INTRODUCTION
Nasal carriage and hand contamination of food handlers with Staphylococcus aureus have been implicated in food poisoning outbreaks [Reference Bennett, Walsh and Gould1, Reference Wei and Chiou2]. Investigation of staphylococcal food poisoning has revealed that colonizing strains of food handlers were attributable in more than 50% of cases [Reference Greig3]. Contamination of food with S. aureus and subsequent growth at ambient temperature leads to production of enterotoxins which are secreted into the food matrix [Reference Bennett, Walsh and Gould1].
It is estimated that S. aureus is persistently carried by 20% of healthy individuals who act as reservoirs for its dissemination, while others are occasionally colonized [Reference van Belkum4]. The percentage of transient carriers can vary affecting the overall prevalence of nasal carriage in a community at a given time.
To investigate patterns of colonization and hand contamination with S. aureus in Hong Kong food handlers, a longitudinal study was initiated in autumn 2002. Collection of the second round of samples commenced in spring 2003 but because of the severe acute respiratory syndrome (SARS) outbreak most of the follow-up samples were collected 6–8 months after the initial sampling which coincided with the end of the outbreak. During the SARS epidemic, improved hand hygiene practices and use of face masks and gloves were strictly implemented for food handlers to reduce viral transmission. Surveillance data indicated that there was a significant decrease in episodes of food poisoning in 2003 which was attributed to compliance with the enhanced hygiene measures [5]. Measures undertaken during the SARS outbreak were supplemented in early 2005 by several strategies including mandatory employment of a certified hygiene officer at all catering establishments and restaurants whose major role was supervision of hand washing and other personal hygiene practices. The hygiene officers were also responsible for providing instructions to all food handlers regarding the proper hand washing technique. In addition, government inspection of food premises was made more rigorous. However, whether the effect of the changes in food handling practices has led to a sustainable effect on nasal carriage and hand contamination has not been investigated. This study aimed to compare prevalence of S. aureus carriage in food handlers over a decade and investigate characteristics of the isolates.
METHODS
Study participants
Between mid-October and mid-December 2002, samples were collected from 619 food handlers at 15 catering establishments. Fourteen of these premises, which included supermarkets, college canteens, centralized kitchens for local hospitals and for a major sporting facility, were revisited in May/June 2003 and a further 527 samples collected. Of the participants, 499 were sampled on both occasions and those repeatedly positive for S. aureus were defined as persistent carriers. In 2011, 434 food handlers were recruited from six large catering establishments which included hotels, college canteens, and centralized kitchens for local hospitals and for a major sporting facility. The sample size was based on the prevalence observed in a recent study of the general public [Reference Zhang6]. Participants in the studies of 2002 and 2011 completed a short questionnaire concerning risk factors for colonization, including demographic data, smoking, years of working experience, and foods handled. In 2011 additional questions involving family members working in healthcare, pet ownership, wearing gloves at work, and recent hospitalization or antibiotic use were added. Details of participating establishments are given in Table 1.
Table 1. Distribution of subjects and sites sampled
* These facilities were visited on all three occasions.
Ethical statement
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, revised in 2008. Ethical approval was obtained from the University Human Subjects Ethics Committee. Each subject was provided with a brief description of the study and gave written consent.
Specimen collection
Nasal swabs were collected from the anterior nares of study participants 1–2 hours after commencement of their shift by the research team using saline-moistened transport swabs (Oxoid, UK), placed in brain heart infusion broth (Oxoid) supplemented with 5% sodium chloride for enrichment and incubated overnight at 37 °C. In 2002 and 2003, specimens were subsequently subcultured onto mannitol salt agar (Oxoid), while for 2011 samples MRSASelect™ chromogenic agar (Bio-Rad UK Ltd, UK) was used. The use of this recently introduced chromogenic agar allowed S. aureus to be more easily differentiated from other staphylococci. On all sampling occasions, hand contamination was determined by fingertip impression of the dominant hand onto mannitol salt agar. Food handlers were instructed not to wash their hands prior to providing the sample. All agar plates were incubated overnight at 37 °C. Colonies resembling S. aureus were picked from primary selective agars and pure cultures obtained by subculture on tryptic soy agar (Oxoid). In order to assess reproducibility of the sampling methodology, sampling was repeated at one institution 2 weeks after the initial investigation of carriage rates.
