Aetiology of community-acquired neonatal sepsis in low and middle income countries.

2011·0.67 MB·English

by  Waters, Donald

journal of
global
health
S
T
N
I
OS aetiology of community-acquired neonatal
Pr
WE
EP
Ia sepsis in low- and middle-income countries
VP
Donald Waters1,
Background 99% of the approximate 1 million annual neonatal
Issrah Jawad1, aziez ahmad2,
deaths from life-threatening invasive bacterial infections occur in
Ivana Lukšić3, Harish Nair1, developing countries, at least 50% of which are from home births
Lina Zgaga1, Evropi or community settings. Data concerning aetiology of sepsis in these
settings are necessary to inform targeted therapy and devise man-
Theodoratou1, Igor rudan1*,
agement guidelines. This review describes and analyses the bacte-
anita K.M. Zaidi2*,
rial aetiology of community-acquired neonatal sepsis in developing
Harry Campbell1* countries.
Methods A search of Medline, Embase, Global Health and Web of
1 Centre for Population Health Sciences and Knowledge, limited to post-1980, found 27 relevant studies. Data
Global Health academy, The University of on aetiology were extracted, tabulated and analysed along with data
Edinburgh, Scotland, UK
on incidence, risk factors, case fatality rates and antimicrobial sen-
2 Department of Paediatrics and Child Health, sitivity.
aga Khan University, Karachi, Pakistan
Results The most prevalent pathogens overall were Staphylococcus
3 Department of Microbiology, Dubrava
aureus (14.9%), Escherichia coli (12.2%), and Klebsiella species
University Hospital, Zagreb, Croatia
(11.6%). However, variations were observed both between global
* Joint senior authorship
regions and age-of-onset categories. Staphylococcus aureus and Strep-
tococcus pneumoniae were most prevalent in Africa, while Klebsiella
was highly prevalent in South-East Asia. A notably higher prevalence
of Group B Streptococcus was present in neonates aged 7 days or less.
The highest case fatality rates were recorded in South-East Asia. Kleb-
siella species showed highest antimicrobial resistance.
Conclusion Data on community-acquired neonatal sepsis in devel-
oping countries are limited. Future research should focus on areas
of high disease burden with relative paucity of data. Research into
maternal and neonatal vaccination strategies and improved diagnos-
tics is also needed. All of this could contribute to the formulation of
community-based care packages, the implementation of which has
significant potential to lower overall neonatal mortality and hence
advance progress towards the attainment of Millennium Develop-
Correspondence to:
ment Goal 4.
Prof. Harry Campbell
Centre for Population Health Sciences
University of Edinburgh
Approximately 1 million deaths a year occurring in the neonatal period
Teviot Place
Edinburgh EH8 9aG (0–28 days) are caused by infection, accounting for over 25% of global
Scotland, UK neonatal deaths and 10% of all mortality in infants under the age of 5
[email protected] (1); 99% of these deaths occur in developing countries (2). Neonatal sep-
154 December, 2011 • Vol. 1 No. 2 / journal of global health • www.jogh.org
Aetiology of community-acquired neonatal sepsis in developing countries
sis is classically defined as the presence of symptoms of unsterile cord cutting and potentially unsafe cultural cus-
sepsis in the neonatal period combined with bacteriologi- toms such as spreading dung on the newborn’s umbilicus
cal isolation of an infectious agent from blood or cerebro- (10). Other predisposing risk factors for infection in neo-
spinal fluid (CSF) (3). It is classified as ‘early-onset’ if it nates include low birth-weight, prematurity, prolonged
occurs within the first 7 days of life and as ‘late-onset’ if it rupture of membranes and a long delivery period (3).
occurs after this time. Typically, early-onset sepsis is con-
Health education can also be a problem with early detec-
sidered maternally-acquired, usually from the maternal S
tion of CANS in developing countries often being low, po- T
genital tract, and late-onset sepsis is generally regarded to N
tentially due to mothers failing to notice important symp- I
originate from the care-giving environment – either a SO
toms and seek healthcare. The role of women in some rP
healthcare or community setting. Consequently early- and EW
late-onset sepses are also associated with different distri- societies is also an issue, with woman having a low social aPE
butions of pathogens (4). standing and a lack of autonomy resulting in delays in, or PVI
absence of, care seeking for infant’s health, poorer sanita-
The majority of babies in developing countries are born at tion and a decrease in access to healthcare facilities (10).
home and at least a half of neonatal deaths occur in home
births (5). Reasons for this include poor health system cov- Management
erage or provision and limited or no access to referral fa-
Management of neonatal sepsis in a hospital setting is com-
cilities (6). There is significant evidence that in rural areas
monly through parenteral antibiotic therapy and support-
neonates often do not receive required healthcare and that
ive care, which has shown positive impacts (10). However,
this is associated with an increase in mortality (7). How-
it is important to note that most neonates in developing
ever, other reasons for the high prevalence of home births
countries do not receive this therapy because they do not
can be suggested, such as financial constraints. This is be-
have access to the necessary health services, or their par-
cause even when health care services are available, and of
ents do not seek care. A recent review showed a significant
respectable quality, they may still remain beyond the finan-
reduction in mortality from CANS as a result of introduc-
cial means of many (8). There are also potential sociocul-
ing perinatal care packages including injectable antibiotics
tural issues related to the rejection of health care services
to the community (6). Research shows that the aetiology
for newborns, because research has demonstrated a high
of neonatal sepsis is continually evolving, and therefore
prevalence of refusal of hospital referrals by their families
continuing updating of aetiological data is necessary to in-
and highlighted the need for education programmes on ap-
form appropriately targeted therapy (10). A previous re-
propriate care seeking (9).
view of CANS showed a predominance of gram-negative
The predominance of home births in developing countries
organisms over gram-positive, with the main causative
is not reflected in related research as this mainly provides
pathogens being Klebsiella species, Escherichia coli and
data on neonatal sepsis in hospitals, a large percentage of
Staphylococcus aureus (4). The focus of this review will be
which is nosocomial (10). There are several potential rea-
to provide updated data on the aetiology of CANS globally.
sons behind the lack of aetiological data on neonatal sepsis
acquired in the community, including lack of sufficient lab- Prevention
oratory facilities in rural areas and also potentially low lev-
In addition to efforts to improve diagnosis and treatment
els of care seeking, resulting in much unreported morbid-
ity and mortality. This may particularly be true for the of CANS, efforts to prevent this life-threatening illness are
cases that occur in the areas without access to health care, also important to consider. A review of possible preventa-
or in areas with poorly developed care-seeking behaviour tive interventions for improving neonatal health highlight-
(4). This review is concerned specifically with the bacte- ed a need for universal provision of antenatal care for moth-
rial aetiology of life-threatening, community-acquired neo- ers in developing countries as a means of decreasing
natal sepsis (CANS). mortality from neonatal sepsis (13). This involves educat-
ing mothers about hygienic birth practice, promoting
Pathogenesis and risk factors breast feeding and also detecting and treating important
maternal risk factors for neonatal sepsis, such as asymp-
Due to their immature immune systems and incompletely
tomatic bacteriuria (13). Another important potential way
developed skin barriers, neonates are more susceptible to
to prevent neonatal sepsis is to train and provide adequate
infection (11). In developing countries, the likelihood of
numbers of skilled birth attendants in the community (10).
infection is increased due to other additional risk factors.
