4.1. (Pseudo)outbreaks and Cross-Sectional Surveys: Overview
Figure 1 summarizes the search and screening process. Among the excluded articles with reason, some originated from LMICs: one article from India was excluded because it was a purely experimental study [
37]; three articles from Tunisia reported about a single outbreak by contaminated eosin (a chemical dye) as a reservoir [
38,
39,
40]. In addition, two outbreak investigations found contaminated antiseptics but were excluded, as another fomite was identified as the source of the outbreaks [
41,
42] (
Supplementary Table S5). One cross-sectional survey was removed from the analysis because the name and product category of the investigated product were not mentioned [
43].
The final panel consisted of 38 articles, among which were 12 outbreaks, 1 pseudo-outbreak (together representing 34.2% of articles), and 25 (65.8%) cross-sectional surveys, versus 114 articles (71.1% (pseudo)outbreaks and 28.9% cross-sectional surveys) from HICs. Only a single article from LMICs reported a (pseudo)outbreak [
44] compared to 13 in HICs. As there was only a single (pseudo)outbreak, the text further groups this article together with the outbreak reports, unless otherwise stated. The earliest publication dated from 1977 [
45], exactly 20 years after the first publications from HICs [
46]. During the 1980s and 1990s, publications from LMICs were rare, but from 2000 onwards, numbers increased and also comprised liquid soap products, in line with observations from HICs; the 2:1 rate of cross-sectional surveys versus (pseudo)outbreak articles remained stable over time (
Figure 2) but was the opposite of what was observed in HICs [
9].
A total of 26 (68.4%) out of 38 articles originated from middle-income countries, with only 2 outbreaks reported from low-income countries [
47,
48]. Over half of the articles (20/38, 52.6%) originated from Asia, particularly from India, Malaysia, and Thailand (
Table 1). South America and the Caribbean accounted for 5 articles and Africa for 13 articles, 12 of which originated from sub-Saharan Africa, including 7 from Nigeria and only 1 from a French-speaking country. Central Africa and LMICs from Oceania were not represented [
49]. For sub-Saharan Africa, the poor representation reflected the lack of bacteriology testing services and poor IPC in healthcare facilities: according to a recent survey, only 1.3% of 50.000 medical laboratories in 14 African countries conducted bacteriology testing [
50], and only one-quarter (26%) of healthcare facilities in sub-Saharan Africa had basic environmental cleaning services in place [
1]. There were no time or geographical clusters of publications among the outbreak reports and the cross-sectional surveys.
Out of 12 outbreak articles providing information, 11 occurred in an urban tertiary care setting. All outbreaks involved a single hospital; the affected wards comprised pediatrics (n = 7), surgery and related wards (n = 5), neonatology (n = 4), intensive care units, and hematology–oncology and dialysis (n = 3 each). The outbreak median (range) duration was 12 weeks (1 week–7.25 years); five outbreaks extended more than six months, of which four exceeded a one-year duration. The median (range) number of patients affected was 18 (5–361). In two outbreaks, colonized patients were reported [
51,
52]. Bloodstream infections were reported in 11/13 outbreaks, either alone (n = 4 outbreaks) or associated with specific foci, such as meningitis, wounds, and urinary tract infections (n = 7). These data were comparable to those reported from HICs (29 (1–151) patients affected and 11 (1–104) weeks duration), particularly when the
Burkholderia cepacia outbreak associated with contaminated ethanol (411 bloodstream infection episodes in 361 patients and 7-year duration) [
53] was subtracted from the comparison.
Ten articles reported patient outcome: the median (range) of the case–fatality ratio was 26.0% (0.0–88.5%), and the aggregated case–fatality ratio was 9.1%. This figure was significantly higher than that observed in HICs (median: 0.0% (0.0–60.0%); aggregated case–fatality ratio: 6.1%;
p = 0.027, chi square). The median (range) number of deaths was four (from one to eight). In three outbreaks, case–fatality ratios were ≥40% [
41,
49,
51]. Case–fatality ratios ≥ 20% were reported from five outbreaks in high-risk wards (surgery, neonatology (n = 2), pediatrics, and hemato-oncology) or were associated with interventions. The implicated organisms were MDR
Serratia marcescens,
Elizabethkingia meningosepticum, MDR
Klebsiella pneumoniae,
Enterobacter cloacae, and carbapenem-resistant MDR
Enterobacter cloacae) [
48,
49,
51,
54,
55]. In one outbreak (postsurgical infections associated with chlorhexidine gluconate (CHG) contaminated with
Achromobacter spp.), a single death (among 59 affected patients) was considered unrelated to the infection [
52]. When reported, triggers pointing to a possible outbreak were an unusual high incidence of postoperative wound infections by
Pseudomonas aeruginosa [
47] and the occurrence of a previously rarely observed species, i.e.,
Achromobacter denitrificans [
56].