Bacterial identification and characterization
All isolates were confirmed as S. aureus by DNase production, Staphaurex Plus latex agglutination test (Remel, USA) and amplification of the 16S rRNA gene [Reference Monday and Bohach7]. Methicillin resistance was detected by disc diffusion using cefoxitin and oxacillin [8] and confirmed by mecA detection which was performed simultaneously with SCCmec typing by a multiplex PCR [Reference Kondo9]. Clonal relationships of isolates were identified by sequencing the polymorphic region of the spa gene [Reference Harmsen10].
Susceptibility to other antimicrobial agents; fusidic acid, quinupristin-dalfopristin, erythromycin, clindamycin, co-trimoxazole, ciprofloxacin, gentamicin, chloramphenicol and tetracycline, was determined by disc diffusion [8] except for fusidic acid susceptibility which was performed according to EUCAST guidelines [11]. All zone sizes were determined using an automatic reader, Mastascan Elite (Mast Group Ltd, UK).
Data analysis
Differences in carriage rates over time were compared by trend analysis of the total food-handler population as well as subgroups of food handlers in different types of workplace. Association of potential risk factors with carriage was determined using χ 2 test followed by multivariate analysis using a backward logistic regression model to control for confounders. Analysis was performed using SPSS software v. 19·0 for Windows (SPSS Inc., USA). A P value <0·05 was considered to be statistically significant. Characteristics of the spa loci of the isolates from persistent carriers were compared and the mutation rate calculated as number of mutations divided by time in months [Reference Blumental12]. The epidemic index (E) was calculated as E = 1 – D, where D refers to Simpson's index of diversity, in order to compare diversity of spa types in 2002 and 2011 [Reference Sakwinska13, Reference van Belkum14]. Isolates were grouped into clonal clusters using the criteria of Ruppitsch et al. [Reference Ruppitsch15].
RESULTS
Comparison of the baseline characteristics of the subjects revealed that there were significant differences in proportions of male food handlers and food handlers who were smokers (Table 2). Smoking was found to be significantly associated with male gender (P < 0·001). The nasal carriage rate in 2002 prior to the SARS outbreak was 35% and hand contamination was 41·2%. Repetition of sampling for consistency at one institution yielded an identical nasal carriage rate (28·9%) although some of the nasally colonized participants differed between the samplings. The hand contamination rate was also similar (15·5% first sample, 17·8% second sample). This repeated estimate was not included in the overall calculation.
Table 2. Comparison of characteristics of participating subjects in 2002 and 2011
* Pearson χ 2 test.
Comparison of nasal carriage rates in the different establishments showed there was no significant difference in the rates between institutions in 2002 (P = 0·31), 2003 (P = 0·76) and 2011 (P = 0·86). Stratification of workplace also showed no significant difference in nasal colonization rates (2002: P = 0·20; 2003: P = 0·56; 2011: P = 0·86). In the immediate post-SARS period (2003), both nasal and hand colonization rates were reduced to 23·5% and 11·6%, respectively (P < 0·001). In 2011, the nasal carriage rate was similar to 2003 (22·9%) while the hand contamination rate dropped further to 3·7%. Trend analysis showed this reduction to be significant (P trend < 0·001) (Table 3). In order to determine if this trend was similar in all types of establishment, the trend analysis was repeated for each type of workplace: canteens (P trend < 0·001); hospital kitchens (P trend = 0·06); sporting facility (P trend = 0·009); others (P trend = 0·02) (Table 3). Similarly, trend analysis of hand contamination showed an overall significant reduction (P trend < 0·001), while that of different establishment types varied (P trend < 0·001 to 0·01) (Table 3). The percentage of both colonized and non-colonized subjects whose hands were contaminated was significantly reduced over the period (Table 4).