Unsafe birthing practices are common, with only 35% of Possibly one of the most important preventative interven-
births in some of the least developed countries being at- tions after birth is early and exclusive breastfeeding. Breast
tended by a skilled birth attendant (12), often resulting in milk contains important immunological factors, some of
unhygenic practices such as delivery onto a unsterile floor, which have the potential to inhibit causative pathogens of
www.jogh.org • journal of global health / December, 2011 • Vol. 1 No. 2 155
Donald Waters et al.
neonatal sepsis (11). This is a major issue, as recent re- view were to determine the bacterial aetiology of CANS in
search has shown that only 37% of infants younger than developing countries through systematic literature review,
six months of age in developing countries are exclusively to investigate aetiological variations between global regions
breastfed (14). Consequently, promotion of breastfeeding and different ages-of-onset and to explore potential sugges-
in community settings is the subject of an extensive World tions from information presented for future policy and re-
Health Organization (WHO) strategy document (15). search.
S
T
N International responses to neonatal sepsis
I METHODS
OS
Pr Neonatal sepsis is an important issue internationally, espe-
WE cially with relation to the United Nations Millennium De- A review of published literature was undertaken using the
EP
Ia velopment Goals. Without a reduction in newborn deaths, electronic databases Medline, Embase, Global Health and
VP
of which sepsis is a major cause, the fourth goal of reduc- Web of Knowledge. The search involved combinations of
ing mortality in children under five by two-thirds cannot Medical Subject Headings (MeSH) and keywords in con-
be achieved (2). There is therefore a need to investigate junction with a search for each individual developing coun-
prevention, diagnosis and treatment strategies and their try. These were defined as low- or middle-income countries
potential for implementation or improvement globally (8). from World Bank classifications (19). Search terms used for
Medline and Embase are shown in Table 1. Terms for oth-
Founded in 1992 by the WHO and the United Nations
Children’s Fund (UNICEF), the Integrated Management of
Childhood Illness initiative (IMCI) is an integrated ap- Table 1 Search terms for Medline/Embase
proach to improving child health globally, which provides 1. Developing Countries/ or Algeria/ or Egypt/ or Libya/ or Morocco/
guidelines including curative and preventative elements in or Tunisia/ or Cameroon/ or Central African Republic/ or Chad/ or
Congo/ or “Democratic Republic of the Congo”/ or Gabon/ or Bu-
both healthcare and community settings. Rather than an
rundi/ or Djibouti/ or Eritrea/ or Ethiopia/ or Kenya/ or Rwanda/ or
individual disease-specific approach, IMCI has a wide and Somalia/ or Sudan/ or Tanzania/ or Uganda/ or Angola/ or Botswa-
integrated strategy, aiming at addressing the varied risk fac- na/ or Lesotho/ or Malawi/ or Mozambique/ or Namibia/ or South
Africa/ or Swaziland/ or Zambia/ or Zimbabwe/ or Benin/ or Burki-
tors for childhood illness (16). This approach is highly rel-
na Faso/ or Cape Verde/ or Cote d’Ivoire/ or Gambia/ or Ghana/ or
evant in the case of neonatal sepsis as the disease has many Guinea/ or Guinea-Bissau/ or Liberia/ or mail/ or Mauritania/ or Ni-
ger/ or Nigeria/ or Senegal/ or Sierra Leone/ or Togo/ or “Antigua
risk factors and can be both healthcare- and community-
and Barbuda”/ or Cuba/ or Dominica/ or Dominican Republic/ or
acquired. In many settings where CANS is prevalent, high- Grenada/ or Haiti/ or Jamaica/ or “Saint Kitts and Nevis”/ or Saint
quality diagnostic facilities are not widely available and the Lucia/ or “Saint Vincent and the Grenadines”/ or Belize/ or Costa
Rica/ or El Salvador/ or Guatemala/ or Honduras/ or Nicaragua/ or
determination of commonly observed clinical signs provid-
Panama/ or Mexico/ or Argentina/ or Bolivia/ or Brazil/ or Chile/ or
ed by IMCI guidelines could be key in increasing diagnosis Colombia/ or Ecuador/ or Guyana/ or Paraguay/ or Peru/ or Suri-
of neonatal sepsis and improving health outcomes (10). name/ or Uruguay/ or Venezuela/ or Antarctic Regions/ or Arctic
Regions/ or Kazakhstan/ or Kyrgyzstan/ or Turkmenistan/ or Uz-
Emerging antibiotic resistance is also important to consid- bekistan/ or Borneo/ or Cambodia/ or East Timor/ or Indonesia/ or
Laos/ or Malaysia/ or Mekong Valley/ or Myanmar/ or Philippines/
er in relation to CANS. A recent review in this area con-
or Thailand/ or Vietnam/ or Bangladesh/ or Bhutan/ or India/ or Sik-
cluded that data concerning antibiotic resistance in CANS kim/ or Afghanistan/ or Iran/ or Iraq/ or Jordan/ or Lebanon/ or Syr-
are very limited, but nevertheless highlighted potential ia/ or Turkey/ or Yemen/ or Nepal/ or Pakistan/ or Sri Lanka/ or Chi-
na/ or Hong Kong/ or Macau/ or Tibet/ or Korea/ or “Democratic
cause for concern resulting from studies showing emerging
People’s Republic of Korea”/ or Mongolia/ or Taiwan/ or Albania/ or
resistance in Klebsiella species and E. coli although levels of Lithuania/ or Bosnia-Herzegovina/ or Bulgaria/ or “Republic of Be-
larus”/ or “Macedonia (republic)”/ or Moldova/ or Montenegro/ or
resistance were noted to be lower than in hospital settings
Russia/ or Bashkiria/ or Dagestan/ or Moscow/ or Siberia/ or Serbia/
(17). Emerging antibiotic resistance is a major internation- or Ukraine/ or Yugoslavia/ or Armenia/ or Azerbaijan/ or “Georgia
al concern (18) and the information about the aetiological (republic)”/ or Melanesia/ or Fiji/ or Papua New Guinea/ or Vanuatu/
or Micronesia/ or Palau/ or Polynesia/ or Samoa/ or “Independent
spectrum of CANS and the prevalence of antibiotic resis-
State of Samoa”/ or Tonga/ or Comoros/ or Madagascar/ or Mauri-
tance among major causal pathogens are important to build tius/ or Seychelles/ or Solomon Islands.mp. or Marshall Islands.mp.
a broader understanding of this important public health or (Sao Tome and Principe).mp. or Maldives.mp. or Tuvalu.mp. or
(West Bank and Gaza).mp. or American Samoa/ or Romania/
issue.
2. exp Infant, Newborn/ or (newborn* or neonat*).tw.
3. exp sepsis/ or exp infection/ or (infection* or pathogen* or or-
Aims of this study
ganism* or bacter* or etiology).tw.
