and micrococcus species (7/333, 2%). Of the CoNS isolates, five episodes were deemed pathogenic on the basis of isolation of the same species on a second blood culture obtained within 24 hours of the original specimen. Of 717 nosocomial BSI episodes, 395 (55%) were in males. The risk of acquiring nosocomial BSI was 30/1000 neonatal admissions (95% CI 27.9–32.2). Median age at time of blood culture sampling was
Table 1 Microbiological profile of nosocomial bloodstream infection episodes (n5717) BSI episodes
n (%)
Total pathogen count*
Monomicrobial Polymicrobial: 2 pathogens 3 pathogens 4 pathogens Gram-negatives Enterobacteriaceae: Klebsiella pneumoniae Escherichia coli Enterobacter spp Serratia marcescens Other (10 different genera) Non-fermenting Gram-negative bacilli: Acinetobacter baumannii Pseudomonas. aeruginosa Other (2 different genera) Gram-positives Staphylococcus aureus Coagulase negative staphylococci (CoNs){ Group B Streptococcus Enterococcus spp Other (2 different genera) Fungi Candida albicans Candida parapsilosis Candida glabrata All other candida spp
650 (91) 67 (9) 57 8 2 n5519 (% of Gram-negatives)
796
235 (45) 58 (12) 15 (3) 84 (16) 34 (7)
1 6 9 4 –
69 (13) 12 (2) 12 (2) n5244 (% of Gram-positives) 112 (46) 5 (2) 36 (15) 88 (36) 3 (1) n533 (% of fungi) 18 (55) 9 (27) 2 (6) 4 (12)
5 10 – Organism rankz 2 – 7 3 – Organism rankz 8 – – –
Organism rankz
* Total pathogens isolated from 717 BSI episodes: 796 [650 monomicrobial z67 polymicrobial (57|2 isolates) z(8|3 isolates) z(2|4 isolates)]; { coagulase-negative staphylococci were deemed pathogens if the same species was isolated on a second blood culture drawn within 24 hours of the original specimen; z organism rank reported for the top ten isolates only.
268
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
Nosocomial bloodstream infections
Figure 1 Trends in neonatal nosocomial BSI, pathogen yield and blood culture contamination rates (2009–2013) (——— BSI rate, . . .. . .. . . pathogen yield, - - - - - - contamination rate)
culture-negative sepsis (n59). Of the 112 laboratoryconfirmed nosocomial BSI deaths, the majority were of very low birthweight (v1500 g) (86/112, 77%). Table 2 Profile of neonatal deaths nosocomial BSI episodes (n5112) Patient characteristics Overall BSI-associated deaths Male Median gestational age (IQR) Median birthweight, g (IQR) Median age at death, days (IQR) Trend* in overall mortality attributable to nosocomial BSI Microbial characteristics: Death from monomicrobial BSI Death from polymicrobial BSI Crude mortality ranked by pathogen (for monomicrobial BSI only):{ Crude nosocomial BSI-associated mortality P. aeruginosa A. baumannii S. marcescens E. coli C. albicans K. pneumoniae Death rate from top six pathogens associated with mortality Death rate from all other pathogens associated with mortalityz Mortality by pathogen category (for monomicrobial BSI only):{ Gram-negative Gram-positive Fungal
associated
n (%)
with
P-value
112 (100) – 58 (52) 28 (27–31) 1017 (866–1445) 9 (7-13) 112/979 (11) 0.01 n5112 96 (86) 16 (14)
0.07
112/717 (16)
–
4/9 (46) 12/59 (20) 13/66 (20) 10/53 (19) 3/17 (18) 33/204 (16) 75/408 (18)
0.01
21/197 (10)
78/433 (18) 15/186 (8) 3/31 (10)
0.007
* BSI attributable mortality over ‘all-cause’ neonatal mortality by year, assessed by x2 for trend analysis; { deaths from monomicrobial BSI divided by total monomicrobial BSI episodes for that particular pathogen; z Group B streptococcus, Citrobacter spp, Enterobacter spp, Enterococcus spp, K. oxytoca, S. aureus.
Median birthweight was 1017 g (IQR 866–1445) and median gestational age at birth was 28 weeks (IQR 27–31) (Table 2). Median age at death was 9 days (IQR 7–13). Overall mortality attributed to nosocomial BSI declined significantly over time (x2 for trend P50.01). Prevalence of antimicrobial resistance was assessed in a subset of pathogens, focusing on fluconazole-resistant Candida species and six bacterial pathogens: MRSA, multidrug-resistant (MDR) A. baumannii, ESBL-producing and MDR E. coli and K. pneumonia, MDR P. aeruginosa and MDR Serratia marcescens (Table 3). MRSA rates were high (74/112, 66%). The vast majority of A. baumannii isolates fulfilled the criteria for MDR (62/69; 90%). Of Pseudomonas BSI, 4/12 (25%) were classified as MDR isolates. Of 84 S. marcescens BSI, four isolates (5%) were MDR. ESBL carriage rates were very high among K. pneumoniae isolates (172/235, 73%), but much lower among Escherichia coli isolates (7/58, 12%). No carbapenem-resistant Enterobacteriaceae or vancomycin-resistant enterococci were isolated during the study period. Of the 33 Candida isolates, all 18 C. albicans were fluconazole-susceptible, while three of the 15 (20%) non-albicans candida species were fluconazole-resistant. The prevalence of antimicrobial resistance among the selected pathogens did not differ significantly between the first and second halves of the study period.
Discussion The overall rate of neonatal nosocomial BSI determined in the 5-year review (3.9 infections per 1000 patient days) is far lower than that reported from other LMIC settings.4,5,7 This rate, however, includes all hospitalized neonates (not only NICU), and only reports BSI rate (some studies include all types of nosocomial infection).
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
269
Dramowski et al.
