Factors Influencing the Recovery of Microorganisms Using Swabs

410 Dairy, Food and Environmental Sanitation JUNE 2002

A peer-reviewed article.

*Author for correspondence: Phone: 44.29.2041.6453;Fax: 44.29.2041.6306; E-mail: [email protected]

Factors InfluencingRecovery of Micro-

organisms from Surfacesby Use of TraditionalHygiene Swabbing

G. Moore* and C. GriffithFood Safety Research Group, University of Wales Institute, Cardiff (UWIC),

Colchester Avenue, Cardiff, CF23 9XR, UK

Dairy, Food and Environmental Sanitation, Vol. 22, No. 6, Pages 410-421Copyright International Association for Food Protection, 6200 Aurora Ave., Suite 200W, Des Moines, IA 50322

SUMMARY

Although swabbing is widely used in hygiene monitoring and as a reference point forevaluating new methods, information is lacking with regard to the variables affecting theaccuracy of the swabbing technique. A systematic evaluation of the individual components ofthe swabbing procedure removal of bacteria from a surface, release of bacteria from theswab, and overall bacterial recovery was conducted. Stainless steel squares were inoculatedwith known levels of bacteria and were sampled using different swab types and swab-wettingsolutions. Increasing the amount of mechanical energy generated during swabbing significantlyincreased (P < 0.05) the number of bacteria removed from a surface. In general, however,protocols that allowed a higher percentage of bacteria to be removed from the surface hinderedtheir release from the bud. Using swabs in conjunction with a swab-wetting solutionsignificantly improved (P < 0.05) bacterial release, but in the majority of cases this broughtabout no significant improvement (P > 0.05) of overall bacterial recovery. The results of thisstudy have been used to recommend ways in which the traditional swabbing protocol couldbe improved. However, the results also illustrate why traditional microbiology should not bepresumed to be either the gold standard or the optimum means for monitoring surfacecleanliness.

JUNE 2002 Dairy, Food and Environmental Sanitation 411

INTRODUCTION

Traditionally, microbiologicaltesting within the food industry hasfocussed on end-product analysis.Although results of such tests canindicate that problems have oc-curred during processing, they can-not establish the causes of micro-bial contamination (35). Addition-ally, particularly with regard tohigh-risk products with a shortshelf-life, by the time a defect is dis-covered a large amount of unsatis-factory or unsafe food may havebeen produced and sold. In an at-tempt to reduce the incidence offood poisoning, risk-based foodsafety management systems such asHACCP (hazard analysis criticalcontrol point) have been intro-duced and are increasingly being in-corporated into legislation aroundthe world (17). This has led to agreater emphasis being placedupon the monitoring of in-processpreventative control measures(22).

Within the food industr y,cleaning schedules are designed toreduce both food debris and micro-organisms to levels that pose mini-mal risk to the safety and quality ofthe product (19). However, sur-faces that appear clean visually maystill harbor large numbers of micro-organisms that could contaminatethe food. Such cross contaminationhas been identified as an importantcontributory factor in a significantproportion of general foodbornedisease outbreaks recorded in boththe UK and the USA (11, 21). There-fore, for many foods, particularlythose that are ready-to-eat, thecleanliness of food contact surfacesis likely to be identified as beingcritical to food safety (9).

No standard protocol has yetbeen adopted by the food industryfor surface hygiene monitoring(17). Visual inspection of surfacesafter cleaning can reveal gross defi-ciencies caused by the presence ofvisible food debris, but, most foodoperations require information onsurface cleanliness that extends farbeyond the sensitivity of this test

(23). Many authors have recom-mended the use of ATP biolumines-cence as a means to provide, in realtime, an estimate of total surfacecontamination, which is an indica-tion of overall cleaning efficacy (7,9, 18, 28). However, although ATPbioluminescence can have greatvalue in the initial evaluation of sur-face cleanliness, it does have its limi-tations. Examples are its inability todifferentiate between microbial andsoil ATP unless non-bacterial ATP isremoved enzymatically before theassay, its inability to identify differ-ent organism types (37), and in theabsence of food debris, its inabilityto detect the presence of low num-bers of bacteria on a surface (26).Therefore, despite the fact that,within HACCP plans, routine hy-giene monitoring should provideresults rapidly and in time for reme-dial action to be implemented, theenumeration of microorganisms onfood contact surfaces by use of con-ventional microbiological methodsremains an important means of as-sessing the hygienic status of a va-riety of processing environments(6, 25, 30).

The recommended procedurefor the microbiological examinationof food contact surfaces involvesuse of cotton-tipped hygiene swabsor direct surface contact methodssuch as dipslides (27). Hygieneswabs may often be preferred be-cause of the ease with which theycan be used to sample difficult-to-clean, irregular, and uneven sur-faces as well as the fully quantita-tive nature of the results attained.Accurate detection and enumera-tion of microbial contaminants, byuse of the traditional swabbing tech-nique relies initially upon the abil-ity of the swab to remove the mi-croorganisms from the surface, fol-lowed by their effective release fromthe swab bud and their subsequentrecovery and cultivation. However,bacteria become increasingly diffi-cult to remove by use of hygieneswabs, once they have adhered tothe surface, particularly if they havebecome associated with a biofilm(4, 32). Furthermore, the buds of

cotton-tipped swabs retain some ofthe microorganisms removed fromthe surface, again resulting in anapparent reduction of recovery(12). Additionally, inherent limita-tions are associated with the swab-bing technique (2, 15, 33). Theseinclude, for example, the lack ofstandardization of both the swab-bing pattern and the pressure ap-plied to the swab during sampling,both of which can lead to extremevariability in results (16).