4. Limit 3 to “etiology (sensitivity)” [Limit not valid in Embase; re-
The aims of this study were to provide information on the
cords were retained]
bacterial aetiology of CANS in developing countries and to
5. (neonat* adj3 sepsis).tw.
discuss the implications of the information generated for 6. 2 and 4
future research and international child health policy in this 7. 5 or 6
field. The specific objectives of this systematic literature re- 8. 1 and 7
156 December, 2011 • Vol. 1 No. 2 / journal of global health • www.jogh.org
Aetiology of community-acquired neonatal sepsis in developing countries
er databases were slightly modified where necessary, to fit days. For several studies data were not reported in the cat-
the search terms offered in the respective databases. Final egories described above and so it was necessary to redis-
searches on all databases were undertaken on 30 January tribute so as to standardise for analysis. Some studies re-
2011. Searches were supplemented by screening reference ported overlapping aetiological data for CSF and blood
lists of selected papers and including literature discovered isolates (i.e. more than one isolate for an individual pa-
that corresponded with inclusion criteria. tient), in these studies it was decided to only extract data
on blood isolates to avoid distortion of the results. S
T
Inclusion and exclusion criteria N
It was decided through the study of primary data descrip- I
SO
Although it started with no time limits, the review was tions and previous reviews not to extract data for certain rP
EW
eventually restricted to literature published after 1980, to organisms, including Myma polymorpha and Micrococci sp. PE
a
limit the number of studies and to present current aetio- as they were deemed to be contaminants. Although previ- PVI
logical data. No restrictions were used concerning publica- ous studies have excluded altogether data specifying infec-
tion type or language of publication. All life-threatening tion from coagulase-negative Staphylococci, which are
invasive bacterial infections (bacteraemia, pneumonia and known to be common opportunistic pathogens in hospital-
meningitis) affecting infants of 0–90 days of age were in- acquired infections, but are seen as likely to be contami-
cluded.Studies reporting viral or fungal infection, nosoco- nants in CANS (4,47), it was decided to extract data for
mial infection, congenital infection or other infections such these organisms but exclude them from summary tables.
as ophthalmia neonatorum, malaria, tetanus or tuberculosis
Data analysis
were excluded. All studies reporting CANS were included,
along with studies where the setting or infection type re- For the purpose of analysis, data tables from studies were
ported suggested CANS. A certain number of studies were separated into 6 WHO global regions as illustrated below
found where it was thought that CANS was indicated, how- (Figure 1). Summary tables using the aetiological catego-
ever. However, the study data were deemed inconclusive ries above were assembled for each region and relative per-
to justify this assumption with a sufficient degree of cer- centages of each organism calculated (Supplementary Ta-
tainty. These studies were included for data extraction, but ble 4). Tables for each region were then compiled, detailing
not for the final data analysis. Isolated organisms from both only potentially pathogenic organisms so as to gain clearer
blood and CSF were included. insight into aetiology. Decisions on pathogenicity were
based on those of previous reviews and also the informa-
Studies reporting less than 50 cases were excluded, to in-
tion from other published literature (2,4,48). The ‘other/
crease the potential to generalise from results and prevent
unspecified’ category for Gram-positives and Gram-nega-
large deviations in suggested prevalence due to small sam-
tives was removed from potential pathogen tables along
ple sizes and chance effects. Studies reporting the inci-
with the Non-stated/Undetermined category.
dence/prevalence of only one organism were also excluded,
for the same reason of likely over-estimation and the effects Regional relative percentages for potential pathogens were
of chance. Review articles were also excluded, because pri- calculated along with 95% confidence intervals and these
mary data were the focus of the review. These were, how- were combined through meta-analysis to counteract issues
ever, used as helpful sources of reference. with data bias and create an ‘All Regions’ category (Supple-
mentary Table 5). Either the fixed effect model (Mantel-
Data extraction Haenszel method) or in cases of heterogeneity the random
effect model (DerSimonian-Laird method) were used (49).
Data were extracted from all selected studies and compiled
Between-study heterogeneity was quantified by calculating
in Microsoft Excel spreadsheets. An overall table of study
the Q statistic with a p-value less or equal than 0.05 being
characteristics was formed (Supplementary Table 1) and
the threshold (49). The meta-analysis results were unstable
individual tables for each study were compiled, charting
for several pathogens due to the small quantity of data,
quantities of organisms isolated from blood-culture proven
therefore it was decided for the purpose of analysis to use
CANS (Supplementary Table 2) (20–46). This data was
median and inter-quartile (IQ) range data from regional
further split into ≤7 days of life, 7–59 days of life and 60–
percentages for all potential pathogens and meta-analytical
90 days of life based on the neonatal age at isolation of or-
data of regional percentages for the cumulative ‘Potentially
ganism, henceforth described as age-of-onset (Supplemen-
pathogenic Gram-positives/negatives’ columns.
tary Table 3). The first category corresponds to early-onset
sepsis as described in the introduction. The second catego- As 30% of studies reported aetiological data using just
ry corresponds to late-onset sepsis but was expanded to fit numbers of positive isolates for each organism as a denom-
the WHO definition of a ‘young infant’ (7–59 days). The inator rather than the number of patients with positive iso-
third category includes any data after this period up to 90 lates for each organism, some studies reported more iso-
www.jogh.org • journal of global health / December, 2011 • Vol. 1 No. 2 157
Donald Waters et al.
European Eastern Mediterranean
S
T
N
I
OS
Pr
WE
EP
a
I
VP
Americas
Africa
South-East Asia
Western Pacific
Figure 1 WHO Regions (adapted from Wikimedia Commons;
http://commons.wikimedia.org/wiki/File:World_Health_Organisation_regional_offices.PNG).
lates than the number of patients in their sample. Therefore To avoid potential compromising of final results it was de-
it was deemed necessary to split aetiological data into that cided to extract data from these 7 papers however exclude
based on patients and that based on isolates so as to deter- them from overall analyses. The literature search process is
mine and analyse any differences. Other data on incidence, outlined in Figure 2.
case fatality rates, risk factors for CANS and antimicrobial
susceptibility patterns were extracted where available and

analysed. Studies identified through
search of Medline, Embase, Additional studies identi fied
Global Health, Web of Knowle dge through refere nce list
Quality control (n=16789) (n=3)
To ensure quality control, another reviewer undertook an
independent second data extraction of a certain proportion
of this review’s selected studies, totaling 520 data points;
Titles and abstracts Records excluded
100% of these were the same and therefore it was conclud- screened (n=16686 )
(n=16792)
ed that the standard of data reliability was likely to be high.
Full-text articles
Full-text articles excluded in accordance
rESULTS assessed for eligibility with inclusion/
(n=106) exclusion criteria
(n=72)
The literature search returned 16 789 studies whose titles
and abstracts were reviewed for relevance. 103 were se-
Data extracted (n=34)
lected for full text examination, however only 100 papers
were sourced in full-text versions. Of these 31 were select- Data inconclusive Studies included in
(extracted and presented qualitative synthesis
ed for inclusion in the review. In addition, 3 studies were in appendices) (n=7) (n=27)
found from other studies’ reference lists and selected for
inclusion, resulting in a total of 34 studies included in the
Studies included in
review. Of these, 27 studies were deemed to be reporting
quantitative synthesis
data concerning CANS and 7 were considered to be less (meta-analysis )
(n=27)
conclusive, possibly reporting neonatal sepsis acquired
from another source (Supplementary Table 6) (50–56). Figure 2 An overview of literature search results.
158 December, 2011 • Vol. 1 No. 2 / journal of global health • www.jogh.org
Aetiology of community-acquired neonatal sepsis in developing countries
Table 2 Study characteristics to fit with the WHO ‘young infant’ criteria. Full tables in-
cluding this data are presented in Supplementary Table 4.
No.
Characteristic
studies
Figures 3 and 4 present the meta-analysis forest plot
Africa 11
Americas 1 graphs for the potentially pathogenic Gram-positive and
Europe 2
WHO Region Gram-negative categories respectively. Forest plot graphs
Eastern Mediterranean 1
South-East Asia 7 for all other potential pathogens are presented in Supple-
Western Pacific 5 S
mentary Table 5. T
< 1 Year 3 N
1–2 Years 9 I
Length of study 3–4 Years 8 The percentages from Table 4 on the six most commonly SO
≥ 5 Years 5 isolated organisms along with Group B Streptococci (GBS) ErWP
Not reported 2
1980–1985 2 are then illustrated in Figure 5. Group B Streptococci was aPE
I
1986–1990 5 included despite its relatively low prevalence so as to pro- PV
1991–1995 13
Number of studies active 1996–2000 11 vide comparison with known high colonization rates ex-
in particular time periods
2001–2005 10 perienced in many developed countries (57). Figure 6 dis-
2006–2010 2
Not reported 2
Blood 16
CSF 5
Culture category
Blood and/or CSF* 5
Urine/Other 2
Isolates 8
Study denominator
Patients 19
0–25 8
26–50 10
51–75 5
Number of positive isolates
76–100 1
101–200 1
>200 2
0–25 11
26–50 10
Number of potentially 51–75 3
pathogenic positive isolates 76–100 1
101–200 1
>200 1
CSF – cerebrospinal fluid
*One study also used antigen detection.