Nosocomial bloodstream infections
Table 3 Antimicrobial resistance in selected neonatal nosocomial BSI pathogens Organism(resistance phenotype) S. aureus (MRSA) A. baumanii (MDR AB) K. pneumoniae (ESBL KP) K. pneumonia (MDR KP) E. coli (ESBL EC) E. coli (MDR EC) S. marcescens (MDR SM) P. aeruginosa (MDR PA) Candida spp (fluconazole-resistant)
Total isolates
No. resistant
% resistant
112 69 235 235 58 58 84 12 33
74 62 172 168 7 5 4 4 3
66 90 73 71 12 9 5 25 9
SI, bloodstream infection; MRSA, methicillin resistant Staphylococcus aureus; MDR, multidrug-resistant (according to published criteria); ESBL, extended spectrum b-lactamase producer; AB, Acinetobacter baumanni; KP, Klebsiella pneumoniae; EC, E. coli; SM, Serratia marcescens; PA, Pseudomonas aeruginosa.
However, the incidence of low birthweight in our institution (36.7%) is markedly higher than that reported for Western Cape Province (16.6%), South Africa (14.2%)22 and sub-Saharan Africa (14%).23 In contrast with non-referral centres, our institution has high rates of invasive care and a consequently elevated risk of nosocomial infection. In addition, the number of neonatal admissions and total inpatient days has grown annually but there has been no increase in the number of neonatal staff. In spite of these factors which increase the risk of neonatal infections, the observed ‘low’ BSI rates have been maintained. With no historical data available for comparison, one can only speculate on the reasons for this relatively ‘low’ nosocomial BSI rate. Improvements in the availability of alcohol hand-rub, infection prevention policies, earlier use of isolation for infected infants and the introduction of a central line-associated bloodstream infection (CLABSI) prevention programme from mid-2012 may have contributed to ‘lower’ BSI rates in the latter half of the study period. The relatively high blood culture contamination rates (double the international norm)24 identified in our analysis of neonatal blood culture trends are concerning and warrant a review of blood culture sampling practice. The annual pathogen yield declined significantly over time (as the number of blood culture requests increased), although in 2013 (at 10.5%) it was still higher than blood culture yields cited in a systematic review of blood cultures in Africa (8.2%, 95% CI 7.9–8.4).25 Other possible explanations for the declining pathogen yield include poor aseptic technique with failure to isolate pathogens because of overgrowth by contaminants, low sensitivity of neonatal blood cultures and sub-optimal blood volume inoculum. It is also possible that the change in the laboratory culturing system may be responsible since differences in pathogen yield between the two systems have been described.26–28 This study has several limitations, most importantly related to the retrospective study design. Lack of standardized patient selection for blood culture, unknown sampling technique and blood
The major pathogens in the study, K. pneumoniae and S. aureus, are also important neonatal pathogens globally.5,6 In contrast with other studies of neonatal nosocomial BSI pathogens,5,6 CoNS were not a major contributor, with only five CoNS BSI episodes (0.7%) identified in the cohort. However, the practice of obtaining a single blood culture specimen may have resulted in under-estimation of CoNS BSI. Of note is the emergence of S. marcescens as a major neonatal nosocomial pathogen (the fourth most frequent BSI pathogen in the study), as has been reported internationally6 but not from Africa. Although rates of ESBL production and MRSA prevalence in the cohort’s K. pneumoniae and S. aureus BSI isolates exceed that described in highincome settings, they were in keeping with resistance estimates from Southern Africa.13,21 This is of major concern, given that access to appropriate antibiotics (i.e. carbapenems and vancomycin) is severely limited in most African countries. The low rates of fluconazole resistance are encouraging, and the low prevalence of nosocomial fungal BSI upholds the decision not to use fluconazole prophylaxis. The BSI-attributable mortality (11%) is below the global figure of an estimated 15% of neonatal deaths attributable to sepsis (although that estimate includes maternally-derived BSI and meningitis).3 Of particular interest is the predominance of very low-birthweight among neonates who died from nosocomial BSI. Encouragingly, although overall BSI rates remained unchanged during the study period, the proportion of deaths attributable to nosocomial BSI declined significantly. The trend remains when BSI deaths are analysed by annual admissions, implying that the declining death rate could be ascribed to improved recognition and appropriate management of nosocomial BSI in neonates. In attempting to understand why the BSI rate is far lower than that described from elsewhere is Africa, both our neonatal population and the resources available for neonatal care were considered. In keeping with most African settings, the neonatal service includes a large proportion of low-birthweight infants.
270
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
inoculum, lack of comprehensive clinical data and recent antibiotic use for all BSI episodes are also limitations. Factors associated with mortality and antimicrobial resistance could not be explored owing to a lack of clinical data on the comparator group (e.g. birthweight, HIV exposure status) of neonates who survived nosocomial BSI. The burden of nosocomial neonatal BSI at this middle-income country, neonatal referral unit is substantial, but markedly less than that of low-income settings elsewhere in Africa. Despite a static BSI rate over the study period, mortality attributable to nosocomial BSI was significantly reduced. Pathogens causing nosocomial BSI exhibited substantial antimicrobial resistance, which is of concern for low-income settings with limited access to effective antibiotics. Very low-birthweight neonates account for most nosocomial BSI-associated deaths and should be a specific target population for infection prevention programmes in low–middle-income settings.
Acknowledgments AD and AB are supported by the Medical Research Council Clinician Researcher Programme and the Discovery Foundation Academic Fellowship (but these funding bodies played no role in the design, data collection, data analysis and interpretation, or writing of the manuscript). The authors thank Professor Andrew Whitelaw (of the NHLS and Stellenbosch University’s Department of Medical Microbiology) for assistance with description of the laboratory methods.
Disclaimer statements Authors’ contributions All authors contributed to study design, writing and critical review of the manuscript. AD and AB carried out the laboratory data collection, data cleaning and statistical analysis. AM collected the mortality data. All authors read and approved the final manuscript.
Funding Conflict-of-Interest We have no conflicts of interest to declare.
Ethics Approval Ethical approval and waiver of individual informed consent was obtained from the Human Health Research Ethics committee of Stellenbosch University (N10/10/341).