Information regarding thecleanliness of food contact surfacescontinues to be of importance tothe food industry and, despite itsacknowledged shortcomings,which can lead to misleading re-sults, use of hygiene swabs remainsa common and accepted means ofdetecting bacteria on food contactsurfaces. Therefore, in an attemptto find ways of optimizing the tra-ditional swabbing protocol, thisinvestigation conducted a system-atic evaluation of each individualcomponent of the swabbing proce-dure (i.e., bacterial removal, re-lease, and overall recovery) to iden-tify key areas where one or moreof these component stages could besignificantly improved.

MATERIALS AND METHODS

Preparation of bacterialculture

A member of the Enterobacte-riaceae was used throughout thisinvestigation and was of particularinterest because of the potentialpathogenicity of this group of bac-teria as well as their wide use asindicator organisms.

A Gram negative, oxidase nega-tive rod was isolated from a foodenvironment and maintained ontryptone soya agar (TSA; Oxoid,Basingstoke, UK). The culture wassub-cultured every 4 weeks and wassubsequently identified, using bio-chemical test strips (API 20E;bioMrieux), as being Salmonellaspp.

A loopful of the bacterial cul-ture was aseptically transferred into


100 ml of nutrient broth no. 2(Oxoid) in a 250 ml conical f laskand incubated at 37oC for 18 h inan orbital shaking incubator (Model4518. Forma Scienti fic Inc.,Marietta, Ohio) at 100 rpm. Theseculture conditions were found toyield approximately 109 CFU/ml.

After incubation, the culturewas serially diluted using 1/4strength Ringer solution (Oxoid).

Preparation of test surfaces

New squares (5 cm 5 cm) offood-grade stainless steel (type 304)were conditioned before use. Thisinvolved them initially being placedin acetone and sonicated for 15 minusing a Sonicleaner (Lucas DaweUltrasonics, London, UK) and thenbeing soaked in a sodium hypochlo-rite solution. This process removedany grease associated with themanufacturing process. Thereafter,between each set of experiments,the coupons were immersed inVirkon (Antec International, Suf-folk, UK) at the manufacturers rec-ommended usage level, before be-ing rinsed, dried and autoclaved(121oC for 15 min).

Assessing removal of bacteriafrom the surface

The methodology used wasbased upon the direct surface agarplate (DSAP) technique describedby Angelotti and Foter (1).

Sterile coupons were asepti-cally transferred to petri dishes and12.5 l of the 10-4 bacterial serial di-lution (approximately 103 CFU) wasinoculated onto each square andspread evenly over the surface. Thesurfaces were then sampled, usinga previously described standardswabbing protocol (9), immedi-ately after inoculation while stillwet or after they had been allowedto air-dry for 1 h.

Once they had been sampled,the coupons were directly overlaidusing plate count agar (PCA;Oxoid). Control surfaces were pre-pared identically to the test couponsbut were directly overlaid without

(1)

having been swabbed. All plateswere incubated at 30oC for 48 h.

After incubation, the numberof colonies present on the surfaceof those coupons that had beenswabbed was compared to the num-ber present on the surface of thecontrol coupons. Each experimentwas based on 10 replicates, and thepercentage of CFU removed fromthe surface during swabbing wasthen calculated using equation 1.

Nrem

= Ncc

Ntc 100

Ncc

where Nrem

= the percentage ofbacteria removed from the surface;Ncc = the mean number of CFUpresent on the surface of the con-trol (un-swabbed) coupons; N

tc =

the number of CFU present on thesurface of the test (swabbed) cou-pons.

During this investigation thesurfaces were sampled using swabstipped with sterile cotton (TSA-6;Technical Service Consultants Lim-ited, Lancashire, UK), dacron (ULH100S; Biotrace, Bridgend, UK),polyurethane foam (HardwoodProducts Company, Guilford, ME),or alginate (TS7, Technical ServiceConsultants Limited).

Cotton swabs were used eitherdry or after being pre-moistenedwith one of a range of swab-wettingagents. These included an MRD-based solution described byBloomfield (3), containing Tween80 (3% w/v), lecithin (0.3% w/v);and sodium thiosulphate (0.1% w/v), a detergent-based blend des-cribed by Tuompo et al. (36) con-sisting of a TRIS-acetate buffer(0.02M, pH 6.7), EDTA (0.1% w/v);and Triton-X-100 (1% w/v), and anMES-based buffer (10mM, pH 6.8)containing Tween 80 (0.03% w/v)and sodium thiosulphate (0.025%w/v). A biofilm disintegrating re-agent (SprayCult; Orion Diag-nostica, Espoo, Finland) and 1/4strength Ringer solution were alsoused to pre-moisten the swabs be-fore use.