Figure 3 Forest plot of the summary estimate and 95% confidence inter-
val of the prevalence of potentially pathogenic Gram-positives. Analysis
Characteristics of the studies that were retained after the
is based on 27 studies in the 5 WHO regions (Africa: 11 studies, Ameri-
process of literature search as they met the minimum qual- cas: 1 study, Europe: 2 studies, Eastern Mediterranean: 1 study, South East
ity criteria are shown in Table 2. A full version of this table Asia: 7 studies and Western Pacific: 5 studies). Weights are from random
effects analysis. ES: estimate, 95% CI: 95% confidence interval; I-squared
can be found in Supplementary Table 1. Of the 27 stud- and p-value are measures for heterogeneity between the studies.
ies, only 2 presented community surveillance data. Anoth-
er 20 either presented CANS-specific or disaggregated non-
nosocomial data and the remaining 5 did not explicitly
report CANS aetiology but were deemed suitable for inclu-
sion due to the infection type or study setting reported.
Four studies were the primary data sources for a WHO
Young Infants Study Group Multicenter Study (47), which
presented overall data. However, in our study we treated
each site as an independent data point, and we analysed
the information from each study individually.
Aetiological data
Individual aetiological data from each study are presented in
Supplementary Table 2. For the purpose of analyses, ae-
tiological data were split by WHO regions, and then by age-
of-onset. Table 3 is a summary table that contains data for Figure 4 Forest plot of the summary estimate and 95% confidence inter-
val of the prevalence of potentially pathogenic Gram-negatives. Analysis
all isolated organisms by region. Table 4 contains data for
is based on 27 studies in the 5 WHO regions (Africa: 11 studies, Ameri-
all potential pathogens isolated by region, both for age-of- cas: 1 study, Europe: 2 studies, Eastern Mediterranean: 1 study, South East
Asia: 7 studies and Western Pacific: 5 studies). Weights are from random
onset categories ≤7 days of life and 8–59 days. Data for the
effects analysis. ES: Estimate, 95% CI: 95% confidence interval; I-squared
60–90 days of life category were excluded from these tables and p-value are measures for heterogeneity between the studies.
www.jogh.org • journal of global health / December, 2011 • Vol. 1 No. 2 159
Donald Waters et al.
EWPOINTSaPErS All RegionsN% (95% CI)MedianIQ range30114.6 (13.1–16.2)12.312.6542.6 (2.0–3.4)0.51.1502.4 (1.8–3.2)0.62.4170.8 (0.5–1.3)0.43.71065.1 (4.3–6.2)3.34.5743.6 (2.9–4.5)1.93.71125.4 (4.5–6.5)0.10.471434.6 (32.536.6)33.9–13.925412.3 (10.9–13.8)8.913.4251.2 (0.8–1.8)01.936617.7 (16.1–19.4)1011.71889.1 (7.9–10.4)6.45.6693.3 (2.6–4.2)1.23.3401.9 (1.4–2.6)00.7130.6 (0.4–1.1)0.10.8271.3 (0.9–1.9)0.31.3321.5 (1.1–2.2)1.43.6130.6 (0.4–1.1)01.41326.4 (5.4–7.5)4.53.61014.9 (4.0–5.9)3.78.5126061.0 (58.963.1)62.8–15.3924.5 (3.6–5.4)4.211.12066100.0 (n/a)   ern Mediterranean – 0 isolates, Europe – 9 iso- All Regions Median %Meta-analysis % P-value for (IQ range)(95% CI)heterogeneity14.9 (14.2)17.4 (13.9–20.9)0.036 0.6 (1.2)3.4 (0.0–7.2)<0.0005 0.6 (3)3.2 (0.0–7.0)<0.0005 0.5 (5)1.1 (0.0–2.5)0.007 7 (12.6)8.3 (0.8–15.9)<0.0005 2.3 (4.2)4.1 (0.9–7.3)<0.0005 31 (22.4)34.8 (20.7–49.0)<0.0005 11.6 (11.3)18.6 (9.9–27.3)<0.000512.2 (12)19.1 (11.1–27.1)<0.00057.9 (8.5)9 (3.5–14.5)<0.00051.4 (3.8)2.9 (0.2–5.6)<0.0005N/a (only 2 –0 (0.7)studies)0.2 (0.8)0.5 (0.1–0.8)0.070.4 (1.6)1.9 (0.0–3.9)0.0061.9 (4.3)2.3 (0.0–5.3)<0.0005N/a (only 2 –0 (1.7)studies)5.5 (4.9)7 (4.0–10.0)0 69 (22.4)65.2 (51.1–79.3)<0.0005    
VIP Western PacificN% (95% CI)16815.2 (13.2–17.5)121.1 (0.6–1.9)30.3 (0.1–0.8)20.2 (0.0–0.7)141.3 (0.8–2.1)444.0 (3.0–5.3)11010.0 (8.3–11.9)35332.0 (29.334.8)–14313.0 (11.1–15.1)00.0 (–)24322.1 (19.7–24.6)14012.7 (10.9–14.8)565.1 (4.0–6.5)393.5 (2.6–4.8)30.3 (0.1–0.8)70.6 (0.3–1.3)20.2 (0.0–0.7)00.0 (–) 958.6 (7.1–10.4)201.8 (1.2–2.8)74867.9 (65.170.6)–10.1 (0.0–0.5)1102100.0 (n/a) Americas – 0 isolates, East N% (95% CI) 30117.1 ( 15.4–18.9) 543.1 (2.4–4.0) 502.8 (2.1–3.7) 171.0 (0.6–1.5) 1066.0 (5.0–7.2) 744.2 (3.4–5.2) 60234.2 (32.036.4)– 27915.8 (14.2–17.6)36620.8 (19.0–22.7)18810.7 (9.3–12.2)693.9 (3.1–4.9) 402.3 (1.7–3.1) 130.7 (0.4–1.3)271.5 (1.1–2.2)321.8 (1.3–2.6) 130.7 (0.4–1.3) 1327.5 (6.4–8.8) 115965.8 (63.6) 1761100.0 (n/a)