Most (84.0%, 21/25) cross-sectional surveys aimed to assess the proportion of contamination of in-use products. Other surveys also aimed to study factors contributing to contamination [
57,
58,
59] and to study the causative organisms, their resistance to antibiotics and antiseptics, and their genetic relatedness [
60,
61,
62]. In two surveys, antiseptics and disinfectants were part of other fomites assessed (high-touch surfaces, cleaning tools, patient care items, and leftover vials of medicines) [
63,
64]. Three surveys assessing soaps compared the contamination between bar and liquid soap, respectively [
65,
66,
67]. Routine infection control monitoring detected an anecdotal observation of contaminated in-use alcohol [
68]. Seventeen surveys studied only one hospital. One survey (5 hospitals) was part of a larger study assessing IPC in public maternity units in Kenya [
69], and a nationwide survey (addressing hospitals of all hierarchic levels (n = 39)) was performed in Thailand [
58]. Three other surveys (in Malaysia and Nigeria) were comprised respectively of 6, 16, and 20 hospitals [
59,
63,
70]. In addition to tertiary hospitals in urban settings (n = 17 articles), secondary hospitals (n = 8) and private facilities (n = 1) were addressed.
The median (range) number of different products tested per survey was 5 (1–10); the median number of samples tested (for 24/25 surveys providing data) was 94 (12–16,142 samples), which is higher than in HICs (median: 48 (1–492 samples) [
9]. The nationwide surveys from Malaysia and Thailand each assessed more than 10,000 samples [
58,
70]. Compared to HICs, cross-sectional surveys also more frequently comprised general and private hospitals, as well as multicenter surveys. The large sample sizes allowed for making comparisons between wards and hospitals [
57,
58,
62,
63,
70,
71,
72], different dilutions and formulations (alcohol- versus water-based products, bar versus liquid soaps) [
57,
58,
62,
65,
66,
67,
69,
70], and different storage conditions [
57,
63,
73]. Products most frequently selected were phenol (12 surveys), chlorine (n = 10), alcohol (n = 9), chlorhexidine gluconate-quaternary ammonium compounds (CHG-QUAT), and liquid soaps (8 surveys each); phenol and chlorine were more frequently assessed compared to HIC surveys (6 and 1 surveys, respectively).
In addition to methodological issues (see below in
Section 4.3), challenges were product names (see footnote of
Supplementary Table S3), formulations, and origins. Some brand names (e.g., Lysol and Mercurochrome) stood for different generic products over time and in different countries, and the active components of the products used in the article were not retrievable. Likewise, products not or no longer marketed as antiseptics were listed (e.g., acriflavine, Mercurochrome (in its mercury-containing formulation) and Methimasol), and some products listed as disinfectants (e.g., Harpic and Biotex) were probably household-grade cleaning products. Further, some papers did not list products’ concentrations [
59].
As was the case for articles from HICs [
9], the terminology of antiseptics versus disinfectants was not always correct; and example is povidone-iodine categorized as a disinfectant [
70]. In some articles, the terms antiseptics and disinfectants were used interchangeably [
58,
60,
74].
4.2. Products Involved
Products associated with outbreaks comprised water-based chlorhexidine gluconate (CHG) and chlorhexidine-quaternary ammonium compound (CHG-QUAT) combinations (representing half (7/13) of the articles), as well as ethanol and chlorine (1 product each) (
Table 2). Almost half (n = 6) of these products were primarily used as skin antiseptics for intravenous catheter care [
44,
53,
56] or for topical wound care [
47,
55,
56]; one CHG product and a Dakin solution (i.e., a stabilized chlorine product used as an antiseptic) were used as disinfectants [
49,
75]. In one article, CHG was used both as a disinfectant (to soak nasal suction catheters) and for hand hygiene [
55]. In addition, liquid soap products (among which were two antiseptic soaps) were reported in four articles [
48,
51,
76,
77].
Among 65 products found contaminated in cross-sectional surveys (
Table 3), CHG, QUAT, and CHG-QUAT products (all but one water-based) represented 17 (26.2%) products, and phenol compounds and chlorine accounted for 15 (23.1%) and 12 (18.5%), respectively. Liquid and bar soaps accounted for eight (12.3%) and four (6.2%) products, respectively. Alcohol products were reported in seven articles (10.8%) [
58,
63,
64,
68,
70,
74,
78]. Products were mainly termed disinfectants (n = 51 articles). Indications for use (described in 13 articles) were disinfection of devices and instruments (forceps, thermometers, and suction and ventilation tubes (n = 6 surveys) [
45,
63,
64,
70,
79,
80], skin antisepsis (n = 2) [
81,
82], and hand hygiene (n = 6) [
65,
66,
67,
71,
73]. In one report products, were used both for antisepsis and hand hygiene [
64].
Concentrations of products assessed in outbreak articles were expressed either as dilutions or concentrations and varied: for water-based CHG, they ranged from a 1/2000 dilution of a 5% stock solution [
47] to 2% and 4%, respectively (which are among the highest concentrations of products marketed) [
44,
52]. In one article, a 1/2000 dilution of a 5% CHG-QUAT solution was contaminated [
75]. Expressed as number of samples contaminated versus total number assessed, cross-sectional surveys consistently showed that contamination of CHG, phenol, and chlorine was related to low concentrations (or high dilutions) and water-based formulations [
57,
70,
79]. These observations were also made in articles from HICs [
9]. Contamination was, however, also noted in alcohol-based CHG [
70]. The high number of surveys (n = 7) showing contaminated ethanol was intriguing, but in two of these surveys, contamination was demonstrated by pre-enrichment of the samples. This is a procedure that overestimates the contamination (see below in
Section 4.3) [
64,
74]. Further, in three other articles, cultures of alcohol and alcohol-based products (including CHG and iodine tinctures) remained negative [
47,
60,
82].