Assessing bacterial releasefrom the swab bud andoverall recovery

Sterile pre-moistened or dryswabs were directly inoculatedwith 12.5 l of the 10-4 bacterial se-rial dilution. These swabs, togetherwith those used to sample the cou-pons, were snapped off into either10 ml 1/4 strength Ringer solutionor 10 ml Calgon Ringers (Oxoid) todissolve the alginate swabs. Theswabs were vortexed to release thebacteria from the bud before 1 mlof the bacterial suspension waspipetted into a petri dish. Approxi-mately 15 ml of PCA was added andthe contents mixed well. Once set,the plates were incubated at 30oCfor 2448 h.

The percentage of CFU re-leased from the swab bud was cal-culated using equations 2a or 2b.The efficiency of the samplingmethod (i.e., the overall percentagerecovery) was calculated usingequation 3; a method previouslydescribed by Whyte et al. (39).

Nrel

= N d 100

Nrem I

100

Nrel = N d 100

Nrem Ncc

100

E = N d 100

I

where Nrel = percentage of bac-teria released from the swab bud,N = mean number of CFU countedon replicate plates, d = dilution fac-tor, N

rem = percentage of bacteria

removed from the surface, I = num-ber of CFU inoculated onto surface/ bud, N

cc = mean number of CFU

present on the surface of the con-trol (un-swabbed) coupons, and E= efficiency of the bacterial surfacesampling technique.

Modifications to the above pro-cedures will be discussed in rela-tion to the results to which theyapply.

(2a)

(2b)

(3)


Statistical analysis

Data analysis was performedusing Microsoft Excel 97. Statisticalsignificance set at a level of P < 0.05was determined by use of t-tests oranalysis of variance (ANOVA) com-bined with analysis of the least sig-nificant difference.

RESULTS

Factors affecting overallefficiency of the traditionalhygiene swabbing technique

The efficiency of a bacterialsurface sampling technique can bedefined as its ability to recover mi-croorganisms from a surface. Table1 shows the effect of swab wetness,surface dryness, and sampling pro-cedure on the efficiency of the tra-ditional swabbing technique.

When the coupons were inocu-lated with a small sample volume(12.5 l), the efficiency of the tra-ditional swabbing procedure, whenused to sample a wet surface, wassignificantly greater (P < 0.05)when a pre-moistened swab wasused than when a dry swab wasused. In both cases, the samplingefficiency could be significantly im-proved (P < 0.05) by applying thepour plate rather than the spreadplate methodology.

Swabbing a dry surface ratherthan a wet surface significantly re-duced (P < 0.05) the efficiency ofsurface hygiene swabbing. Further-more, in this case neither swab wet-ness nor sampling procedure sig-nificantly improved the efficiencyof the swabbing technique, whichin all cases did not exceed 0.2%.Therefore, swabbing a dry surfacewith a dry swab was omitted fromthe majority of subsequent experi-ments.

Conversely, when couponswere inoculated with a largersample volume (100 l), surfacedryness did not appear to affect theefficiency of the swabbing tech-nique (P > 0.05). Swab wetness alsohad no significant effect; however,application of the spread and pourplate procedures resulted in a sam-pling efficiency of approximately2% and 13%, respectively. Thisagain suggests that pour plate meth-odology can significantly improve(P < 0.05) the efficiency of tradi-tional hygiene swabbing.

Assessing removal of bacteriafrom the surface

Table 2A shows the percentageof bacteria removed from a wet anda dry stainless steel surface, using

a sterile cotton swab pre-moistenedwith various swab-wetting agents.Depending on which agent wasused to pre-moisten the swab, thepercentage of bacteria removedfrom the wet and dr y surfaceranged from approximately 62% to88% and from 60% to 89% respec-tively.

When a wet surface wassampled, dry swabs or swabs pre-moistened with either 1/4 strengthRinger solution or the MES buffer-based solution removed a signifi-cantly greater (P < 0.05) percentageof bacteria present on the surfacethan swabs pre-moistened with ei-ther the TRIS buffer-based solution,the 3% Tween solution, or Spray-cult. The percentage of bacteria re-moved from a surface when swabswere pre-moistened with these lat-ter three swab-wetting agents did,however, significantly increasewhen the surface sampled was dry.Consequently, there were no signifi-cant differences in the number ofbacteria removed from a dry surfacewhen cotton swabs were pre-moist-ened with any of the different swab-wetting agents, and in all cases a sig-nificantly greater percentage of bac-teria were removed when the swabused was wet than when it was dry.