No competing interests to declare
Nosocomial bloodstream infections
References 1 Lawn JE, Cousens S, Zupan J, Lancet Neonatal Survival Steering Team. Neonatal survival 4 million neonatal deaths: When? Where? Why? Lancet. 2005;365:891–900. 2 World Bank. World Bank Atlas Method. Available from: https:// datahelpdesk.worldbank.org/knowledgebase/articles/378832-whatis-the-world-bank-atlas-method 3 Black RE, Cousens S, Johnson HL, Lawn JE, Rudan I, Bassani DG, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet. 2010; 375:1969–87. 4 Zaidi AK, Huskins WC, Thaver D, Bhutta ZA, Abbas Z, Goldmann DA. Hospital-acquired neonatal infections in developing countries. Lancet. 2005;365:1175–88. 5 Couto RC, Carvalho EA, Pedrosa TM, Pedroso ER, Neto MC, Biscione FM. A 10-year prospective surveillance of nosocomial infections in neonatal intensive care units. Am J Infect Control. 2007;35:183–9. 6 Zingg W, Posfay-Barbe KM, Pittet D. Healthcareassociated infections in neonates. Curr Opin Infect Dis. 2008; 21:228–34. 7 Allegranzi B, Bagheri Nejad S, Combescure C, Graafmans W, Attar H, Donaldson L, et al. Burden of endemic health-careassociated infection in developing countries: systematic review and meta-analysis. Lancet. 2011;377:228–41. 8 Perlman SE, Saiman L, Larson EL. Risk factors for lateonset health care-associated bloodstream infections in patients in neonatal intensive care units. Am J Infect Control. 2007;35: 177–82. 9 Couto RC, Pedrosa TM, Tofani Cde P, Pedroso ER. Risk factors for nosocomial infection in a neonatal intensive care unit. Infect Control Hosp Epidemiol. 2006;27:571–5. 10 Rothe C, Schlaich C, Thompson S. Healthcare-associated infections in sub-Saharan Africa. J Hosp Infect. 2013;85: 257–67. 11 Folgori L, Bielicki J, Sharland M. A systematic review of strategies for reporting of neonatal hospital-acquired bloodstream infections. Arch Dis Child Fetal Neonatal Ed. 2013;98:518–23. 12 Tansarli GS, Poulikakos P, Kapaskelis A, Falagas ME. Proportion of extended-spectrum b-lactamase (ESBL)-producing isolates among Enterobacteriaceae in Africa: evaluation of the evidence. J Antimicrob Chemother. 2014;69:1177–84. 13 Perovic O, Singh-Moodley A, Duse´ A, Bamford C, Elliott G, Swe-Han KS, et al. National sentinel site surveillance for antimicrobial resistance in Klebsiella pneumoniae isolates in South Africa, 2010 - 2012. S Afr Med J. 2014;104:563–8. 14 Mason D, Butt J, Vollmer L, Gebhardt GS, Tygerberg Hospital PPIP 2013: Tygerberg Hospital Perinatal Mortality Database, Cape Town, 2014Available from: http://obstyger.co.za/doc/ Tygerberg_Hospital_Perinatal_data_2002-2013.pdf. 15 National Department of Health, South Africa, National Antenatal Sentinel HIV and Syphilis Prevalence Survey, 2012Available from: http://www.hst.org.za/publications/2012national-antenatal-sentinel-hiv-herpes-simplex-type-2-prevalenc e-survey. 16 National Department of Health, South Africa, The South African Antiretroviral Treatment Guidelines, PMTCT Guidelines, 2013Available from: http://www.hst.org.za/publica tions/pmtct-guidelines-2013. 17 Goga, AE, Dinh, TH, Jackson DJ for the SAPMTCTE Study Group. Evaluation of the Effectiveness of the National Prevention of Mother-to-Child Transmission (PMTCT) Programme Measured at Six Weeks Postpartum in South Africa, 2010. Available from: http://www.mrc.ac.za/healthsystems/SAPMTCTE2010.pdf 18 Centers for Disease Control and Prevention and the National Healthcare Safety Network, CDC/NHSH Bloodstream Infection Event (Central Line-Associated Bloodstream Infection and Non-central Line-associated Bloodstream Infection), 2015Available from: www.cdc.gov/nhsn/XLS/master-organism-Com-Commensals-Lists.xlsx. 19 Magiorakos AP, Srinivasan A, Carey RB, Carmeli Y, Falagas ME, Giske CG, et al. Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect. 2012;18:268–81. 20 Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Susceptibility Testing, Twenty-first Informational Supplement. M100-S21. Wayne, PA: Clinical and Laboratory Standards Institute; 2011.
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
271
Dramowski et al.
Nosocomial bloodstream infections
21 Ballot DE, Nana T, Sriruttan C, Cooper PA. Bacterial bloodstream infections in neonates in a developing country. ISRN Pediatr. 2012;:508512. doi: 10.5402/2012/508512. 22 Pattinson RC, Saving Babies, 2010–2011, 2013Eighth Report on Perinatal Care in South Africa, Available from: http:// www.ppip.co.za/saving-babies/. 23 UNICEF, Low Birthweight: Country, Regional and Global Estimates, 2004Available from: http://www.unicef.org/publica tions/index_24840.html. 24 Hall KK, Lyman JA. Updated review of blood culture contamination. Clin Microbiol Rev. 2006;19:788–802. 25 Reddy EA, Shaw AV, Crump JA. Community-acquired bloodstream infections in Africa: a systematic review and meta-analysis. Lancet Infect Dis. 2010;10:417–32.
272
Paediatrics and International Child Health
2015
VOL .
35
26 Sullivan KV, Turner NN, Lancaster DP, Shah AR, Chandler LJ, Friedman DF, et al. Superior sensitivity and decreased time to detection with the Bactec Peds Plus/F system compared to the BacT/Alert Pediatric FAN blood culture system. J Clin Microbiol. 2013;51:4083–6. 27 Akan OA, Yildiz E. Comparison of the effect of delayed entry into 2 different blood culture systems (BACTEC 9240 and BacT/ALERT 3D) on culture positivity. Diagn Microbiol Infect Dis. 2006;54:193–6. 28 Zadroga R, Williams DN, Gottschall R, Hanson K, Nordberg V, Deike M, et al. Comparison of 2 blood culture media shows significant differences in bacterial recovery for patients on antimicrobial therapy. Clin Infect Dis. 2013;56: 790–7.
NO .