Table 3. Temporal changes of nasal and hand carriage rates by workplace
* χ 2 test for trend.
† Refers to supermarkets and hotels.
Table 4. Temporal changes in prevalence of nasal colonization and hand contamination with S. aureus with respect to nasal carriage status
In 2002, contact with cooked meat was found to be a risk factor for colonization (OR 1·21, 95% CI 1·0–1·46, P = 0·044), and contact with raw meat almost reached significance (OR 1·12, 95% CI 1·0–1·42, P = 0·062) (Table 5). Of the total 151 colonized workers handling meat, 82 (54%) were in contact with both cooked and raw meat. Raw meat as a risk factor for colonization was confirmed in 2011 (OR 2·7, 95% CI 1·7–4·5, P < 0·001) [Reference Ho, O'Donoghue and Boost16]. No other factors were found to be significantly associated with carriage on either occasion.
Table 5. Analyses of risk factors for nasal carriage with S. aureus (2002)
OR, Odds ratio, aOR, adjusted odds ratio; CI, confidence interval.
* The difference is statistically significant.
Spa-typing for 2002 and 2011 isolates revealed 39 spa types clustered into 13 spa clonal clusters (spa CCs) and 42 spa types clustered into 10 spa CCs, respectively. Between two and 65 isolates were present in each spa CC (Table 6). The epidemic index showed that there was a slight increase of spa type diversity over the period (2002: Epersistent = 0·056; Eoverall = 0·061 vs. 2011: Eoverall = 0·052), reflecting the absence in 2011 of 28 spa types observed in 2002 and 29 types newly detected in 2011. In particular, the number of spa CC15-associated types reduced significantly from 23% to 11% (P = 0·046) while spa CC45-associated types increased from 4% to 12% (P = 0·004) (Table 6).
Table 6. Comparison of spa clonal clusters and types of S. aureus nasal carriage strains in 2002 and 2011
MLST, Multilocus sequence type; CC, clonal cluster; ST, sequence type; n.s., non-significant.
* Defined by visual analysis as described by Ruppitsch et al. [Reference Ruppitsch15].
† Bold font represents spa types present in both 2002 and 2011.
‡ Comparison of number of isolates in the spa CC (χ 2 test).
§ nt = new type nt1, repeat succession: 07-23-12-21-17-254; nt2, repeat succession: 23-34-34-16-34-33-13.
The remaining isolates were singletons not belonging to any of the clusters above, 51 in 2002 and 10 in 2011.
Of the 499 participants sampled in both 2002 and 2003, 89 (17·8%) were persistently nasally colonized. The percentage of transient carriers fell from 20·2% (n = 101) to only 5·4% (n = 27) between these samplings. Eighty (89·8%) of the persistent carriers yielded an isolate with the same spa type on both occasions. Common spa types were t189 (18·8%), t084 (12·5%), t091 (7·5%), t437 (7·5%) and t034 (6·3%). The remaining nine subjects had either a change or a mutation in the colonizing strain (Table 7). Spa types seen in transient carriers were similar to those of persistent carriers. The mutation rate of spa locus in persistent carriers is one mutation per 223 months.
Table 7. Characteristics of altered spa types in persistent carriers
NID, Not in database.
* Bold font indicates change of spa repeats.
† Upper row denotes spa type for the isolate in 2002 and lower row for that in 2003.
There was a higher prevalence of methicillin resistance in nasal isolates in 2011 (1·2%) compared to 2002 (0·6%) and 2003 (0·8%) but this did not reach statistical significance. The majority of isolates were SCCmec types IV or V. However, there was a notable change to a predominance of one spa type, t1081, in 2011. Characteristics of these isolates are shown in Table 8. All isolates of persistent carriers were methicillin sensitive.