TABLE 1. Effect of swab wetness, surface dryness and sampling procedure on the efficiency ofthe traditional hygiene swabbing technique

Efficiency of sampling method(mean % SD)

Wet swab* Dry swab

Smaller sample volume Wet surface Spread plate 3.22 0.19 0.29 0.47

Pour plate 6.32 2.82 1.12 1.23

Dry surface Spread plate 0.07 0.05 0

Pour plate 0.15 0.33 0

Larger sample volume Wet surface Spread plate 2.57 0.64 2.12 1.16

Pour plate 12.84 1.88 12.93 3.83

Dry surface Spread plate 2.82 0.78 1.13 0.38

Pour plate 12.28 2.60 13.02 4.58

* cotton-tipped swabs pre-moistened using 1/4 strength Ringer solution 12.5l of 10-4 dilution (approx. 103 CFU) inoculated onto 5cm 5cm coupon 100l of 10-5 dilution (approx. 103 CFU) inoculated onto 5cm 5cm coupon


Table 2B shows that the per-centage of bacteria removed froma wet and a dry stainless steel sur-face using a variety of differentswab types, ranged from approxi-mately 36% to 72% and from 14%to 85% respectively.

When a wet surface wassampled, wet cotton and foamswabs (pre-moistened with 1/4strength Ringer solution) removedsignificantly more (P < 0.05) bacte-ria from the surface than pre-moist-ened alginate swabs. Additionally,dry foam swabs removed signifi-cantly more bacteria from thewet surface than dry cotton swabs.Wet or dry, these three swab typesremoved significantly more (P < 0.05)bacteria from a wet surface than wetor dry dacron swabs.

When a dry surface wassampled, pre-moistened foam swabsremoved a significantly greater per-

centage of bacteria from the sur-face than pre-moistened cottonswabs. Pre-moistened dacron swabsagain removed significantly fewer(P < 0.05) bacteria from the surfacethan any of the other three swabtypes.

Assessing the release ofbacteria from the swab

Table 3A shows the percentageof bacteria released from differentswab types, that prior to direct in-oculation, had been pre-moistenedwith a range of different swab-wet-ting agents. The percentage of bac-teria released from the bud of thecotton, dacron, and foam swabswas significantly lower (P < 0.05)when the swabs were dry thanwhen the swabs were pre-moist-ened with any of the swab-wettingagents. However, except when the3% Tween solution was used to pre-

moisten the bud, the percentage ofbacteria released from a directlyinoculated dry alginate swab didnot differ significantly from thatreleased from pre-moistened algi-nate swabs.

In general, the percentage ofbacteria released from any directlyinoculated pre-moistened swab wassignificantly higher when the swabwas foam or alginate-tipped. How-ever, the percentage of bacteria re-leased from cotton and dacron-tipped swabs could be significantlyimproved by pre-moistening theswabs with 1/4 strength Ringer so-lution.

Tables 3B and 3C show the per-centage of bacteria released fromdifferent swab types pre-moistenedwith a variety of swab-wettingagents and used to sample a wet ordry stainless steel surface. The re-sults take into consideration thepercentage of either the original

TABLE 2B. The mean percentage SD of a bacterial population removed from a wet and drystainless steel surface using a variety of different swab types

Percentage of bacterial population (CFU) removed from surface

Swab type

COTTON DACRON FOAM ALGINATE

Wet surface Wet swab * 69.58 8.76 38.18 23.49 70.38 6.25 55.85 8.50

Dry swab 50.64 13.95 36.34 19.56 71.79 7.69 62.74 10.33

Dry surface Wet swab * 63.24 21.65 14.17 37.04 85.32 6.20 72.79 11.56

* swabs pre-moistened using 1/4 strength Ringer solution

TABLE 2A. The mean percentage SD of a bacterial population removed from a wet and drystainless steel surface using a sterile cotton swab pre-moistened using a range of differentswab-wetting solutions


Swab-wetting solution

Dry swab 1/4 strength TRIS buffer- MES buffer- 3% Tween SpraycultRinger solution based solution based solution

solution

Wet surface 88.20 3.52 85.20 1.74 67.16 9.52 82.42 2.31 62.96 7.48 62.28 4.72

Dry surface 60.32 8.89 79.81 12.02 89.33 5.59 87.47 3.97 80.05 9.11 80.74 8.90


inoculum or the number of colonieson the control coupons that wereremoved during sampling (Tables2A and 2B) and are, therefore, basedon the number of colonies theoreti-cally present on the bud of the dif-ferent swab types.

When a wet surface wassampled, the percentage of bacte-ria released from a cotton swab wassignificantly lower when the swabwas dry than when it was wet. Al-though the majority of swab-wet-

ting agents did not appear to signifi-cantly affect the percentage of bac-teria released from a cotton swab(Table 3B), bacterial release didappear to be significantly increased(P < 0.05) when the swab was pre-moistened with the 3% Tween so-lution.

When a dry surface wassampled, neither swab wetness norswab-wetting agent significantly af-fected the percentage of bacteriareleased from a cotton swab (Table

3B). Furthermore, the swab typeused to sample either a wet or a drysurface did not significantly increasethe percentage of bacteria releasedfrom a swab bud (Table 3C).

Factors improving the efficiencyof the swabbing technique

Tables 4A and 4B show the ef-fect of swab-wetting agent andswab type on the efficiency of thetraditional swabbing technique.