The panel of contaminated products in outbreaks and cross-sectional surveys in LMICs was, overall, comparable to that observed in HICs, but some differences were noted, e.g., the proportion of chlorine and phenol products was higher in LMICs. As was the case for HICs, soap products were assessed from the 2000s onwards [
9], and the proportion of soaps associated with outbreaks in LMICs was higher compared to HICs (4/13 (30.7%) versus 11/81 (13.5%), respectively) [
9]. In LMICs, contamination was rarely observed for povidone-iodine, but this product was assessed in only three surveys [
58,
70,
80]; further, an additional outbreak investigation from India (in which an alternative reservoir was identified) showed contamination of povidone-iodine with
Pseudomonas aeruginosa [
42] (
Supplementary Table S5).
Multicenter surveys revealed substantial variation in contamination ratios per hospital [
58,
59,
69,
70,
72,
73]. A nationwide survey in Malaysia found product contamination rates per hospital ranging from 0.5% to 19.5% [
70]; in one from Thailand, contamination ratios per hospital level were 0.0%, 0.7%, 3.3%, 2.3%, and 1.0% at the university, regional, provincial, district, and private levels, respectively [
58]. One study in Nigeria found no large differences between product contamination ratios from different wards (overall ratio: 63.1%; range per ward: 50.0–72.2% [
57]; another survey (assessing liquid and bar soap) found no relation between time of sampling (morning versus afternoon) and contamination ratio [
62].
Among the four surveys that included both bar and liquid samples, three found higher contamination ratios among bar soap samples [
65,
66,
67]; in two of them (assessing > 40 samples for each of bar and liquid samples), the proportions of contaminated bar soaps were 4-fold and 20-fold higher than in liquid soaps (30/50 versus 7/44 and 61/99 versus 2/60, respectively) (
Supplementary Table S4) [
66,
67]. The remaining survey found 11.4% (44/378) of liquid soap products contaminated versus none among the bar soap samples, but the latter comprised only five samples [
62]. Data allowing comparison between plain and antiseptic soaps were not available.
4.3. Epidemic and Microbiological Methods Used
Most (12/13) outbreak investigations conducted a clinical epidemic study to orient the environmental sampling, but half of them (n = 6) provided no details. The other seven investigations used case definitions and conducted case-control studies (5 and 2 investigations, respectively). One investigation combined a cohort and a case-control study [
53], and another sent a questionnaire to clinicians, together with a laboratory report [
56]. In one investigation, AS, DI, and HH products were directly targeted (i.e., without a prior epidemic study) [
47].
All but one environmental investigation (n = 12) assessed a wide range of fomites, such as liquids (e.g., milk for neonates, dialysate, and intravenous and irrigation solutions), high-touch surfaces (e.g., door handles, bedrails, trollies, and trays), and medical devices (e.g., catheters, tubing, endotracheal tubes, ultrasound scanners, stethoscopes, and infusion pumps). One investigation assessed the contamination of indoor air [
55]. Upstream analyses along the chain of supply were conducted in four investigations and covered stock solutions and distilled water used for dilution in pharmacies [
47,
53,
54,
55]. Four investigations mentioned culturing unopened sealed bottles or sachets [
44,
51,
52,
77]. Four investigations assessed healthcare provider hands [
54,
76,
77], patients (throat and rectal swabs) [
51,
55], and mothers of newborns (throat swabs [
55]).
Most (72.0%, 18/25) cross-sectional surveys did not mention the method of sample selection. Three surveys used random selection [
63,
70,
82], one survey selected products according to acceptability and frequency of use [
80], and another (assessing soap products) selected samples from the sinks of toilets and working rooms [
66]. One survey calculated a sample size targeted to the precision of the proportion of all fomites assessed [
64], and another randomly selected 5% of all the products combined [
70]. A nationwide study in Thailand enrolled hospitals (n = 39) according to representativity for hierarchic level and geographical localization [
58].
All the surveys assessed in-use products. In addition, three, six, and one survey assessed freshly prepared products at pharmacies [
58,
70,
81], stock samples in pharmacies and wards [
57,
59,
60,
61,
81,
82], and sealed original containers [
71], respectively. In addition, one survey assessed boiled and tap water used for dilution [
57]. Two-thirds (66.7%, 10/15) of surveys reporting information were conducted in multiple hospital wards, most frequently surgery and related wards (n = 10 surveys), pediatrics (n = 6), intensive care units (n = 5), obstetrics and gynecology (n = 4), and neonatology (n = 3).