3
Introduction The overwhelming majority of the annual three million neonatal deaths globally occur in low- and middleincome countries (LMIC)1 (with gross annual income per capita of v$12,746).2 About one-third of these deaths are caused by neonatal infections including sepsis (15%), pneumonia (10%), diarrhoea (2%) and tetanus (2%).3 The burden of deaths attributable to nosocomial infection in LMICs is uncertain, but is estimated to be 3–20 times higher than in
Correspondence to: A Dramowski, Department of Paediatrics and Child Health, Faculty of Medicine and Health Sciences, Stellenbosch University, PO Box 19063, Tygerberg, 7505, South Africa. Fax: þ 27 21 938 9138; email: [email protected]
ß W. S. Maney & Son Ltd 2015 DOI 10.1179/2046905515Y.0000000029
high-income settings.4,5 Rates of neonatal nosocomial infection (including nosocomial bloodstream infections) reported from high-income settings range from 6 to 9 infections/1000 patient days6 vs 15–62 infections/1000 patient days reported in a review of developing country data.7 Bloodstream infection (BSI) is the most prevalent nosocomial infection affecting neonates and is associated with high mortality.6 Well recognized risk factors for nosocomial BSI in this vulnerable population include prematurity (v37 wks), low birthweight (v2500 g), prolonged hospital stay and the use of indwelling devices.8,9 Additional risk factors for nosocomial infection in LMICs include neonatal HIV exposure and infection, understaffing,
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
265
Dramowski et al.
266
Nosocomial bloodstream infections
overcrowding, lack of isolation rooms, lack of provision for hand hygiene and low levels of awareness and training of healthcare workers in infection prevention and control (IPC).4,10 Despite efforts to establish the nosocomial BSI burden in LMICs, published data are likely to underestimate rates as many LMICs lack the resources for laboratory confirmation of infection, and treat for suspected neonatal sepsis (BSI and/or meningitis) on clinical grounds alone. In addition, most LMIC have limited or no resources to implement rigorous infection surveillance, reporting and post-discharge monitoring for nosocomial infection.4 In contrast, most high-income settings use standardized, computerized reporting systems to track infection rates at individual institutions, and at regional and national level.11 These reports are used to inform development of targeted interventions to reduce neonatal infection. It is likely that the lack of organized surveillance and reporting of nosocomial BSI in LMIC neonates downplays their important and significant contribution to overall neonatal mortality. In addition to accurate estimates of the burden of nosocomial BSI, recent data on trends in antimicrobial resistance in neonatal BSI pathogens in LMICs is required. Of note is the increasing prevalence of enterobacteriaceae producing extended-spectrum b-lactamases (ESBLs) in Africa. Given the limited availability of appropriate antibiotics for drug-resistant infection in these settings, this trend is worrying.12,13 A better understanding of the epidemiology of nosocomial BSI is needed, especially in Africa. This study describes the burden, causative pathogens, mortality and antimicrobial resistance of nosocomial BSI over 5 years at a South African referral neonatal service.
two combined special-care and low-dependency care wards (with bed occupancy rates of 83–138%). Critically ill neonates are nursed in the neonatal intensive care unit (NICU) or high-care wards, with facilities available for respiratory support [nasal continuous positive airway pressure (nCPAP), conventional ventilation and oscillation], inotropic support and nitric oxide therapy. However, given the extreme shortage of NICU beds, nCPAP and central lines are used in both special-care baby wards. In the Western Cape Province, antenatal HIV prevalence increased between 2009 and 2013, from 16.1% to 16.9% (vs 29.5% nationally).15 Combination antiretroviral therapy (cART) has been available since 2004; however, universal cART in pregnancy (irrespective of CD4 count) was introduced only in 2013.16 Between 2009 and 2011, a national prevention of mother-tochild HIV infection transmission (PMTCT) programme achieved reductions in transmission rates in the Western Cape Province, from 3.6% to 1.9%.17 The hospital has an on-site Unit for Infection Prevention and Control (IPC). Routine microbiology laboratory data are used for the surveillance of selected nosocomial infection types (including BSI) and specified bacterial ‘alert’ pathogens. Four infection prevention nurse practitioners are employed (ratio 1:350 patients), with one dedicated to the neonatal and paediatric wards. Institutional changes which might have affected neonatal nosocomial BSI rates were the installation of automated alcohol hand-rub dispensers on all wards and the commencement of an NICU central line-associated BSI (CLABSI) programme in mid-2012. Other than increasing admission volumes, there were no significant changes in patient profile or physician practices during the study period.
Methods Setting
Study population
Tygerberg Hospital is a 1384-bed public teaching hospital in Cape Town, South Africa (an uppermiddle-income country). The hospital’s obstetric service is extremely busy, and annual deliveries increased by 24% between 2009 and 2013, from 5975 to 7855. The neonatal service provides neonatal medical and surgical care for sick and/or low-birthweight (v2500 g) neonates (36.7% low birthweight rate in 2013),14 with prematurity, perinatal asphyxia and neonatal sepsis being the most common reasons for admission. Between 2009 and 2013, annual neonatal admissions increased by 22%, from 4470 to 5756 [including inborn and outborn neonates (neonates from home at v10 days of life or from surrounding clinics and hospitals)]. A 128-bed neonatal unit comprises an intensive care unit, high-care unit, two special-care baby wards and
Nosocomial (hospital-acquired) neonatal BSIs at Tygerberg Hospital were retrospectively reviewed for the period 1 January 2009 to 31 December 2013 (excluding the paediatric wards). Requests for neonatal ward blood-culture during the 5-year study period were reviewed and data on positive cultures were extracted from the electronic laboratory records; additional demographic data on the neonates who died were obtained from the departmental mortality registry (clinical records are routinely reviewed and discussed by the neonatal consultants responsible for monitoring and reporting neonatal deaths, in order to determine cause of death and identify remedial factors). As this study was retrospective and laboratory data-based, additional clinical information regarding the probable source of BSIs, risk factors for BSIs and co-morbidities, e.g. HIV exposure status, was not available.