South-East AsiaN% (95% CI)2010.0 (6.6–14.9)21.0 (0.3–3.6)63.0 (1.4–6.4)00.0 (–)73.5 (1.7–7.0)10.5 (0.1–2.8)10.5 (0.1–2.8)3718.5 (13.724.5)–6733.5 (27.3–40.3)52.5 (1.1–5.7)189.0 (5.8–13.8)189.0 (5.8–13.8)31.5 (0.5–4.3)00.0 (–)00.0 (–)31.5 (0.5–4.3)10.5 (0.1–2.8)00.0 (–)94.5 (2.4–8.3)115.5 (3.1–9.6)13567.5 (60.773.6)–2814.0 (9.9–19.5)200100.0 (n/a) : Africa – 28 isolates, occi Western Pacific N% (95% CI) 16817.3 (15.1–19.8) 121.2 (0.7–2.1) 30.3 (0.1–0.9) 20.2 (0.1–0.7) 141.4 (0.9–2.4) 444.5 (3.4–6.0) 24325.0 (22.427.8)– 14314.7 (12.6–17.1)24325.0 (22.4–27.8)14014.4 (12.3–16.8)565.8 (4.5–7.4) 394.0 (3.0–5.4) 30.3 (0.1–0.9)70.7 (0.3–1.5)20.2 (0.1–0.7) 00.0 (–) 959.8 (8.1–11.8) 72875.0 (72.277.6)– 971100.0 (n/a)
Table 3 Distribution of all microorganisms isolated by region AfricaAmericasEastern MediterraneanEuropeMicroorganism isolatedN% (95%CI)N% (95%CI)N% (95%CI)N% (95% CI)8514.6 (11.7–17.4)00.0 (–)00.0 (–)2826.4 (19.0–35.6)Staphylococcus aureusGroup A 406.9 (4.8–8.9)00.0 (–)00.0 (–)00.0 (–)Streptococci/ Streptococcus pyogenesGroup B 406.9 (4.8–8.9)00.0 (–)00.0 (–)10.9 (0.2–5.2)StreptococciGroup D 40.7 (0.0–1.4)00.0 (–)24.8 (1.3–15.8)98.5 (4.5–15.4)Streptococci/ Enterococcus8113.9 (11.1–16.7)13.0 (0.5–15.3)37.1 (2.5–19.0)00.0 (–) Streptococcus pneumoniaeOther/unspecified species193.3 (2.1–5.0)00.0 (–)1023.8 (13.5–38.5)00.0 (–)Streptococcus Other/ unspecified Gram-positives*10.2 (0.0–1.0)00.0 (–)00.0 (–)00.0 (–) Total Gram-positives27046.3 (42.350.4)131535.7 (23.050.8)3835.8 (27.445.3)–––223.8 (2.5–5.6) 00.0 (–)24.8 (1.3–15.8)2018.9 (12.6–27.4)Klebsiella pneumoniaeOther/unspecifiedspecies203.4 (2.2–5.2)00.0 (–)00.0 (–)00.0 (–) Klebsiella 6411.0 (8.7–13.8)00.0 (–)37.1 (2.5–9.0)3835.8 (27.4–45.3)Escherichia colispecies223.8 (2.5–5.6)00.0 (–) 49.5 (3.8–22.1)43.8 (1.5–9.3)Pseudomonas species50.9 (0.4–2.0)00.0 (–)49.5 (3.8–22.1)10.9 (0.2–5.2)Enterobacter species00.0 (–)00.0 (–)00.0 (–)10.9 (0.2–5.2)Serratia species91.5 (0.8–3.0)00.0 (–)00.0 (–)10.9 (0.2–5.2)Proteus species172.9 (1.8–4.6)00.0 (–)00.0 (–)00.0 (–)Salmonella 254.3 (2.9–6.3)39.1 (3.1–23.6)12.4 (0.4–12.3)00.0 (–)Haemophilus influenzae111.9 (1.1–3.3)26.1 (1.7–19.6)00.0 (–)00.0 (–)Neisseria meningitidisspecies264.5 (3.1–6.5)00.0 (–)24.8 (1.3–15.8)00.0 (–)Acinetobacter Other/unspecified Gram-negatives**5910.1 (7.9–12.8)00.0 (–)1126.2 (15.3–41.1)00.0 (–)Total Gram-negatives28048.0 (44.052.1)515.2 (6.731.0)2764.3 (49.1––77.0)6561.3 (51.870.0)––Non-stated/Undetermined335.7 (4.1–7.8)2781.8 (65.6–91.4)00.0 (–)32.8 (1.0–8.0)Total583100.0 (n/a)33100.0 (n/a)42100.0 (n/a)106100.0 (n/a)95% CI – 95% confidence interval, IQ range – interquartile range* Includes data sp., sp and others.for Aerococcus Bacillus .**Includes data for sp., sp., sp., sp. and others. Data was also extracted for coagulase-negative Citrobacter Moraxella Shigella Aeromonas Staphyloclates, South-East Asia – 35 isolates, Western Pacific – 811 isolates. Table 4 Distribution of potential pathogens isolated by region Eastern EuropeSouth-East AsiaAfricaAmericasMediterraneanPotential pathogen N% (95% CI)N% (95% CI)N% (95% CI)N% (95% CI)N% (95% CI)isolated8517.3 (14.3–21.0)00.0 (–)00.0 (–)2827.2 (19.5–36.5) 2012.5 (8.2–18.5)Staphylococcus aureusGroup A Streptococci/ 408.2 (6.1–10.9)00.0 (–)00.0 (–)00.0 (–)21.3 (0.3–4.4)Streptococcus pyogenesGroup B 408.2 (6.1–10.9)00.0 (–)00.0 (–)11.0 (0.2–5.3)63.8 (1.7–7.9)StreptococciGroup D Streptococci/ 40.8 (0.3–2.1)00.0 (–)26.5 (1.8–20.7)98.7 (4.7–15.8)00.0 (–)Enterococcus8116.5 (13.5–20.1)116.7 (3.0–56.4)39.7 (3.3–24.9)00.0 (–) 74.4 (2.1–8.8)Streptococcus pneumoniaeOther/unspecified 193.9 (2.5–6.0)00.0 (–)1032.3 (18.6–49.9)00.0 (–)10.6 (0.1–3.5) speciesStreptococcusPotentially pathogenic 26954.9 (50.559.2)116.7 (3.056.4)1548.4 (32.065.2)3836.9 (28.246.5)3622.5 (16.729.6)–––––Gram-positivesspecies428.6 (6.4–11.4)00.0 (–)26.5 (1.8–20.7)2019.4 (12.9–28.1)7245.0 (37.5–52.7)Klebsiella 6413.1 (10.4–16.3)00.0 (–)39.7 (3.3–24.9)3836.9 (28.2–46.5)1811.3 (7.2–17.1)Escherichia coli species224.5 (3.0–6.7)00.0 (–)412.9 (5.1–28.9)43.9 (1.5–9.6)1811.3 (7.2–17.1)Pseudomonasspecies51.0 (0.4–2.4)00.0 (–)412.9 (5.1–28.9)11.0 (0.2–5.3)31.9 (0.6–5.4)Enterobacter species00.0 (–)00.0 (–)00.0 (–)11.0 (0.2–5.3)00.0 (–)Serratia species91.8 (1.0–3.5)00.0 (–)00.0 (–)11.0 (0.2–5.3)00.0 (–)Proteusspecies173.5 (2.2–5.5)00.0 (–)00.0 (–)00.0 (–)31.9 (0.6–5.4)Salmonella 255.1 (3.5–7.4)350.0 (18.8–81.2)13.2 (0.6–16.2)00.0 (–)10.6 (0.1–3.5)Haemophilus Influenzae 112.2 (1.3–4.0)233.3 (9.7–70.0)00.0 (–)00.0 (–)00.0 (–)Neisseria Meningitidis species265.3 (3.7–7.7)00.0 (–)26.5 (1.8–20.7)00.0 (–)95.6 (3.0–10.3)AcinetobacterPotentially pathogenic 22145.1(40.849.6)583.3 (43.497.0)1651.66563.112477.5 (70.483.3)–––Gram-negativesTotal490100.0 (n/a)6100.0 (n/a)31100103100.0 (n/a)160100.0 (n/a) 95% CI – 95% confidence interval, IQ range – interquartile range