Microbiological culture methods were detailed in 76.3% (29/38) of the articles (7 outbreaks and 22 cross-sectional surveys). Four articles used direct plating on agar media [
47,
58,
66,
81]. Seventeen articles used the Kelsey–Maurer method (or a modification of this method), which consists of a 1/10 dilution step of a product (to dilute the biocide effect) and subsequent plating on solid culture media [
9,
45,
83]. Among these articles, 10 used a neutralizer to inactivate the biocide activity of the products, mostly 3% Tween 80. In seven articles, subculturing was performed in duplicate, with incubations at 37 °C and room temperature; incubation times varied considerably (between 24 h and 72 h at 35–37 °C and between 72 h and 7 days at room temperature). Two articles used a modified Kelsey–Maurer method but incubated the products diluted in broth from 24 h to 7 days instead of plating on culture media within 1 h after dilution [
74,
77]. Likewise, two articles inoculated and incubated the products in enrichment broths before subculturing (one article combined both methods) [
51,
64]. Culture techniques relying on broth enrichment cultures only (used in references [
64,
74,
77]) can recover very low concentrations of organisms and, as such, overestimate the contamination rate, as AS, DI, and HH products are not sterile [
9].
Quantitative cultures were performed by the pour plate method (n = 3) [
53,
59,
61] or by inoculating 10-fold dilutions (n = 4) [
60,
71,
72,
84]. Two of these papers also used membrane filtration to assess viable counts [
59,
71]. Five articles mentioned the bacterial count at which the Kelsey–Maurer test was interpreted as positive, i.e., >250 and >1000 colony-forming units/mL (CFU/mL), respectively [
45,
57,
63,
70,
79]. Seven articles reported high bacterial counts but did not provide details about the method used [
45,
59,
60,
61,
65,
71,
75].
Culture media used for direct plating and subculturing were mostly blood agar and nutrient agar (13 and 11 articles, respectively) as general media; one study used Thayer Martin agar as a selective medium for
Elizabethkingia meningoseptica [
55], and another used cetrimide agar to detect
Burkholderia cepacia [
53]. Methods for isolate identification (for 33/38 articles describing these methods) were mostly conventional phenotypic testing (n = 26 articles; in some articles, complemented by commercial kits [
53,
66,
70,
77] and serotyping [
55]. Other methods were automate-based phenotypic testing (n = 3) [
44,
56,
68] and MALDI-TOF [
67]. In 15 articles, antibiotic susceptibility testing (AST) was performed, with all but one by disk diffusion methods. Fourteen studies specified guidelines for interpretation; eight of them mentioned the version number or year of publication (for details, see
Section 4.4).
Relatedness between index and environmental isolates was assessed by phenotypical methods (serotyping, pyocine typing, colony pigment, and AST results [
44,
47,
53,
55,
76]) or molecular testing (plasmid testing, Pulsed-Field Gel Electrophoresis, PCR-based methods, Rep-PCR, RAPD, and whole-genome sequencing [
44,
48,
51,
52,
53,
54,
56,
75,
77]) (
Supplementary Table S2). Two cross-sectional surveys assessed relatedness between contaminating bacteria and bacteria isolated from clinical specimens by phenotypic identification and AST results [
81] and Pulsed-Field Gel Electrophoresis [
62].
As part of a root cause analysis, two articles used membrane filtration to assess water used for dilution [
53,
57], and one of them used sodium thiosulphate to neutralize chlorine [
53].
As part of outbreak investigation, procedures (water treatment, cleaning, maintenance, dialysis, standard IPC procedures, and procedures for catheter care) were reviewed in six investigations [
44,
51,
53,
55,
75,
76], combined with interviews of staff [
53,
56,
75]. Three cross-sectional surveys (from Trinidad and Tobago and Thailand) combined a microbiological survey with a questionnaire for staff to understand and map the lifecycle of products in a health facility [
58,
60,
73]. Procedure reviews, interviews of staff, and observations of practices were part of other surveys [
61,
62,
69,
70,
71,
73,
74,
79].
In addition to the methodological limitations described in
Section 4.1, none of the outbreak reports referred to the ORION guidelines [
32,
33], cross-sectional studies failed to describe sample selection, and there was a plethora of culture techniques. All of these observations are in line with findings from HICs [
9]. In addition to these limitations, outbreak investigations from LMICs were well-performing in terms of clinical epidemic studies and samplings of high-touch, high-risk fomites. Compared with HICs, outbreak investigations in LMICs more frequently assessed staff and patients for colonization (conducted by only 9/68 outbreaks in HICs) but less frequently assessed the upstream tracks of the products (58.0% among 68 outbreak investigations in HICs) [
9]. Among both outbreak investigations and cross-sectional surveys, many used state-of-the-art molecular testing and conducted in-depth interviews and observations to understand the root cause analysis of the contamination; some used advanced techniques (visualization of biofilm or whole-genome sequencing) [
48,
75] and addressed high sample sizes among multiple centers [
58,
70].
Figure 1.
Flow chart of literature search for bacterial contamination of antiseptics, disinfectants, and hand hygiene products in low- and middle-income countries.
Figure 1.
Flow chart of literature search for bacterial contamination of antiseptics, disinfectants, and hand hygiene products in low- and middle-income countries.
Figure 2.