TABLE 3A. The mean percentage SD of bacteria released from a range of directly inoculatedswabs that had been pre-moistened using a variety of different swab-wetting solutions

Percentage of inoculum (CFU) released from bud

Swab type



Dry swab 3.20 0.43 1.75 1.13 19.76 3.92 34.52 10.51

1/4 strength Ringer solution 46.55 1.47 25.79 10.75 45.53 4.95 28.65 11.57

TRIS buffer-based solution 8.92 1.62 15.61 5.36 44.51 6.79 20.56 6.16

MES buffer-based solution 10.72 1.99 18.38 5.39 42.81 3.84 39.22 5.03

3% Tween solution 18.96 4.48 4.99 2.81 48.82 6.18 49.84 6.18

Spraycult 25.55 7.49 16.15 2.76 34.96 6.20 25.89 6.52

TABLE 3B. The mean percentage SD of bacteria released from cotton swabs pre-moistenedusing a range of different swab-wetting agents, taking into consideration the number of bacte-ria each had removed from a wet and dry stainless steel surface during sampling




solution

Wet surface 1.08 0.96 7.98 0.56 9.22 2.94 8.67 3.40 15.23 9.89 9.18 2.26

9.00 8.01 66.58 4.71 76.86 24.52 72.26 28.58 127.03 15.93 76.50 18.89

Dry surface 0 0.17 0.23 0 0.39 0.48 0.42 0.42 0.02 0.02

0 5.81 7.96 0 13.26 16.24 14.49 14.49 0.57 0.79

taking into consideration the percentage of bacteria removed from the original inoculum

taking into consideration the percentage of bacteria removed from the number of colonies counted on control(un-swabbed) coupon


A lthough swab type didnot significantly affect swabbingefficiency (Table 4B), using anypre-moistened swab significantlyimproved (P < 0.05) the efficiencyof the swabbing technique com-pared with results when dry swabswere used (Table 4A). Pre-moisten-ing a cotton swab with the 3%Tween solution resulted in a sam-pling efficiency of 9.6%, signifi-cantly higher than the efficiencywith any other swab-wetting agent.

DISCUSSION

Becase of the time involved inobtaining microbial data, it is notfeasible to use microbiologicalanalysis for monitoring withinHACCP (22); nevertheless, micro-biological methods can be used forvalidation and verification purposes(10).

After being used to sample asurface, a swab can be rubbed overthe surface of an agar plate. Al-though this spread plate techniquecan be used to make a gross esti-mate of surface contamination, asan enumeration technique it is very

TABLE 3C . The mean percentage SD of bacteria released from different swab types, takinginto consideration the number of bacteria each had removed from a wet and dry stainless steelsurface during sampling

Percentage of bacterial population (CFU) released from swab bud

Swab type



23.51 10.26 31.17 25.32 48.04 23.28 65.66 36.37

Dry swab 0.10 0.22 0.14 0.31 0.42 0.38 0.56 0.45

0.98 2.19 1.36 3.05 4.14 3.78 5.53 4.50

Dry surface Wet swab* 3.60 5.48 0 0.49 0.66 4.84 9.25

35.71 54.33 0 4.81 6.59 47.95 91.73


taking into consideration the percentage of bacteria removed from the original inoculum

taking into consideration the percentage of bacteria removed from the number of colonies counted on control(un-swabbed) coupon

inaccurate (12). Vortexing theswab in a diluent, a more effectivemeans of breaking up clumps ofbacteria, is more likely to measurethe number of individual bacterialcells present on a surface (14). Thisis ref lected in Table 1, which sug-gests that in the majority of casesthe efficiency of the swabbing tech-nique can be significantly improvedif the pour plate rather than thespread plate procedure is applied.However, in terms of the minimumdetection limit, the extra dilutionfactor created by adding the swabto 10 ml of diluent lowers the sen-sitivity of the pour plate technique.It is acknowledged, therefore, thatin comparison to the pour plateprocedure, spread plate methodol-ogy is capable of detecting lowernumbers of bacteria on a surface(26).

Although the pour plate tech-nique is widely used and accepted,previous studies have highlightedproblems associated with the recov-ery of bacteria, particularly from adry surface, with traditional hy-giene swabbing (9, 26). This isagain illustrated in Table 1, which

shows that when a surface was in-oculated with a relatively smallsample volume (12.5 l), the effi-ciency of the swabbing techniquewas greater when the surfacesampled was wet than when it wasdry. It has been suggested that thisreduction in swabbing efficiency isdue to a loss in microbial viabilityduring drying (9). However, Stoneand Zottola (34) have shown that,despite appearing smooth to the un-aided eye, stainless steel viewedunder a microscope is very roughand has distinct f laws, that couldharbor bacterial cells. Should waterand/or nutrients also be present,then microbial survival may be en-hanced, and this was evident in thecurrent study. The number of colo-nies on the un-swabbed control cou-pons, directly after inoculation, wasreduced by 45% after the 12.5 l in-oculum had been allowed to dry onthe surface for 1 h. By contrast,when the coupons were inoculatedwith a larger sample volume (100l), there was no apparent reduc-tion in the number of coloniespresent on the coupons after 1 hand no reduction in the efficiencyof the swabbing technique.


ences in results obtained with hy-giene swabs (9). This raises issuesregarding the numerous swab-wet-ting agents available and their effec-tiveness in picking up bacteria fromsurfaces. However, the results pre-sented in Tables 2A and 2B suggestthat mechanical rather than chemi-cal factors have the greatest effecton the number of bacteria removedfrom a surface. Any condition orpractice that increases the amountof mechanical energy generated hasbeen shown to improve the hy-gienic efficiency during handwash-ing. The use of coarse paper towel,for example, results in a greater pro-portion of the resident f lora beingremoved from the hands than whena softer cloth towel is used (24).Similarly, during the present inves-tigation, the use of a coarse foam

swab resulted in a greater propor-tion of bacteria being removed froma surface than when swabs tippedwith a softer material were used(Table 2B).