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
Nosocomial bloodstream infections
Data analysis
Blood culture sampling and analysis
Nosocomial BSI rates, blood culture contamination, pathogen profile, antimicrobial resistance and BSIassociated mortality were determined. Positive blood cultures obtained on or after day 3 of hospitalization (i.e. w48 hours of life or hospitalization) were classified as nosocomial BSIs. Organisms were categorised using the United States Centers for Disease Control’s (US CDC) list of pathogens and contaminants.18 Coagulase-negative staphylococci (CoNs) were classified as pathogens if a repeat blood culture (at least two separate blood draws collected on the same or consecutive calendar days) isolated the same species. A discrete BSI episode was defined as a blood culture that yielded a pathogen/s but excluded any cultures isolating the same pathogen within 14 days of the original specimen. The 14-day cut-off is used to avoid duplicate inclusion of pathogens obtained on subsequent blood cultures prior to appropriate therapy or pathogens that persisted owing to an untreated source of infection. Fluconazole-resistant Candida species, methicillin-resistant Staphylococcus aureus (MRSA), multi-drug resistant Acinetobacter baumannii (resistant to at least three classes of antimicrobials) and ESBL-producing Enterobacteriaceae were classified as antimicrobialresistant pathogens using proposed standard definitions.19 For the Enterobacteriaceae, resistance to cefotaxime was used as a marker of ESBL phenotype. Mortality associated with BSI was determined by consensus opinion of consultant neonatology staff participating in routine mortality review discussions, and had to occur within 7 days of the BSI episode. The following standard definitions were used to stratify the mortality cohort: low birthweight (v2500 g), very low birthweight (v1500 g) and extremely low birthweight (v1000 g).
In most neonates with suspected BSI, a single blood culture sample (one bottle) is submitted at the discretion of attending clinicians. Blood culture specimens may be obtained by peripheral blood draw (the majority) or through a central or arterial line for infants in the NICU (site of blood draw was not available from laboratory records). Local guidelines recommend inoculation of at least 1–2 ml of blood into a paediatric blood culture bottle (paediatric bottles contain specialized media for small-volume samples and charcoal for antibiotic neutralization). Before April 2011, the Bactec system (Becton Dickinson, New Jersey, USA) and BACTEC Peds Plus/F were used, and then replaced according to changes in laboratory protocol with the BacT/Alert system (BioMe´rieux, Marcy l’Etoile, France) and the BacT/ALERTH PF bottle. Specimens are transferred to the on-site National Health Laboratory Service (NHLS) microbiology laboratory for processing; if bacterial growth is detected, a Gram stain is performed and the sample sub-cultured onto appropriate media and incubated overnight. Further identification and antimicrobial susceptibility testing of clinically significant isolates is performed with the automated Vitek II system (BioMe´rieux, Marcy l’Etoile, France) using breakpoints annually published by the Clinical and Laboratory Standards Institute (CLSI).20
Investigation and management of BSIs Blood cultures to identify possible nosocomial BSI pathogens are obtained (iday 3 of life) when there are any clinical, radiological and/or laboratory features which suggest infection. These may include lethargy, apnoea, need for increased respiratory support, poor feeding, temperature instability, abdominal distention, raised white cell count or C-reactive protein, among others. When meningitis is considered a possibility and the clinical condition allows, a cerebrospinal fluid specimen is submitted in addition to blood culture. Given the high prevalence of ESBL-producing Enterobacteriaceae in our institution, empirical treatment of hospital-acquired infection usually includes meropenem. Vancomycin is added if MRSA is considered a likely pathogen, e.g. with suspected central line or soft tissue infection. Fluconazole prophylaxis is not routinely used.
Statistical analysis The nosocomial BSI rate was calculated by dividing the total number of BSI episodes (on or after day 3 of life) by the total inpatient days accumulated during the 5-year period. The pathogen and contamination rates were calculated by dividing the number of blood cultures yielding pathogens and contaminants, respectively, by the total number of blood cultures requested. Continuous and categorical variables were compared using Student’s t-tests and the x2 test, respectively. A x2 test for linear trend was used to assess change in rates over the study period. To determine factors associated with mortality from BSI and antimicrobial resistance, binary logistic regression analyses were performed. Pv0.05 was considered statistically significant. Stata Statistical Software version 13.0 IC (College Station, TX: StataCorp LP) was used.
Ethics approval Ethical approval and waiver of individual informed consent was obtained from the Human Health Research Ethics committee of Stellenbosch University (N10/10/341).
Results A total of 6521 blood cultures specimens were submitted over 5 years for investigation of suspected
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
267
Dramowski et al.
Nosocomial bloodstream infections
10 days (IQR 7–19). The 717 BSI episodes yielded 796 individual pathogens (650 single pathogens, 57 cultures with 2 pathogens, 8 with 3 pathogens and 2 with 4 pathogens). Gram-negative pathogens predominated (519, 65%), followed by Gram-positives (244, 31%) and fungi (33, 4%) (Table 1). K. pneumoniae (235, 30%), S. aureus (112, 14%) and enterococci (88, 11%) were most prevalent (Table 1). Crude nosocomial BSI mortality was 16% (112/717, 95% CI 13–18.5%) (Table 2). Mortality rates varied by type of pathogen, with Gram-negative BSI resulting in significantly greater mortality than Gram-positive and fungal BSI (Pv0.007). Of all monomicrobial BSI pathogens, P. aeruginosa, A. baumannii and S. marcescens were associated with the highest case fatality rates (Table 2). A total of 969 neonates died from all causes over the 5 years, including 139 (14%) neonates with at least one documented episode of BSI. In 112 of these 139 neonates (81%), death was directly attributed to nosocomial BSI. Among the 27 (19%) neonates whose demise was not thought to be BSI-related, death occurred at a median of 51 days after the BSI episode. In addition, many of these neonates had alternative causes of death, e.g. necrotizing enterocolitis (n57), congenital abnormalities (n55), cot death (n52), cardiac tamponade, TPN-related pleural effusion, pneumothorax, massive intraventricular haemorrhage (n51 each) and presumed but