Distribution of (pseudo)outbreak reports and cross-sectional surveys investigating bacterial contamination of antiseptics, disinfectants, and hand hygiene products in high, low-, and middle-income countries among decades. Solid, filled bars represent high-income countries (HICs), while dashed bars refer to low- and middle-income countries (LMICs).
Figure 2.
Distribution of (pseudo)outbreak reports and cross-sectional surveys investigating bacterial contamination of antiseptics, disinfectants, and hand hygiene products in high, low-, and middle-income countries among decades. Solid, filled bars represent high-income countries (HICs), while dashed bars refer to low- and middle-income countries (LMICs).
Table 1.
Geographic distribution of articles reporting bacterial contamination of antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. The numbers represent articles; countries are categorized according to the United Nations geoscheme [
85].
Table 1.
Geographic distribution of articles reporting bacterial contamination of antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. The numbers represent articles; countries are categorized according to the United Nations geoscheme [
85].
United Nations Geoscheme/Countries | Cross-Sectional | Outbreak a | Total |
---|
Eastern Africa | 3 | - | 3 |
Ethiopia | 2 | - | 2 |
Kenya | 1 | - | 1 |
Northern Africa | - | 1 | 1 |
Tunisia | - | 1 | 1 |
Southern Africa | - | 1 | 1 |
South Africa | - | 1 | 1 |
Western Africa | 6 | 2 | 8 |
Nigeria | 6 | 1 | 7 |
Senegal | - | 1 | 1 |
Eastern Asia | 1 | - | 1 |
China | 1 | - | 1 |
Southeastern Asia | 5 | 3 | 8 |
Malaysia | 2 | 2 | 4 |
Thailand | 3 | 1 | 4 |
Southern Asia | 4 | 2 | 6 |
India | 4 | 1 | 5 |
Nepal | - | 1 | 1 |
Western Asia | 4 | 1 | 5 |
Iraq | 1 | - | 1 |
Lebanon | - | 1 | 1 |
Palestine | 2 | - | 2 |
Turkey | 1 | - | 1 |
Latin America and the Caribbean | 2 | 3 | 5 |
Argentina | - | 1 | 1 |
Brazil | 1 | - | 1 |
Colombia | - | 1 a | 1 |
Mexico | - | 1 | 1 |
Trinidad and Tobago | 1 | - | 1 |
Total | 25 | 13 | 38 |
Table 2.
Products implicated in outbreaks and (pseudo)outbreaks (n = 13) associated with contaminated antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. Numbers represent the number of articles. Abbreviations: CHG = chlorhexidine gluconate; QUAT = quaternary ammonium compounds.
Table 2.
Products implicated in outbreaks and (pseudo)outbreaks (n = 13) associated with contaminated antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. Numbers represent the number of articles. Abbreviations: CHG = chlorhexidine gluconate; QUAT = quaternary ammonium compounds.
Decades | Alcohol | CHG a | CHG-QUAT b | Chlorine | Liquid Soap c | Total |
---|
1980s | - | 2 | - | 1 | - | 3 |
2000s | 1 | 1 | 2 | - | 1 | 5 |
2010s | - | 1 a | - | - | 2 | 3 |
2020s | - | 1 | - | - | 1 | 2 |
Total | 1 | 5 | 2 | 1 | 4 | 13 |
Table 3.
Products implicated in cross-sectional surveys (n = 25) reporting contaminated products of antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. Numbers represent contaminated products per survey; numbers exceed the number of articles since several articles reported more than one contaminated product. CHG = chlorhexidine gluconate; H2O2 = Hydrogen peroxide; QUAT = quaternary ammonium compound.
Table 3.
Products implicated in cross-sectional surveys (n = 25) reporting contaminated products of antiseptics, disinfectants, and hand hygiene products in healthcare facilities in low- and middle-income countries. Numbers represent contaminated products per survey; numbers exceed the number of articles since several articles reported more than one contaminated product. CHG = chlorhexidine gluconate; H2O2 = Hydrogen peroxide; QUAT = quaternary ammonium compound.
Decades | Alcohol | CHG a | CHG-QUAT | QUAT | Iodophor | Phenol b | Chlorine | H2O2 | Liquid Soap c | Bar Soap d | Total |
---|
1970s | - | - | - | - | - | 1 | - | - | - | - | 1 |
1990s | 2 | 1 | 2 | 1 | 1 | 5 | 3 | - | - | - | 15 |
2000s | 2 | 4 | 5 | 2 | - | 7 | 6 | 1 | 2 | 3 | 32 |
2010s | 2 | - | 2 | - | - | 2 | 3 | - | 6 | 1 | 16 |
2020s | 1 | - | - | - | - | - | - | - | - | - | 1 |
Total | 7 | 5 | 9 | 3 | 1 | 15 | 12 | 1 | 8 | 4 | 65 |
4.4. Microorganisms Involved
A total of 18 isolates were recovered from 13 outbreak-related articles (
Table 4). One article retrieved two Enterobacterales species from a soap product, while another obtained five species of coagulase-negative staphylococci from contaminated QUAT-CHG [
51,
77]. Enterobacterales were associated with five outbreaks, with three of them related to liquid soap products [
48,
51,
76]. In the two other outbreaks,
Enterobacter cloacae was associated with a Dakin solution (a chlorine product) [
49], and
Serratia marcescens was obtained from a CHG solution [
54]. Non-fermentative Gram-negative rods accounted for seven outbreaks, six of which were obtained from CHG or CHG-QUAT products [
44,
47,
52,
55,
56,
75].