Table 8. Methicillin-resistant S. aureus strain characteristics
CIP, Ciprofloxacin; CHL, chloramphenicol; DA, clindamycin; ERY, erythromycin; FD, fusidic acid; GEN, gentamicin; TET, tetracycline.
Resistance to ciprofloxacin increased significantly over the study period from 1·4% to 7% (P = 0·012), whereas susceptibility to erythromycin, clindamycin, gentamicin, tetracycline, and chloramphenicol remained unchanged. All nasal isolates collected were sensitive to linezolid.
DISCUSSION
Our study revealed that the prevalence of nasal carriage and hand contamination with S. aureus decreased remarkably between 2002 and 2003 and that this reduction appears to have been sustained as evidenced by rates observed in 2011. Although nasal colonization rates can vary temporally, the change does not usually reach statistical significance, as was observed in a 9-month longitudinal follow-up of carriers, in which the prevalence of S. aureus varied by only 2% [Reference Sakwinska13]. The repeated sampling at one site showed no variation in the nasal carriage rate and the difference in the hand contamination rate did not reach significance (P = 0·76). This suggests that there was an extraneous factor which affected the carriage rate between years in our study. Baseline characteristics of participants were similar except for the proportion of male food handlers and the increase in smokers. These were found to be related as there is a strong association of smoking with male gender. Neither of these factors was found to be associated with colonization. There was a wide variety of establishments employing food handlers sampled on each occasion, but comparison of nasal colonization rates showed no significant difference between places of employment. In the hospital kitchens included, staff in the kitchens had no contact with either patients or healthcare workers, which may have reduced, although not totally eliminated, environmental pressures typical of hospitals. In the case of the sports facility, once again there was no direct contact between food handlers and sports participants and they each had separate bathroom and changing facilities. Trend analysis showed a significant reduction in both hand and nasal contamination/colonization over the decade of study and this significant trend was detected in almost all of the individual groupings of food handlers. The one exception, i.e. hospital kitchen workers, which was very near to significance (P trend = 0·06), may be attributable to small sample size.
We attributed the considerable decrease in both nasal and hand colonization rates to the enhanced hygiene practices implemented in the food industry during and after the SARS epidemic. In response to the outbreak, use of gloves and masks was strictly enforced for food handlers in Hong Kong. This was accompanied by increased emphasis on the importance of regular hand washing, with the implementation of a government-enforced hygiene supervisor scheme with a major role for monitoring hand washing practices. There were also more in-depth inspections of food premises and practices. These measures could minimize the spread to transient carriers resulting in lower colonization rates as was evidenced in the 2003 sampling in which the majority of positive subjects were persistently colonized. Transient carriers are most likely to become colonized as a result of transmission to their nose from their contaminated hands. This hand contamination is most likely to occur as a result of contact with work surfaces in their environment. Improved hand hygiene and use of gloves would reduce these opportunities for self-colonization. Studies have shown that persistent carriers have higher rates of skin contamination and of shedding of S. aureus than transiently colonized subjects [Reference Wertheim17] Interestingly, a significant decrease in nasal colonization was noted in a national survey of non-institutionalized US citizens, from 32·4% in 2001/2002 to 28·6% in 2003/2004 [Reference Gorwitz18]. This may be the result of global efforts made at that time to improve hygiene for prevention of coronavirus transmission. A similar effect was observed in studies conducted in Hong Kong on subjects from the general population. A carriage rate of 35% was reported in young adults in samples collected in early 2002 [Reference O'Donoghue and Boost19]. A more recent study revealed that the carriage rate had fallen to 24% [Reference Zhang20]. In the months after the epidemic in Hong Kong, telephone surveys revealed a sustainable increase in self-reported hand hygiene practices in the general public [Reference Lau21]. This was confirmed to be sustainable in an investigation in 2009 which revealed that 91% of the general population wash their hands more than six times per day with 46·6% washing more than 10 times per day [Reference Lau22]. Since 2003, although use of gloves and improvements in hand hygiene has been sustained, use of masks has decreased in the food industry except for handlers of some sensitive foods, such as sushi, and for operatives suffering from respiratory infections or rhinitis. The sustained low levels of nasal carriage and hand contamination strongly suggest that prevention of dissemination of S. aureus is largely a result of hand hygiene measures including use of gloves.