The dacron swabs used in thisinvestigation removed significantlyfewer (P < 0.05) bacteria from botha wet and a dry surface than eitherthe cotton, sponge or alginateswabs did (Table 2B). The main rea-son was possibly not the dacronbud itself, but rather the greatershaft f lexibility in comparison tothe other swab types, which en-abled less pressure and conse-quently less mechanical energy tobe applied during swabbing. Thisresulted in removal of only 38% ofbacterial colonies from a wet sur-face; however, the effect of such areduction in mechanical energy

TABLE 4B. The effect of swab type upon the efficiency of the traditional hygiene swabbingtechnique

Efficiency of surface sampling procedure (mean % SD)

Swab type



Dry swab 0.05 0.11 0.05 0.11 0.30 0.27 0.35 0.29

Dry surface Wet swab* 0.55 0.84 0 0.10 0.14 0.85 1.63


These results do suggest that aloss in microbial viability can con-tribute to the reduced efficiency ofsurface hygiene swabbing. How-ever, even when a loss in microbialviability did not occur, as when awet surface was sampled, and whenthe pour plate procedure was ap-plied, the efficiency of the swab-bing technique still ranged, depend-ing on inoculum size, from only ap-proximately 6% to 13% (Table 1).Other factors, therefore, must be in-f luencing the recovery of microor-ganisms from the surface.

Removal of bacteria fromthe surface

The increased adhesion of bac-teria to dry surfaces has also beensuggested as a reason for differ-

TABLE 4A. The effect of swab-wetting agent upon the efficiency of the traditional hygieneswabbing technique

Efficiency of sampling procedure (mean % SD)



solution

Wet surface 0.95 0.85 6.80 0.48 6.19 1.98 7.14 2.81 9.59 1.19 5.71 1.41

Dry surface 0 0.14 0.19 0 0.34 0.42 0.34 0.34 0.01 0.02


was even more apparent whendacron swabs were used to samplea dry surface. During this investi-gation the bacteria were re-sus-pended in 1/4 strength Ringersolution, a non-growth enhancingmedium, before being inoculatedonto the steel surfaces. Researchershave speculated that low-nutrientsystems may enhance adherence(5) and this possible increase in thestrength of bacterial attachment tothe surface coupled with a low levelof mechanical energy, resulted inonly 14% of bacterial colonies be-ing removed by the dacron swabsfrom a dry surface.

The cotton and alginate swabsused in this study had similarwooden shafts, allowing equalpressure to be applied with bothswab types. However, when a wetsurface was sampled, pre-moist-ened cotton swabs removed a sig-nificantly greater proportion of thebacteria present than did pre-moist-ened alginate swabs (Table 2B).This concurs with findings of pre-vious studies (36) and may be dueto the natural absorbency of cottonfibers. Cotton, a natural fiber, iscomposed primarily of cellulose,which is very absorbent (8). There-fore, in comparison to other swabtypes, cotton swabs may be capableof absorbing a greater volume of liq-uid present on a surface, togetherwith any bacteria contained withinit that become dislodged from thesurface during swabbing. When adry surface was sampled and theeffective removal of bacteria reliedsolely upon the attachment of thecells to the swab bud, the percent-age of bacteria removed using acotton swab did not differ signifi-cantly from that removed by an al-ginate swab.

The two main factors inf luenc-ing the amount of mechanical en-ergy that can be generated duringswabbing, and consequently thenumber of bacteria that can be re-moved from the surface, appear tobe the inherent properties of theswab bud itself and the degree ofpressure that can be applied to theswab during sampling. However,various substances can also be usedto improve the detachment of bac-teria from surfaces. The addition of

a detergent to a swabbing solution,for example, lowers the surface ten-sion of that solution, increasing itsability to contact the entire surfacearea being sampled (its wetting ef-fect) and helping it to detach cellsto be f lushed from the surface (itsrinsing effect) (31, 36). Further-more, the incorporation of a deter-gent may prevent the re-dispositionof lifted organisms back onto thesurface.

During this investigation, cot-ton swabs were pre-moistened us-ing a range of different swabbingsolutions. When used to sample awet surface, those swabs which hadbeen pre-moistened with solutionscontaining comparatively high con-centrations of detergent-type sub-stances removed significantly fewerbacteria (P < 0.05) than both the dryswabs and those pre-moistenedwith solutions containing little orno detergent (Table 2A). In thiscase, the detergent in the swabbingsolution may also have reduced thesurface tension of the liquid on thesurface. This enhanced wetting ef-fect may have reduced the mechani-cal energy generated by the swab-bing action and thus reduced thenumber of bacteria removed fromthe surface.