Burkholderia cepacia accounted for three outbreaks, followed by
Achromobacter spp. (n = 2),
Pseudomonas aeruginosa, and
Elizabethkingia meningoseptica (one outbreak each). A single alcohol product (ethanol 70%) was contaminated by
Burkholderia cepacia [
53].
Among a total of 164 single species obtained in 25 cross-sectional surveys (
Table 5), Enterobacterales and non-fermentative Gram-negative rods represented 34.1% and 42.6% of isolates, respectively; the remaining species were Gram-positive cocci and rods, including
Staphylococcus aureus and
Bacillus spp., respectively. Among the Enterobacterales,
Enterobacter spp. ranked first (17, 10.3% of all isolates), followed by
Escherichia coli and
Klebsiella spp. (11 and 10 isolates, respectively). Among the Gram-negative non-fermentative bacteria,
Pseudomonas aeruginosa was the most frequent (retrieved in 20 surveys).
The spectrum of bacteria involved in outbreaks and detected as contaminating flora in cross-sectional surveys is in line with the findings from HICs [
9]. Notable differences were the high proportion of Gram-positive bacteria and the low frequency of
Burkholderia cepacia in cross-sectional surveys in LMICs. However, the high proportion of Gram-positive bacteria may be biased, as five non-aureus staphylococci species were reported from a single outbreak investigation that exclusively used enrichment broths [
77]. Likewise, one survey reporting two Gram-positive isolates cultured the caps (stoppers) of containers and leftover samples [
64]. Further, using numbers of surveys that recovered particular species is not necessarily a reliable proxy for frequency. For example, in the single cross-sectional survey mentioning
Burkholderia cepacia, this species represented the most frequently cultured organism [
71].
For a brief overview of the nature, habitat, and clinical significance of the non-fermentative Gram-negative rods implicated in healthcare-associated outbreaks, we refer to the complementing review [
9]. Of note, some of these bacteria are intrinsically resistant to CHG and QUAT and produce biofilms protecting bacteria from biocides (see
Section 4.5).
Elizabethkingia meningoseptica (formerly
Flavobacterium meningosepticum and
Chryseobacterium meningosepticum) shares the ability of non-fermentative Gram-negative rods to thrive in humid environments but stands out for invasiveness and high case–fatality ratio [
86]. A large-scale outbreak caused by
Burkholderia cepacia in 70% ethanol was, in part, ascribed to its nutritional versatility, i.e., metabolization of alcohol [
53]. Enterobacterales are common hospital-associated bacteria: as an example,
Klebsiella pneumoniae (responsible for an outbreak related to contaminated soap containers in a pediatric oncology unit) represented 20% of blood cultures isolates in a ward [
51]. In such settings, suspicion of an outbreak can be easily overlooked [
87].
Articles assessing bacterial counts above the threshold of the Kelsey–Maurer method reported a wide range of counts, of which the maximum exceeded 10
5 CFU/mL (up to 10
7 CFU/mL) in CHG, CHG-QUAT, para-chloro-meta-xylenol (PCMX), and chlorine- and phenol-based products [
45,
61,
75], as well as in bar and liquid soaps [
65,
71]. These high counts are in line with those obtained from HICs [
9]. In one paper, colony counting provided clues to the reservoir of an outbreak caused by
Burkholderia cepacia in a dialysis unit: the dialysate showed low (30 CFU/mL) bacterial counts, whereas the QUAT disinfectant (later confirmed as the reservoir by molecular testing) showed very high counts (10
5 CFU/mL) [
75].
Of the 15 articles that performed AST, some used an incomplete or inappropriate panel of antibiotics [
61,
74], did not display complete results [
57,
60,
73,
81], or listed uninterpretable or inconsistent results (no denominator, antimicrobial resistance pattern not compatible with species resistance phenotype) [
61,
73,
76]. The remaining panel of interpretable bacteria–antibiotics combinations comprised several bacteria classified as MDR, including third-generation cephalosporin-resistant
Enterobacter cloacae (one carbapenem-resistant) [
48,
49],
Klebsiella pneumoniae [
51], and
Serratia marcescens [
54]. The frequency of MDR bacteria was higher than that reported from HICs [
9], and the above bacteria figure on the WHO list of pathogens prioritized for research [
88]. Of note, some of the non-fermentative Gram-negative bacteria described displayed wild-type antibiotic resistance and did not fulfill the definition of MDR, but their wild-type (i.e., natural) antibiotic resistance by itself entailed resistance to multiple classes of antibiotics [
21] and jeopardized effective antibiotic treatment with antibiotics locally available in LMICs. Examples were
Elizabethkingia meningoseptica [
55] and
Burkholderia cepacia [
44,
53,
75], both of which were resistant to third-generation cephalosporins. Additionally,
Elizabethkingia meningoseptica was resistant to carbapenem antibiotics and the
Burkholderia cepacia complex to aminoglycosides [
21,
86].