nosocomial BSI from 23,920 hospitalised neonates (one blood culture per 3.7 neonatal admissions). Of 1145 (17.6%) culture-positive specimens, 717 discrete nosocomial BSI episodes were identified [650 (91%) monomicrobial BSI, 67 (9%) polymicrobial, Table 1]. Twenty-three neonates had multiple nosocomial BSI episodes (4 patients with 3 BSI episodes and 19 patients with 2 BSI episodes). The overall nosocomial BSI rate was 3.9/1000 patient days (95% CI 3.6–4.2), and remained unchanged over time (x2 for trend P50.23, Fig. 1). Pathogen yield declined significantly over time from a peak of 12.9 in 2010 to 10.5 in 2013 (x2 for trend P50.03, Fig. 1). Blood culture contamination rates were high, and remained unchanged over time [333/6251 (5.1%), 95% CI 4.6–5.7%, x2 for trend P50.18, Fig. 1). Coagulase-negative staphylococci (CoNS) were the most commonly isolated contaminants (289/333, 87%), followed by non-pathogenic streptococci (15/333, 5%), bacillus species (10/333, 3%), diptheroids (8/333, 2%) and micrococcus species (7/333, 2%). Of the CoNS isolates, five episodes were deemed pathogenic on the basis of isolation of the same species on a second blood culture obtained within 24 hours of the original specimen. Of 717 nosocomial BSI episodes, 395 (55%) were in males. The risk of acquiring nosocomial BSI was 30/1000 neonatal admissions (95% CI 27.9–32.2). Median age at time of blood culture sampling was
Table 1 Microbiological profile of nosocomial bloodstream infection episodes (n5717) BSI episodes
n (%)
Total pathogen count*
Monomicrobial Polymicrobial: 2 pathogens 3 pathogens 4 pathogens Gram-negatives Enterobacteriaceae: Klebsiella pneumoniae Escherichia coli Enterobacter spp Serratia marcescens Other (10 different genera) Non-fermenting Gram-negative bacilli: Acinetobacter baumannii Pseudomonas. aeruginosa Other (2 different genera) Gram-positives Staphylococcus aureus Coagulase negative staphylococci (CoNs){ Group B Streptococcus Enterococcus spp Other (2 different genera) Fungi Candida albicans Candida parapsilosis Candida glabrata All other candida spp
650 (91) 67 (9) 57 8 2 n5519 (% of Gram-negatives)
796
235 (45) 58 (12) 15 (3) 84 (16) 34 (7)
1 6 9 4 –
69 (13) 12 (2) 12 (2) n5244 (% of Gram-positives) 112 (46) 5 (2) 36 (15) 88 (36) 3 (1) n533 (% of fungi) 18 (55) 9 (27) 2 (6) 4 (12)
5 10 – Organism rankz 2 – 7 3 – Organism rankz 8 – – –
Organism rankz
* Total pathogens isolated from 717 BSI episodes: 796 [650 monomicrobial z67 polymicrobial (57|2 isolates) z(8|3 isolates) z(2|4 isolates)]; { coagulase-negative staphylococci were deemed pathogens if the same species was isolated on a second blood culture drawn within 24 hours of the original specimen; z organism rank reported for the top ten isolates only.
268
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
Nosocomial bloodstream infections
Figure 1 Trends in neonatal nosocomial BSI, pathogen yield and blood culture contamination rates (2009–2013) (——— BSI rate, . . .. . .. . . pathogen yield, - - - - - - contamination rate)
culture-negative sepsis (n59). Of the 112 laboratoryconfirmed nosocomial BSI deaths, the majority were of very low birthweight (v1500 g) (86/112, 77%). Table 2 Profile of neonatal deaths nosocomial BSI episodes (n5112) Patient characteristics Overall BSI-associated deaths Male Median gestational age (IQR) Median birthweight, g (IQR) Median age at death, days (IQR) Trend* in overall mortality attributable to nosocomial BSI Microbial characteristics: Death from monomicrobial BSI Death from polymicrobial BSI Crude mortality ranked by pathogen (for monomicrobial BSI only):{ Crude nosocomial BSI-associated mortality P. aeruginosa A. baumannii S. marcescens E. coli C. albicans K. pneumoniae Death rate from top six pathogens associated with mortality Death rate from all other pathogens associated with mortalityz Mortality by pathogen category (for monomicrobial BSI only):{ Gram-negative Gram-positive Fungal
associated
n (%)
with
P-value
112 (100) – 58 (52) 28 (27–31) 1017 (866–1445) 9 (7-13) 112/979 (11) 0.01 n5112 96 (86) 16 (14)
0.07
112/717 (16)
–
4/9 (46) 12/59 (20) 13/66 (20) 10/53 (19) 3/17 (18) 33/204 (16) 75/408 (18)
0.01
21/197 (10)
78/433 (18) 15/186 (8) 3/31 (10)
0.007
* BSI attributable mortality over ‘all-cause’ neonatal mortality by year, assessed by x2 for trend analysis; { deaths from monomicrobial BSI divided by total monomicrobial BSI episodes for that particular pathogen; z Group B streptococcus, Citrobacter spp, Enterobacter spp, Enterococcus spp, K. oxytoca, S. aureus.
Median birthweight was 1017 g (IQR 866–1445) and median gestational age at birth was 28 weeks (IQR 27–31) (Table 2). Median age at death was 9 days (IQR 7–13). Overall mortality attributed to nosocomial BSI declined significantly over time (x2 for trend P50.01). Prevalence of antimicrobial resistance was assessed in a subset of pathogens, focusing on fluconazole-resistant Candida species and six bacterial pathogens: MRSA, multidrug-resistant (MDR) A. baumannii, ESBL-producing and MDR E. coli and K. pneumonia, MDR P. aeruginosa and MDR Serratia marcescens (Table 3). MRSA rates were high (74/112, 66%). The vast majority of A. baumannii isolates fulfilled the criteria for MDR (62/69; 90%). Of Pseudomonas BSI, 4/12 (25%) were classified as MDR isolates. Of 84 S. marcescens BSI, four isolates (5%) were MDR. ESBL carriage rates were very high among K. pneumoniae isolates (172/235, 73%), but much lower among Escherichia coli isolates (7/58, 12%). No carbapenem-resistant Enterobacteriaceae or vancomycin-resistant enterococci were isolated during the study period. Of the 33 Candida isolates, all 18 C. albicans were fluconazole-susceptible, while three of the 15 (20%) non-albicans candida species were fluconazole-resistant. The prevalence of antimicrobial resistance among the selected pathogens did not differ significantly between the first and second halves of the study period.
Discussion The overall rate of neonatal nosocomial BSI determined in the 5-year review (3.9 infections per 1000 patient days) is far lower than that reported from other LMIC settings.4,5,7 This rate, however, includes all hospitalized neonates (not only NICU), and only reports BSI rate (some studies include all types of nosocomial infection).
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
269
Dramowski et al.