It has been documented thatremoval of bacteria from a dry sur-face can be significantly improvedby using a wet swab (31). This isillustrated in Table 2A, which sug-gests that although dry swabs arecapable of removing 60% of the bac-teria present on a dry surface, useof a swab-wetting solution can re-sult in removal of 80% to 90% of thebacterial population. However, de-spite the possibility of increasedbacterial adherence, the rinsing ef-fect of the detergent-based solutionsdid not appear to remove a signifi-cantly greater proportion of bacte-ria from a dry surface than did ba-sic 1/4 strength Ringer solution. Itis acknowledged, however, that thebacteria used in this study were al-lowed to adhere to the surface foronly 1 h; true biofilms include notonly the adherent cells but also thematrix of extracellular material pro-duced by the bacteria (20). Deter-gent-based swabbing solutions canalso possess emulsifying, saponify-ing, peptizing and dissolving prop-

erties (31) and may, therefore, bemore effective than non-detergent-based solutions in removing bacte-ria associated with a biofilm.

In this case, however, the re-sults presented in Tables 2A and 2Bsuggest that despite the possibilityof bacteria adhering more stronglyto a dry surface, the reduced effi-ciency of the swabbing techniqueshown in Table 1 is probably notdue to a reduced number of bacte-ria removed from the surface.

Release of bacteria fromswab bud

Reliable plate counts will beobtained only if microorganismsthat have been removed from thesurface, are effectively releasedfrom the swab bud.

The two sets of data presentedin Tables 3B and 3C take into con-sideration the percentage of eitherthe original inoculum or the num-ber of colonies on the control cou-pons that were removed duringsampling. The results therefore rep-resent the percentage of bacteriatheoretically present on the swabthat was released from the bud dur-ing vortexing. Angelotti et al. (2)reported that the direct surface agarplate technique was capable of de-tecting between 88.5% and 99.3%of bacterial spore contamination onnonporous surfaces. However, inthe current investigation, the num-ber of CFUs present on the wet con-trol coupons was approximately10% of the initial inoculum. Thismay have been due to a relativelyhigh degree of cellular aggregationand may account for the differencesbetween the two sets of results pre-sented in both Tables 3B and 3C.

When a wet surface wassampled with a cotton swab pre-moistened using the 3% Tween so-lution, the percentage of bacteriareleased from the bud was signifi-cantly higher than when any otherswab-wetting agent was used (Table3B). Previous studies have used a 1%Tween 80 solution to prevent cellclumping (13). The amount ofTween 80 present on the bud of theswabs during this study may havebeen sufficient to facilitate breakupof clumps of bacterial cells during


vortexing, thus improving bacterialrecovery.

Although not statistically sig-nificant (P > 0.05), the percentageof bacteria released from the cottonswabs was less than that releasedfrom the other three swab typeswhen a wet surface was sampled(Table 3C). This could have beendue to the greater absorbency of thecotton fibres leading to liquid, to-gether with any microorganismspresent in it, being absorbed intoand becoming trapped within thefibres of the cotton bud. Those char-acteristics, therefore, that may en-able a cotton swab to remove a highproportion of bacteria from a sur-face may be the same characteris-tics that prevent those bacteria frombeing released from the swab bud.Conversely, despite removing a sig-nificantly smaller number of bacte-ria from a wet surface, dacronswabs appeared to release a statis-tically similar proportion to thatreleased from the other three swabtypes. Dacron is a polyester, andpolyester fiber is one of the least ab-sorbent of all fibers; consequently,almost all moisture will lie on thesurface of a dacron swab ratherthan penetrate the bud (8). As aresult, fewer bacteria may becometrapped within the bud, allowingvortexing to remove a greater pro-portion of them.

How the absorbency of theswab appears to inf luence the per-centage of bacteria removed fromthe bud is further illustrated inTable 3B. These results indicate thatsignificantly fewer bacteria werereleased from a dry cotton swabthan from pre-moistened swabs.Saturation or near-saturation of theswab bud prior to sampling a wetsurface may prevent all of the liq-uid removed from the surface dur-ing swabbing, together with thosemicroorganisms in it, being ab-sorbed and becoming trappedwithin the swab bud.

The results presented in Table3A also suggest that trapping ofbacteria in the swab bud is an im-portant factor in reducing the re-covery of microorganisms, particu-larly from a dry swab. Researchershave previously reported on the ad-vantages of placing a calcium algi-

nate swab in a sodium hexameta-phosphate solution (Calgon Ring-ers). It has been stated that in ashort period of vigorous shaking,the calcium alginate dissolves,thereby freeing trapped organismsand reesulting in bacterial countshigher than those obtained with acotton swab (38). In the presentstudy, the percentage of bacteriareleased from a directly inoculated,dry alginate swab was significantlygreater (P < 0.05) than that releasedfrom the other three swab types(Table 3A). Furthermore, exceptwhen a directly inoculated alginateswab was pre-moistened with the3% Tween solution, the percentageof bacteria released from a dry swabdid not significantly differ (P > 0.05)from that released from a pre-moist-ened swab.