Table 4.
Bacteria contaminating antiseptics, disinfectants, and hand hygiene products listed in 12 outbreaks and 1 (pseudo)outbreak in healthcare facilities in low- and middle-income countries. Numbers in cells represent articles citing bacterial species; numbers exceed the number of articles since one outbreak reported more than one bacterial species [
77]. Details about the isolates (bacterial load and references) can be found in
Supplementary Table S2. Abbreviations: CHG = chlorhexidine gluconate; CHG-QUAT = chlorhexidine-quaternary ammonium compound.
Table 4.
Bacteria contaminating antiseptics, disinfectants, and hand hygiene products listed in 12 outbreaks and 1 (pseudo)outbreak in healthcare facilities in low- and middle-income countries. Numbers in cells represent articles citing bacterial species; numbers exceed the number of articles since one outbreak reported more than one bacterial species [
77]. Details about the isolates (bacterial load and references) can be found in
Supplementary Table S2. Abbreviations: CHG = chlorhexidine gluconate; CHG-QUAT = chlorhexidine-quaternary ammonium compound.
Contaminating Bacteria | Alcohol | CHG | CHG-QUAT | Chlorine | Liquid Soap a | Total |
---|
Enterobacterales | - | 1 | - | 1 | 0/2/2 | 6 |
Enterobacter cloacae | - | - | - | 1 | 1 | 2 |
Serratia marcescens | - | 1 | - | - | 1 | 2 |
Citrobacter spp. | - | - | - | - | 1 e | 1 |
Klebsiella pneumoniae | - | - | - | - | 1 e | 1 |
Non-fermentative Gram-negative rods | 1 | 4 | 1 | - | 1/0/0 | 7 |
Burkholderia cepacia | 1 | 1 b | 1 | - | - | 3 |
Achromobacter spp. c | - | 1 | - | - | 1 | 2 |
Pseudomonas aeruginosa | - | 1 | - | - | - | 1 |
Elizabethkingia meningoseptica | - | 1 | - | - | - | 1 |
Gram-positive cocci d | - | - | 5 | - | - | 5 d |
Coagulase-negative staphylococci | - | - | 5 | - | - | 5 |
Total | 1 | 5 | 6 | 1 | 1/2/2 | 18 |
4.6. Attribution and Transmission
Among the outbreak investigations, attributions of AS, DI, and HH products as reservoirs were mainly provided by culturing of an index organism from a suspected product and demonstrating its identity with clinical isolates. In 8/13 investigations, identity was demonstrated by phenotypic [
47,
49,
76] or by genotypic methods [
44,
52,
53,
75,
77]. In five outbreaks, identity was probable but not ascertained given inconsistent results among techniques used (e.g., same PFGE patterns but different AST profiles) or given co-occurrence of different phenotypes or genotypes [
48,
51,
54,
56,
77]. Genetic diversity of healthcare-associated bacteria is not unusual and depends on the evolutionary rate (for
Pseudomonas aeruginosa, this is high) [
95]. Additional evidence was provided by time-relatedness: the onset of one outbreak coincided with the use of a water-based CHG [
47], and in seven outbreaks, cases diminished or stopped after interventions [
44,
51,
52,
53,
56,
75,
77].
In cases of intrinsic contamination, it was plausible that the product itself was the primary source of contamination [
44,
52,
77], but in other outbreaks, it was not always clear whether the contaminated product was the unique reservoir and responsible for transmission. In one outbreak, the ubiquitous presence of the index organisms (
Elizabethkingia meningoseptica) in other fomites and in the air precluded the authors from defining the contaminated CHG containers as the definite reservoir, although contaminated CHG containers were proportionally most frequently affected [
55]. Likewise, in two other outbreaks that did not assess other fomites in addition to a liquid soap product, the latter was considered as a potential but not a definitely proven reservoir [
48,
77].
Transmission routes from reservoir to patients were discussed in 11/13 investigations. Direct contact was plausible when CHG was used as an antiseptic during central venous and urinary catheter insertion, pre-operative skin antisepsis, and wound care. In neonates, transmission was assumed to occur by direct invasion through the respiratory tract and abraded skin [
44,
47,
52,
55,
56].
Burkholderia cepacia in contaminated 70% ethanol was assumed to be transmitted through intravascular catheter insertion by both skin antisepsis and disinfection of the rubber stoppers of heparin vials [
53]. Products used as disinfectants spread by contact with (semi)critical devices, such as a transferring forceps (used to transfer gauzes during the insertion of intravascular catheters) standing in a jar filled with contaminated CHG-QUAT or soaking intravascular catheters in a Dakin solution [
49,
75]. Handborne transmission (i.e., transmission by contamination of healthcare workers’ hands and, subsequently, patients) was assumed in three outbreaks associated with contaminated liquid soap [
51,
76,
77].