Nosocomial bloodstream infections
Table 3 Antimicrobial resistance in selected neonatal nosocomial BSI pathogens Organism(resistance phenotype) S. aureus (MRSA) A. baumanii (MDR AB) K. pneumoniae (ESBL KP) K. pneumonia (MDR KP) E. coli (ESBL EC) E. coli (MDR EC) S. marcescens (MDR SM) P. aeruginosa (MDR PA) Candida spp (fluconazole-resistant)
Total isolates
No. resistant
% resistant
112 69 235 235 58 58 84 12 33
74 62 172 168 7 5 4 4 3
66 90 73 71 12 9 5 25 9
SI, bloodstream infection; MRSA, methicillin resistant Staphylococcus aureus; MDR, multidrug-resistant (according to published criteria); ESBL, extended spectrum b-lactamase producer; AB, Acinetobacter baumanni; KP, Klebsiella pneumoniae; EC, E. coli; SM, Serratia marcescens; PA, Pseudomonas aeruginosa.
However, the incidence of low birthweight in our institution (36.7%) is markedly higher than that reported for Western Cape Province (16.6%), South Africa (14.2%)22 and sub-Saharan Africa (14%).23 In contrast with non-referral centres, our institution has high rates of invasive care and a consequently elevated risk of nosocomial infection. In addition, the number of neonatal admissions and total inpatient days has grown annually but there has been no increase in the number of neonatal staff. In spite of these factors which increase the risk of neonatal infections, the observed ‘low’ BSI rates have been maintained. With no historical data available for comparison, one can only speculate on the reasons for this relatively ‘low’ nosocomial BSI rate. Improvements in the availability of alcohol hand-rub, infection prevention policies, earlier use of isolation for infected infants and the introduction of a central line-associated bloodstream infection (CLABSI) prevention programme from mid-2012 may have contributed to ‘lower’ BSI rates in the latter half of the study period. The relatively high blood culture contamination rates (double the international norm)24 identified in our analysis of neonatal blood culture trends are concerning and warrant a review of blood culture sampling practice. The annual pathogen yield declined significantly over time (as the number of blood culture requests increased), although in 2013 (at 10.5%) it was still higher than blood culture yields cited in a systematic review of blood cultures in Africa (8.2%, 95% CI 7.9–8.4).25 Other possible explanations for the declining pathogen yield include poor aseptic technique with failure to isolate pathogens because of overgrowth by contaminants, low sensitivity of neonatal blood cultures and sub-optimal blood volume inoculum. It is also possible that the change in the laboratory culturing system may be responsible since differences in pathogen yield between the two systems have been described.26–28 This study has several limitations, most importantly related to the retrospective study design. Lack of standardized patient selection for blood culture, unknown sampling technique and blood
The major pathogens in the study, K. pneumoniae and S. aureus, are also important neonatal pathogens globally.5,6 In contrast with other studies of neonatal nosocomial BSI pathogens,5,6 CoNS were not a major contributor, with only five CoNS BSI episodes (0.7%) identified in the cohort. However, the practice of obtaining a single blood culture specimen may have resulted in under-estimation of CoNS BSI. Of note is the emergence of S. marcescens as a major neonatal nosocomial pathogen (the fourth most frequent BSI pathogen in the study), as has been reported internationally6 but not from Africa. Although rates of ESBL production and MRSA prevalence in the cohort’s K. pneumoniae and S. aureus BSI isolates exceed that described in highincome settings, they were in keeping with resistance estimates from Southern Africa.13,21 This is of major concern, given that access to appropriate antibiotics (i.e. carbapenems and vancomycin) is severely limited in most African countries. The low rates of fluconazole resistance are encouraging, and the low prevalence of nosocomial fungal BSI upholds the decision not to use fluconazole prophylaxis. The BSI-attributable mortality (11%) is below the global figure of an estimated 15% of neonatal deaths attributable to sepsis (although that estimate includes maternally-derived BSI and meningitis).3 Of particular interest is the predominance of very low-birthweight among neonates who died from nosocomial BSI. Encouragingly, although overall BSI rates remained unchanged during the study period, the proportion of deaths attributable to nosocomial BSI declined significantly. The trend remains when BSI deaths are analysed by annual admissions, implying that the declining death rate could be ascribed to improved recognition and appropriate management of nosocomial BSI in neonates. In attempting to understand why the BSI rate is far lower than that described from elsewhere is Africa, both our neonatal population and the resources available for neonatal care were considered. In keeping with most African settings, the neonatal service includes a large proportion of low-birthweight infants.
270
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
Dramowski et al.
inoculum, lack of comprehensive clinical data and recent antibiotic use for all BSI episodes are also limitations. Factors associated with mortality and antimicrobial resistance could not be explored owing to a lack of clinical data on the comparator group (e.g. birthweight, HIV exposure status) of neonates who survived nosocomial BSI. The burden of nosocomial neonatal BSI at this middle-income country, neonatal referral unit is substantial, but markedly less than that of low-income settings elsewhere in Africa. Despite a static BSI rate over the study period, mortality attributable to nosocomial BSI was significantly reduced. Pathogens causing nosocomial BSI exhibited substantial antimicrobial resistance, which is of concern for low-income settings with limited access to effective antibiotics. Very low-birthweight neonates account for most nosocomial BSI-associated deaths and should be a specific target population for infection prevention programmes in low–middle-income settings.
Acknowledgments AD and AB are supported by the Medical Research Council Clinician Researcher Programme and the Discovery Foundation Academic Fellowship (but these funding bodies played no role in the design, data collection, data analysis and interpretation, or writing of the manuscript). The authors thank Professor Andrew Whitelaw (of the NHLS and Stellenbosch University’s Department of Medical Microbiology) for assistance with description of the laboratory methods.
Disclaimer statements Authors’ contributions All authors contributed to study design, writing and critical review of the manuscript. AD and AB carried out the laboratory data collection, data cleaning and statistical analysis. AM collected the mortality data. All authors read and approved the final manuscript.
Funding Conflict-of-Interest We have no conflicts of interest to declare.
Ethics Approval Ethical approval and waiver of individual informed consent was obtained from the Human Health Research Ethics committee of Stellenbosch University (N10/10/341).