160 December, 2011 • Vol. 1 No. 2 / journal of global health • www.jogh.org
Aetiology of community-acquired neonatal sepsis in developing countries
y 90.0 However, all studies (from 14 countries, and in all regions)
or neit reported data for the category of neonates and young in-
P-value fheteroge0.036 <0.0005 <0.0005 0.007 <0.0005 <0.0005 <0.0005 <0.0005<0.0005<0.0005<0.0005 – 0.070.006<0.0005 – 0 <0.0005   8700..00 fdaanttas swpehcoifi weder aes 8 s–e5p9si sd oayccs uorlrdin. gIn a ta d8d–5it9io dna tyos aoeft liiofelo, gdiactaal
All Regions Median %Meta-analysis % (IQ range)(95% CI)14.9 (14.2)17.4 (13.9–20.9) 0.6 (1.2)3.4 (0.0–7.2) 0.6 (3)3.2 (0.0–7.0) 0.5 (5)1.1 (0.0–2.5) 7 (12.6)8.3 (0.8–15.9) 2.3 (4.2)4.1 (0.9–7.3) 31 (22.4)34.8 (20.7–49.0) 11.6 (11.3)18.6 (9.9–27.3)12.2 (12)19.1 (11.1–27.1)7.9 (8.5)9 (3.5–14.5)1.4 (3.8)2.9 (0.2–5.6)N/a (only 2 0 (0.7)studies)0.2 (0.8)0.5 (0.1–0.8)0.4 (1.6)1.9 (0.0–3.9)1.9 (4.3)2.3 (0.0–5.3)N/a (only 2 0 (1.7)studies)5.5 (4.9)7 (4.0–10.0) 69 (22.4)65.2 (51.1–79.3)    6543200000.....00000 cdblW6ayas0ea, teiy–sfiensse9gtvg,peo 0e 7rfsr rndi–siosst a5 emuPoy9ddasc ‘c cidnoaeiuasfifes ryo ac<rfsgn)ri2, noe ar 7.gmtem e–I pnas 5oofi/t 5nan rv6dt etedeh0wd da–sciby ,to9 idsoo1u0,arn –n tn7da2 tt–sar o ’fi2 my oewa8ssreo e otdinrtinfheoat helylatio sswflcesg ,a ao,oi8 cnt dei–ardngale6 ctogd‘0allriaua oycttddnea aaoe- stsyodef p s(ng.iA,e nos Fc8ferfiiraitn–fiis’nc 3aeeoatldd0sr-, aPErSEWPOINTS
% (95% CI) 17.1 ( 15.4–18.9) 3.1 (2.4–4.0) 2.8 (2.1–3.7) 1.0 (0.6–1.5) 6.0 (5.0–7.2) 4.2 (3.4–5.2) 34.2 (32.036.4)– 15.8 (14.2–17.6)20.8 (19.0–22.7)10.7 (9.3–12.2)3.9 (3.1–4.9) 2.3 (1.7–3.1) 0.7 (0.4–1.3)1.5 (1.1–2.2)1.8 (1.3–2.6) 0.7 (0.4–1.3) 7.5 (6.4–8.8) 65.8 (63.6) 100.0 (n/a) 100..00 Sa1s–t 32u –dm3y om ndothensnt hwose,m r3e1 ian–l9sao0t iodnracylus,d 3ed0–.90 days, 30–59 days and PVI
N 01 54 50 17 06 74 02 79668869 40 132732 13 32 59 61
3 1 6 231 1 11 17 Figure 5 Percentage of selected Nineteen studies used patients as a denominator, while 8
Western Pacific % (95% CI) 17.3 (15.1–19.8) 1.2 (0.7–2.1) 0.3 (0.1–0.9) 0.2 (0.1–0.7) 1.4 (0.9–2.4) 4.5 (3.4–6.0) 25.0 (22.427.8)– 14.7 (12.6–17.1)25.0 (22.4–27.8)14.4 (12.3–16.8)5.8 (4.5–7.4) 4.0 (3.0–5.4) 0.3 (0.1–0.9)0.7 (0.3–1.5)0.2 (0.1–0.7) 0.0 (–) 9.8 (8.1–11.8) 75.0 (72.277.6)– 100.0 (n/a) pripeneaoogdcttiheieocnn narttetsiiegaa. lliNt ohppnueaa.m ttthhobooteaggrlee snnn iussn mii nspob aWleraertsHne dtoOh ff e osre s uoinnsgessde t ai ncsoadtl aedtgieosspr.i leDasya eatdan doin nr e Fpgioigotuennrse twisa el7 rp eaa nstphdlo i8tg ,ei nnrtesos f prtheocmotsi eva el2ll y ag.groe-uopf--
N 8 2 3 2 4 4 3 3306 9 372 0 5 8 1 Other reported information
6 1 1 4 4 4445 3 9 2 7
1 2 121 7 9
outh-East Asia % (95% CI) 12.5 (8.2–18.5) 1.3 (0.3–4.4) 3.8 (1.7–7.9) 0.0 (–) 4.4 (2.1–8.8) 0.6 (0.1–3.5) 22.5 (16.729.6)– 45.0 (37.5–52.7)11.3 (7.2–17.1)11.3 (7.2–17.1)1.9 (0.6–5.4) 0.0 (–) 0.0 (–)1.9 (0.6–5.4)0.6 (0.1–3.5) 0.0 (–) 5.6 (3.0–10.3) 77.5 (70.483.3)– 100.0 (n/a) Toicniend n TeC ansAtbcuNelde dS i6ea-sst awp f errhoecipmicfiohc r8 ftc uecadrost eubh nyefar t3tmrai seloitstur yeid n ric aefotsoen.u sWt,ra airinntehsgd iCr otehAngesaNs rredSe - aptsorope rte htdceeeidfi tc acdoi lamientda--
S N 0 2 6 0 7 1 6 2883 0 031 0 9 4 0 monly reported risk factors for neonatal sepsis among pre-
2 3 711 2 6
1 1
mature or low birth weight neonates, only a single source
Europe % (95% CI) 27.2 (19.5–36.5) 0.0 (–) 1.0 (0.2–5.3) 8.7 (4.7–15.8) 0.0 (–) 0.0 (–) 36.9 (28.246.5)– 19.4 (12.9–28.1)36.9 (28.2–46.5)3.9 (1.5–9.6)1.0 (0.2–5.3) 1.0 (0.2–5.3) 1.0 (0.2–5.3)0.0 (–)0.0 (–) 0.0 (–) 0.0 (–) 63.1 100.0 (n/a) orfwaefnop drtosakr tetianer sdw l qoa4uws6 oi %drteee dsinno t2ucifii2rdec%eden sfocoefetr tC oienAaf gcNCsh AS.( 4NGin3aSc)t ,ici dnhwe ahlnloiicawleen o Mbacinrocdtunh rdc iwoanl-ge wai ginonhdr tkp ceirnoers---
N 8 0 1 9 0 0 8 0841 1 100 0 0 5 3
2 3 23 6 10 term neonates (38). Culture positivity rates varied
n Eastern Mediterranean % (95% CI) 0.0 (–) 0.0 (–) 0.0 (–) 6.5 (1.8–20.7) 9.7 (3.3–24.9) 32.3 (18.6–49.9) 48.4 (32.065.2)– 6.5 (1.8–20.7)9.7 (3.3–24.9)12.9 (5.1–28.9)12.9 (5.1–28.9) 0.0 (–) 0.0 (–)0.0 (–)3.2 (0.6–16.2) 0.0 (–) 6.5 (1.8–20.7) 51.6 100 Fpiogtuernet i6a lD pisattrhiobguetinosn b oyf rGegraiomn-.positive and Gram-negative saTinishg clnilosui wficsoic oaaunsnl d t5cl ysr%,ui tw ge(rg2iitea6hs,) t ,r c seoaipgsreon e ridvfiteeecdnfia nnr3tai%t tdieo isffn foae rosr e rshn ocicmgaehpse a aaacmsgi te6yo 5 ngf%rogo rs u (ta4puc1sdc )iu(e 2arsa4n it)nde.