Although a number of risk factors for colonization were investigated, only contact with meat appeared to increase the risk. The association with raw meat became stronger in 2011 than in the earlier study (2002). This may be due to the association being obscured by more frequent transient colonization due to contamination of surfaces and equipment in the earlier study (2002) by persistent carriers. In addition, over half of workers handled both raw and cooked meat leading to considerable overlap of these exposures. We have previously reported that one S. aureus strain, t189, was associated with food handlers who regularly handled raw meat at work [Reference Ho, O'Donoghue and Boost16].
While resistance rates to most classes of antimicrobial agents remained stable, susceptibility to ciprofloxacin increased throughout the period. This may reflect the increased use of this agent in Hong Kong [Reference Lam23]. A similar trend was also observed in the USA [Reference Gorwitz18].
Isolation of t189 as the most common spa type, representing 18% of isolates in both 2002 and 2011 is noteworthy because this spa type is rarely reported as a human nasal commensal. A large-scale cross-sectional survey of human S. aureus nasal carriage involving more than 30 000 participants from nine European countries [Reference den Heijer24] failed to detect t189 in methicillin-resistant strains. However, spa-typing was not performed on methicillin-susceptible isolates which are known to differ from methicillin-resistant strains. As mentioned above, S. aureus t189 was associated with exposure to raw meat [Reference Ho, O'Donoghue and Boost16]. Therefore, the presence of this clone in the food production chain may increase risk for its carriage.
There was a large change in circulating spa types between 2002 and 2011. In particular, it was observed that spa CC15 incidence fell while that of spa CC45 increased significantly over the period. Although only nine spa types were detected in both samples, it appears that some clones have been circulating in the region for more than a decade.
Examination of the rate of mutations in the spa locus revealed there had been only two mutations over the period of 445 sample months which equates to a molecular clock of one per 223 months in the spa locus. This mutation rate is considerably slower than those previously reported [Reference Sakwinska13, Reference Kahl25] indicating that colonizing strains in our subjects rarely undergo mutations. A possible explanation for the observed differences in mutation rates may be variation in selective pressure on the strains in the studies. The subjects in the previous studies were regularly exposed to healthcare environments where antimicrobial agents are regularly used. They were either cystic fibrosis patients or healthcare workers while in our study subjects were food handlers working in the community where the selective pressure is less.
The prevalence of methicillin resistance in food handlers was comparable to that observed in the general public [Reference Zhang6], indicating that food handling may not be a risk factor for methicillin-resistant S. aureus (MRSA) carriage. This was in contrast to our previous finding that butchers handling raw meat exhibited a fivefold increased risk for colonization with MRSA [Reference Boost26]. The risk for butchers may be increased because they are additionally responsible for primary cut-up of whole carcasses, including the snout which is the site most likely to be colonized in the slaughtered animal.
CONCLUSION
The decrease in colonization rates following the SARS epidemic may be attributed to reduced opportunities for spread to transient carriers due to use of protective measures accompanied by improved hand hygiene. After the epidemic, the reinforcement of use of gloves and improved hand hygiene together with use of masks in the presence of respiratory symptoms has resulted in a sustainable lower nasal colonization level accompanied by greatly reduced incidence of hand contamination. This reinforces the importance of intensive hygiene education and interventions for the control of communicable diseases.
ACKNOWLEDGEMENTS
The authors thank all the establishments that participated in the study for their help and cooperation. The project was funded by a research grant from the School of Nursing, The Hong Kong Polytechnic University, Kowloon, Hong Kong.
DECLARATION OF INTEREST
None.