No competing interests to declare
Nosocomial bloodstream infections
References 1 Lawn JE, Cousens S, Zupan J, Lancet Neonatal Survival Steering Team. Neonatal survival 4 million neonatal deaths: When? Where? Why? Lancet. 2005;365:891–900. 2 World Bank. World Bank Atlas Method. Available from: https:// datahelpdesk.worldbank.org/knowledgebase/articles/378832-whatis-the-world-bank-atlas-method 3 Black RE, Cousens S, Johnson HL, Lawn JE, Rudan I, Bassani DG, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet. 2010; 375:1969–87. 4 Zaidi AK, Huskins WC, Thaver D, Bhutta ZA, Abbas Z, Goldmann DA. Hospital-acquired neonatal infections in developing countries. Lancet. 2005;365:1175–88. 5 Couto RC, Carvalho EA, Pedrosa TM, Pedroso ER, Neto MC, Biscione FM. A 10-year prospective surveillance of nosocomial infections in neonatal intensive care units. Am J Infect Control. 2007;35:183–9. 6 Zingg W, Posfay-Barbe KM, Pittet D. Healthcareassociated infections in neonates. Curr Opin Infect Dis. 2008; 21:228–34. 7 Allegranzi B, Bagheri Nejad S, Combescure C, Graafmans W, Attar H, Donaldson L, et al. Burden of endemic health-careassociated infection in developing countries: systematic review and meta-analysis. Lancet. 2011;377:228–41. 8 Perlman SE, Saiman L, Larson EL. Risk factors for lateonset health care-associated bloodstream infections in patients in neonatal intensive care units. Am J Infect Control. 2007;35: 177–82. 9 Couto RC, Pedrosa TM, Tofani Cde P, Pedroso ER. Risk factors for nosocomial infection in a neonatal intensive care unit. Infect Control Hosp Epidemiol. 2006;27:571–5. 10 Rothe C, Schlaich C, Thompson S. Healthcare-associated infections in sub-Saharan Africa. J Hosp Infect. 2013;85: 257–67. 11 Folgori L, Bielicki J, Sharland M. A systematic review of strategies for reporting of neonatal hospital-acquired bloodstream infections. Arch Dis Child Fetal Neonatal Ed. 2013;98:518–23. 12 Tansarli GS, Poulikakos P, Kapaskelis A, Falagas ME. Proportion of extended-spectrum b-lactamase (ESBL)-producing isolates among Enterobacteriaceae in Africa: evaluation of the evidence. J Antimicrob Chemother. 2014;69:1177–84. 13 Perovic O, Singh-Moodley A, Duse´ A, Bamford C, Elliott G, Swe-Han KS, et al. National sentinel site surveillance for antimicrobial resistance in Klebsiella pneumoniae isolates in South Africa, 2010 - 2012. S Afr Med J. 2014;104:563–8. 14 Mason D, Butt J, Vollmer L, Gebhardt GS, Tygerberg Hospital PPIP 2013: Tygerberg Hospital Perinatal Mortality Database, Cape Town, 2014Available from: http://obstyger.co.za/doc/ Tygerberg_Hospital_Perinatal_data_2002-2013.pdf. 15 National Department of Health, South Africa, National Antenatal Sentinel HIV and Syphilis Prevalence Survey, 2012Available from: http://www.hst.org.za/publications/2012national-antenatal-sentinel-hiv-herpes-simplex-type-2-prevalenc e-survey. 16 National Department of Health, South Africa, The South African Antiretroviral Treatment Guidelines, PMTCT Guidelines, 2013Available from: http://www.hst.org.za/publica tions/pmtct-guidelines-2013. 17 Goga, AE, Dinh, TH, Jackson DJ for the SAPMTCTE Study Group. Evaluation of the Effectiveness of the National Prevention of Mother-to-Child Transmission (PMTCT) Programme Measured at Six Weeks Postpartum in South Africa, 2010. Available from: http://www.mrc.ac.za/healthsystems/SAPMTCTE2010.pdf 18 Centers for Disease Control and Prevention and the National Healthcare Safety Network, CDC/NHSH Bloodstream Infection Event (Central Line-Associated Bloodstream Infection and Non-central Line-associated Bloodstream Infection), 2015Available from: www.cdc.gov/nhsn/XLS/master-organism-Com-Commensals-Lists.xlsx. 19 Magiorakos AP, Srinivasan A, Carey RB, Carmeli Y, Falagas ME, Giske CG, et al. Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect. 2012;18:268–81. 20 Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Susceptibility Testing, Twenty-first Informational Supplement. M100-S21. Wayne, PA: Clinical and Laboratory Standards Institute; 2011.
Paediatrics and International Child Health
2015
VOL .
35
NO .
3
271
Dramowski et al.
Nosocomial bloodstream infections
21 Ballot DE, Nana T, Sriruttan C, Cooper PA. Bacterial bloodstream infections in neonates in a developing country. ISRN Pediatr. 2012;:508512. doi: 10.5402/2012/508512. 22 Pattinson RC, Saving Babies, 2010–2011, 2013Eighth Report on Perinatal Care in South Africa, Available from: http:// www.ppip.co.za/saving-babies/. 23 UNICEF, Low Birthweight: Country, Regional and Global Estimates, 2004Available from: http://www.unicef.org/publica tions/index_24840.html. 24 Hall KK, Lyman JA. Updated review of blood culture contamination. Clin Microbiol Rev. 2006;19:788–802. 25 Reddy EA, Shaw AV, Crump JA. Community-acquired bloodstream infections in Africa: a systematic review and meta-analysis. Lancet Infect Dis. 2010;10:417–32.
272
Paediatrics and International Child Health
2015
VOL .
35
26 Sullivan KV, Turner NN, Lancaster DP, Shah AR, Chandler LJ, Friedman DF, et al. Superior sensitivity and decreased time to detection with the Bactec Peds Plus/F system compared to the BacT/Alert Pediatric FAN blood culture system. J Clin Microbiol. 2013;51:4083–6. 27 Akan OA, Yildiz E. Comparison of the effect of delayed entry into 2 different blood culture systems (BACTEC 9240 and BacT/ALERT 3D) on culture positivity. Diagn Microbiol Infect Dis. 2006;54:193–6. 28 Zadroga R, Williams DN, Gottschall R, Hanson K, Nordberg V, Deike M, et al. Comparison of 2 blood culture media shows significant differences in bacterial recovery for patients on antimicrobial therapy. Clin Infect Dis. 2013;56: 790–7.
NO .
3