gio N 0 0 0 2 3 10 15 2344 0 001 0 2 16 31 microbiological analysis, and therefore potentially data
e
otential pathogens isolated by r AfricaAmericas % (95% CI)N% (95% CI) 17.3 (14.3–21.0)00.0 (–) 8.2 (6.1–10.9)00.0 (–) 8.2 (6.1–10.9)00.0 (–) 0.8 (0.3–2.1)00.0 (–) 16.5 (13.5–20.1)116.7 (3.0–56.4) 3.9 (2.5–6.0)00.0 (–) 54.9 (50.559.2)116.7 (3.056.4)–– 8.6 (6.4–11.4)00.0 (–)13.1 (10.4–16.3)00.0 (–)4.5 (3.0–6.7)00.0 (–)1.0 (0.4–2.4)00.0 (–) 0.0 (–)00.0 (–) 1.8 (1.0–3.5)00.0 (–)3.5 (2.2–5.5)00.0 (–)5.1 (3.5–7.4)350.0 (18.8–81.2) 2.2 (1.3–4.0)233.3 (9.7–70.0) 5.3 (3.7–7.7)00.0 (–) 45.1(40.849.6)583.3 (43.497.0)–– 100.0 (n/a)6100.0 (n/a) val, IQ range – interquartile range pgeTcsspeaarlanaactebemyhctlgesi efird-o ep5dcr giio anspaistoeg raidSntee is- u.fvbeoreponfy -pmta osdlnn e didTmfsafe aetgtearb rne doalnteamfa tt a ar4-la ynl fg ooeoeTgrrf-g a aoptabtfhoin-lveoteiees nm nr3pset.esiol ata Eit ltseic iogpnvalhatetaeitt tah gepsldoo trpugro iiaendpetn isohaes.rl os ltRa girfroeeereogngn impissoo rnn ioen6s--f DTqscptrpeuahinIeetaeShsctl a oiioCatfigvygrce.eUee srF na vaSitnsleno ltSra a idysrmlIaele Oyltl ii,edacm 6c rNiosot inesptbetd luCida adya,Al neiwsNedtesii Snmt irh nseii ni pcotT irolnvaoorlibbwytteyli ea2d apl 7n s7da d.fsato tttmeuarr dicntdhiosede.ns l ecSms eeuirnonintcssaitoin btmpiglver eieC t fvycoAa orpNlu eiennnSr----t
Table 4 Distribution of p Potential pathogen Nisolated85Staphylococcus aureusGroup A Streptococci/ 40Streptococcus pyogenesGroup B 40StreptococciGroup D Streptococci/ 4Enterococcus81Streptococcus pneumoniaeOther/unspecified 19 speciesStreptococcusPotentially pathogenic 269Gram-positivesspecies42Klebsiella 64Escherichia coli species22Pseudomonasspecies5Enterobacter species0Serratia species9Proteusspecies17Salmonella 25Haemophilus Influenzae 11Neisseria Meningitidis species26AcinetobacterPotentially pathogenic 221Gram-negativesTotal490 95% CI – 95% confidence inter cWwwodclocauhaecuyydosusnst e,rew t adrrr0inei nn–eirdn gse7P i atuaindhnctp aiia ≤sfi 4ydt 7occs d ),r ad7i etrt0a egiedoy–gipaosn5ooy n or trsdysoft . oe a l(adilyfAdee sdt.f ,i r aaDodintclaaaodat tgf,aa o i‘ Eecrwreaa uteprlhr lrdooyeera-p cptoteeaarnd, te ses sSpaegeoseton’ uc 4rtwiyte–fih de6oe- rd fEide n na aa aevssylto asss rn,oAe y0pa siit–nsineai6gss-, cvonoi2elifv8isud fiel isseicrfntidaeaou l n cnadll tien,itvl reytdamtls ea lf 6iet iaddan0wnkia –tygeity9na rl oo 0.god bfT fa dp awhttallaha iyhr scefse ioe s cgor ruhiie fnotg vhl nhgiAiefeesaef w sdar.it isgT ’scs se aaham-rn oeeodoa sfrge-ult Sleohg loontsresegaurs.emr e tcaT htoo pp-chfnEhla eecctair aecesesug rt if nozwotAier,coas isuwnies ass a i a o rtno≤lhevsdf7g eos1 irdp to7 ustarni hdoygoses---.f
www.jogh.org • journal of global health / December, 2011 • Vol. 1 No. 2 161
Donald Waters et al.
Table 5 Distribution of all isolates by age-of-onset
≤7 days of life 8–59 days of life 60–90 days of life
Organism Isolated N % (95% CI) N % (95%CI) N % (95%CI)
Staphylococcus aureus 33 11.7 (8.0–15.5) 268 15.0 (13.4–16.7) 1 2.3 (0.0–6.7)
Group A Streptococci/ Streptococcus pyogenes 4 1.4 (0.0–2.8) 50 2.8 (2.0–3.6) 8 18.2 (6.8–29.6)
Group B Streptococci 19 6.7 (3.8–9.7) 31 1.7 (1.1–2.4) 0 0.0 (–)
Group D Streptococci/ Enterococcus 4 1.4 (0.0–2.8) 13 0.7 (0.3–1.1) 0 0.0 (–)
S Streptococcus Pneumoniae 13 4.6 (2.1–7.1) 93 5.2 (4.2–6.3) 14 31.8 (18.1–45.6)
T
N Other/unspecified Streptococcus species 24 8.5 (5.3–11.8) 50 2.8 (2.0–3.6) 0 0.0 (–)
I Other/ unspecified Gram-positives* 0 0.0 (–) 112 6.3 (5.2–7.4) 0 0.0 (–)
OS
Pr All Gram-positives 97 34.4 (28.9–39.9) 617 34.6 (32.4–36.8) 23 52.3 (37.5–67.0)
WE Klebsiella pneumoniae 22 7.8 (4.7–10.9) 232 13.0 (11.4–14.6) 2 4.5 (0.0–10.7)
EaP Other/unspecified Klebsiella species 10 3.5 (1.4–5.7) 15 0.8 (0.4–1.3) 0 0.0 (–)
I
VP Escherichia coli 46 16.3 (12.0–20.6) 320 17.9 (16.2–19.7) 1 2.3 (0.0–6.7)
Pseudomonas species 22 7.8 (4.7–10.9) 166 9.3 (8.0–10.7) 1 2.3 (0.0–6.7)
Enterobacter species 10 3.5 (1.4–5.7) 59 3.3 (2.5–4.1) 0 0.0 (–)
Serratia species 0 0.0 (–) 40 2.2 (1.6–2.9) 1 2.3 (0.0–6.7)
Proteus species 6 2.1 (0.4–3.8) 7 0.4 (0.1–0.7) 0 0.0 (–)
Salmonella species 1 0.4 (0.0–1.0) 26 1.5 (0.9–2.0) 6 13.6 (3.5–23.8)
Haemophilus influenzae 2 0.7 (0.0–1.7) 30 1.7 (1.1–2.3) 4 9.1 (0.6–17.6)
Neisseria meningitidis 0 0.0 (–) 13 0.7 (0.3–1.1) 0 0.0 (–)
Acinetobacter species 19 6.7 (3.8–8.7) 113 6.3 (5.2–7.5) 3 6.8 (0.0–14.3)
Other/unspecified Gram-negatives** 42 14.9 (10.7–19.0) 59 3.3 (2.5–4.1) 1 2.3 (0.0–6.7)
All Gram-negatives 180 63.8 (58.2–69.4) 1080 60.5 (58.3–62.8) 19 43.2 (28.5–57.8)
Non-stated/Undetermined 5 1.8 (0.2–3.3) 87 4.9 (3.9–5.9) 2 4.5 (0.0–10.7)
Totals 282 100.0 (n/a) 1784 100.0 (n/a) 44 100.0 (n/a)
*Includes data for Aerococcus sp., Bacillus sp. and others.