Cultures of patients and staff sampled as part of outbreak investigations revealed different results. In the above-mentioned outbreak of
Elizabethkingia meningoseptica in a neonate unit, 8.7%, 4.1%, and 11.0% of neonates, mothers, and patients from other wards had growth of the index organisms from the upper respiratory tract [
55]. The hands of nurses assessed for index organisms grew
Staphylococcus haemolyticus (2/12 cases) and
Serratia marcescens (1/41 cases) [
54,
77] and were negative in another investigation [
76]. Assessing healthcare workers for colonization by bacteria is not recommended, except in cases of specific bacteria and diseases or when specifically oriented by epidemiological investigation [
27,
96]. In addition, the implicated organisms had variable ability for hand colonization (high for Enterobacterales but low for non-fermentative Gram-negative bacteria), and the colonizing flora were transient and eliminated by hand hygiene [
96]. For the culturing of patients, see
Section 4.7.
Four cross-sectional surveys compared environmental isolates with clinical isolates from healthcare associated infections. Two of them found similarities in species identification and AST profiles [
57,
81], while the two others did not [
62,
73].
Attributions of reservoirs and elucidation of transmission routes among the outbreaks in LMICs are in line with the findings obtained from HICs [
9]. Factors hampering reservoir and transmission investigations in HICs, such as availability of enough clinical isolates and products in use reported from HICs, were not mentioned in the present articles but are probably highly relevant to LMICs, given the low volumes of samples processed [
8,
97].
4.7. Interventions
Outbreak control interventions were reported in 11/13 investigations. In addition to removal of the contaminated products, replacements included substitution of contaminated water-based CHG by alcohol-based CHG [
47] or individually packed alcohol pads [
53] and replacing liquid soap with another soap product or an alcohol-based handrub [
76,
77]). In the two outbreaks associated with intrinsically contaminated CHG (in Columbia and Argentina), the manufacturer and national regulatory authorities were notified [
44,
52].
Outbreak control further focused on the use of safe water (either boiled or distilled) [
47,
53,
55] and the efficacy of containers’ reprocessing steps (sterilization or soaking with chlorine) [
51,
56]. Elbow-commanded and smaller-volume containers were each implemented in one article [
51,
55]. Further, the practice of transfer forceps standing in CHG-QUAT-filled jars was banned [
75], and two articles reinforced hand hygiene practices [
55,
76]. Other interventions included heat sterilization of prepared CHG solutions [
55], “fogging” with stabilized hydrogen peroxide [
48], and contact precautions and the cohorting or isolation of colonized and infected patients (in one article, combined with temporary unit closure) [
44,
51].
Follow-up after interventions was mentioned by seven investigations; the period of follow-up (provided by four articles) ranged from 4 weeks to 18 months. One article did not provide the results of the follow-up period [
55], four articles reported an absence of new cases [
51,
75,
77], and two others reported a considerable but incomplete reduction in cases [
47,
53].
Control measures applied to outbreaks associated with AS, DI, and HH products in LMICs are similar to those reported from HICs [
9], apart from minor differences. Unlike the case for reports from HICs (n = 6), articles from LMICs less frequently mentioned training and education of staff (although this was obviously a risk factor for contamination) and more frequently addressed a safe water source. As was the case for articles from HICs, neither the design of the outbreak investigation nor the applied interventions allowed assessing the efficacy of the investigation [
9]. However, not all interventions were evidence-based or proved to be effective: heat sterilization of CHG [
55] is not possible, as it degrades the product [
98], and H
2O
2 fogging reduced but did not eliminate infections associated with a contaminated soap dispenser [
48]. Lack of access (related to stock rupture and financial barriers) precluded the consistent use of single-packed alcohol pads, reverting staff to the in-house production of 70% ethanol and subsequent re-occurrence of infections [
53].
Two cross-sectionals survey organized educational activities for staff [
62,
94]. In another survey of liquid soap samples, a team installed a cleaning program (with disassembly of dispensers), introduced alcohol-based handrub in risk areas, and organized microbiology control at the reception of products [
71].
All but one surveys further provided recommendations, mostly in line with the above discussed interventions: need for hospital policy, guidelines, and procedures (n = 3); appropriate reprocessing, including sterilization of containers (n = 3); monitoring products for contamination (n = 6); appropriate preparation performed by trained and competent staff (n = 6); and small-volume containers with short storage and in-use duration [
78]. Other recommendations were the need for sterile products [
68,
77], monitoring concentrations of products [
79], and the preference of liquid over bar soap, the latter expressed by three surveys that compared both types of soap [
65,
66].
For contact precautions, the WHO recommends actively screening asymptomatic colonization with carbapenem-resistant Enterobacterales and isolating or cohorting patients colonized or infected with carbapenem-resistant organisms [
88]. In the case of carbapenem-resistant
Pseudomonas aeruginosa and
Acinetobacter spp., which are less effective colonizers compared to Enterobacterales, active screening for colonization depends on the local risk and setting [
99]. The WHO recommends that screening and isolation or cohorting should mitigate potential harm, as well as negative social and psychological consequences, and, particularly in low-resource settings, be balanced against local prevalence, availability of resources, other IPC needs, and cultural perceptions of offensiveness and stigma [
88].