** Includes data for Citrobacter sp., Moraxella sp., Shigella sp., Aeromonas sp. and others. Data were also extracted for coagulase- negative Staphylococci:
≤7 days of life – 3 isolates, 7–59 days of life – 880 isolates, 60–90 days of life – 0 isolates.
Table 6 Reported case fatality rates and incidence data
Region Article Case fatality rate reported* Incidence data reported
Africa Berkley et al (22) 56% of ≤7 days of life, 26% of 8–59 days of life 1457/105 person years (for infants <1year old)
  Mulholland et al (28) 31% of 0–91 days of life –
  English et al (24) 27% of ≤7 days of life, 5% of 8–59 days of life –
  Muhe et al (27) 49% of 0–59 days of life –
  Campagne et al (23) 58% of 0–59 days of life –
Americas Weiss et al (30) – 421/105 person years (for infants <2months old)
South-East Asia Mondal et al (38) 32% of 0–59 days of life –
  Mathur et al (37) 70% of 0–59 days of life –
Europe Taskin et al (33) 5% of ≤7 days of life, 3% of 8–59 days of life –
Western Pacific Quiambao et al (45) 29% of 0–91 days of life –
Gatchalian et al (43) 26% of 0–91 days of life –
Choo et al (42) –  1571/105 live births
*Data were standardised to fit with age categories used in review.
Although the majority of studies being conducted in Africa highlighting the need for new research in this area and im-
can be argued to fit with Africa having the highest neonatal plying potential issues with the representativeness of data
mortality rates globally, it should be noted that seven out presented here. Excluding studies before 1980 to narrow
of ten African studies were based in Nigeria or Kenya, and the literature review may have resulted in missing some
there were little or no data for other countries with com- relevant studies. Although the search was systematic, some
parable neonatal mortality levels. Similarly, 6 out of 7 studies after 1980 may have also been missed due to the
potential of human error in screening results. In addition,
South-East Asian studies were conducted in India, which
several foreign language studies were excluded because it
again has high neonatal mortality rates. However, these are
was not possible to extract enough information to include
reportedly lower than in Pakistan, yet no studies were
them in the review. Several studies had very low culture
found from Pakistan (58).
positivity rates, resulting in small numbers of organisms
The largest numbers of studies found by this review were being reported. The potential to generalise results is lim-
conducted between 1991 and 1995 and only 2 studies ited by their small sample sizes. Some studies presented
were relevant to the most recent period of 2006–2010, considerably larger numbers of isolated organisms than
162 December, 2011 • Vol. 1 No. 2 / journal of global health • www.jogh.org
Aetiology of community-acquired neonatal sepsis in developing countries
others, thereby giving greater weight to their reported ae- ent age-of-onset categories, but they did not combine these
tiological data. two categorisations. Still, in all cases we used the informa-
tion from other studies to impute the data and align it with
To analyse aetiological data by the age-of-onset, it was nec-
the majority of studies, in order to prevent valuable infor-
essary to standardise data into specific categories; however,
these categories did not always fit with reported data. In mation from being lost.
certain cases it was therefore necessary to reassign and im-
Although several studies explicitly reported excluding in-
S
pute data. This was not the case for the significant major- fants who had received prior antimicrobial therapy, the ma- T
N
ity of studies. Despite some of the limitations described
above, data presented in this review should be generally jority of studies did not report inclusion or exclusion criteria SOI
relating to this factor. This is potentially an important issue rP
robust and useful for planning international child health EW
as antimicrobial therapy prior to blood cultures being taken PE
policy on tackling neonatal infections. aI
could significantly change aetiological patterns and therefore PV
Study design bias the reported data. Most studies reported data from ba-
bies who presented to primary facilities or outpatient ser-
Criteria for diagnosing neonatal sepsis varied significantly,
vices of referral facilities. The aetiological distribution may
providing inconsistencies in data and potential biases.
differ from that of babies who are born at home and die be-
Some studies included pneumonia and meningitis in this
fore reaching hospital. Only two studies adopted a commu-
category, while others considered them indistinguishable
nity-surveillance approach that could counter this issue.
from the data for sepsis. The criteria for excluding nosoco-
mial infections also differed, with several studies merely Culture positivity rates were reported to be considerably
separating data into babies born in the study hospital and low by some studies, potentially indicating inadequate lab-
babies born elsewhere (and other studies not defining cri- oratory facilities or high prior antimicrobial use. Future re-
teria at all). The age-of-onset categories also were not fully views could potentially exclude these studies, but this
consistent between studies, and there were some discrep- would further decrease the amount of data. It is likely that
ancies with definitions of ‘early-onset’ and ‘late-onset’ sep- the prevalence of certain pathogens such as Haemophilus in-
sis. Several studies did not collect data for all age-of-onset fluenzae is underestimated in many studies due to the sig-
categories, therefore not providing a complete picture of nificant diagnostic capacity required to isolate these organ-
CANS aetiology. Also, a certain number of studies split data isms (4). For the same reason, it is possible that prevalence
into community-acquired and nosocomial, and into differ- of less fastidious organisms such as S. aureus is overestimat-
Table 7 Percentage sensitivity patterns of most prevalent pathogens to selected antimicrobials
  Region Africa South-East Asia
Organism Antimicrobial Adejuyigbe et al (20) Muhe et al (27) Mathur et al (37) Panigrahi et al (39) Darmstadt et al (35) Tallur et al (41)
Escherichia Amoxycillin (AMX) 60.0 – – – – –
coli Ampicillin (AMP) 40.0 100.0 – – 100.0 29.0
Cefotaxime (CTX) – – – – – 100.0
Ceftazidime (CAZ) – 100.0 – – 100.0 –
Ceftriaxone (CRO) – – – – 100.0 100.0
Ciprofloxacin (CIP) – – – – 100.0 –
Gentamicin (GEN) 80.0 100.0 – – 100.0 71.0
Imipenem (IMP)         100.0 –
Staphylococcus Amoxycillin (AMX) 73.0 – – – – –
aureus  Ampicillin (AMP) – – – – 0.0 21.0
Cefotaxime (CTX) – – – – – –
Ceftazidime (CAZ) – – – – 66.7 –
Ceftriaxone (CRO) – – – – 90.0 –
Ciprofloxacin (CIP) – – – – 80.0 –
Gentamicin (GEN) 85.8 – – – 90.0 29.0
Imipenem (IMP) – – – – 90.0 –
Klebsiella Amoxycillin (AMX) 0.0 – – – – –
species* Ampicillin (AMP) – – 10.0 – 0.0 25.5
Cefotaxime (CTX) – – – – – 76.5
Ceftazidime (CAZ) – – – 22.0 33.3 –
Ceftriaxone (CRO) – – 71.4 – 33.3 81.0
Ciprofloxacin (CIP) – – 64.8 11.0 66.7 –
Gentamicin (GEN) 100.0 – 42.8 – 66.7 59.5
Imipenem (IMP) –   100.0 – 100.0 –
*Averages were taken when more than one variant’s sensitivity patterns were reported.
www.jogh.org • journal of global health / December, 2011 • Vol. 1 No. 2 163