The results in Table 3B alsosuggest that after an alginate swabis used to sample a surface, a greaterpercentage of bacteria are releasedfrom it than from the other swabtypes tested. However, because ofthe extreme variability in the num-ber of bacteria recovered from thereplicate samples, these differenceswere not significantly different(P > 0.05). Close agreement be-tween the number of bacteria re-leased from cotton and from algi-nate swabs has also been observedin a previous study, the authors ofwhich hypothesized that calciumalginate or sodium hexametaphos-phate may exhibit some inhibitoryproperties (2). However, unpub-lished data from experiments car-ried out during the present investi-gation suggest that this is not thecase.

Variations in swab bud sizecould influence the ease and degreeof dissolving of the calcium alginateswabs, and during this study it wasnoted that alginate swabs did notdissolve completely; rather, aslightly viscous suspension wasformed. A more simple scenario,therefore, for the perhaps unex-pected similarity in the percentageof bacteria released from the algi-nate and other three swab types isthat the bacteria may simply remainbound either to undissolved fibersor to those fibers present in solu-tion.

The percentage of bacteria re-leased from a directly inoculatedbud was significantly greater thanthat released from swabs used tosample a wet surface. This may havebeen due to the pressure applied tothe swab during surface samplingcausing the bacterial cells to be-come more firmly adsorbed to thefibers. The percentage of bacteriareleased from a swab used to samplea dry surface was significantly lowerthan that released from swabs usedto sample a wet surface. Further-more, changing swab type or alter-ing the swab-wetting agent did notimprove bacterial release. Changesin environmental conditions dur-ing the process of attachment, forexample, can affect the genetic re-sponses of bacteria, resulting in achange of phenotype (4). However,such morphological changes are un-likely to occur after an attachmenttime of only 1 h. Bacteria are alsosubjected to a certain amount ofstress during the swabbing proce-dure itself, and it has been sug-gested that plate count methodsmay not detect all viable cells, par-ticularly those injured by environ-mental stresses (40). Further work,therefore, could be carried out us-ing epifluorescent microscopy todetermine the number of viable cellspresent on the swab bud before andafter vortexing.

CONCLUSION

The main purposes of ensuringhigh standards of cleanliness infood premises are to maintain shelflife and to protect public health byensuring that food does not becomecontaminated. Although variousmethods of detecting microbial sur-face contamination exist, there isno consensus as to an accepted stan-dard method; however, the method-ology used should permit fully reli-able detection even when organ-isms are present in low numbers.The use of hygiene swabs to samplefood contact surfaces remains animportant means of measuring theeffectiveness of sanitation proce-dures, although the issues previ-ously discussed all affect the over-all efficiency of the swabbing tech-nique (Tables 4A and 4B).


In general, results of this studysuggest that the inherent propertiesof a swab that allow a relatively highpercentage of bacteria to be re-moved from the surface tend tohinder their release from the bud,when a wet surface is swabbed. Asa result, swab type did not appearto have a significant effect upon theoverall efficiency of surface hygieneswabbing (Table 4A). Similarly, themechanical energy generated by theuse of dry swabs allowed them toremove a high proportion of bacte-ria from the surface; however, pre-moistening the swabs appeared toimprove bacterial release and there-fore sampling efficiency.

Although removing compara-tively fewer bacteria from a wetsurface, cotton swabs premoistenedwith a 3% Tween solution showedimproved efficiency of the swab-bing technique. It is unclear, how-ever, whether this was due to an in-crease in the number of bacteriareleased from the swab or to en-hanced anti-clumping properties ofthe swabbing solution. Neverthe-less, within the food industry it isadvisable that solutions used to pre-moisten swabs include agents ca-pable of neutralizing the effects ofresidual detergents and/or disinfec-tants, that may be picked up by theswab during sampling. BecauseTween 80 can neutralize the effectsof quaternary ammonium com-pounds (QAC) (29), the advantagesof its incorporation in a swab-wet-ting solution may be two-fold.

During this investigation, there-fore, the optimum sampling effi-ciency was achieved by swabbing awet surface using a cotton swabpre-moistened with the 3% Tweensolution; however, efficiency wasstill only 9.6%. Results strongly sug-gest that release of bacteria fromthe swab bud is the most importantfactor in the recovery of microor-ganisms from surfaces by use of thetraditional swabbing technique.However, despite evaluating the ef-fects of both sonication and in-creased vortex time, this investiga-tion was unsuccessful in discover-ing a more effective means of re-leasing the bacteria. Additionally,no explanation has been found as towhy the problems associated with

bacterial release appear to be exac-erbated when a dry surface isswabbed, and as a result furtherresearch is warranted. Finally, al-though unable to provide definitiveanswers, this study does demon-strate that traditional microbiologyshould not necessarily be presumedto be either the gold standard orthe optimum means for monitoringsurface cleanliness.

REFERENCES

1. Angelotti, R., and M. J. Foter. 1958.A direct surface agar plate labora-tory method for quantitatively de-tecting bacterial contamination onnonporous surface. Food ResearchInt. 23:170174.

2. Angelotti, R., M. J. Foter, K. A.Busch, and K. H. Lewis. 1958. Acomparative evaluation of methodsfor determining the bacterial con-tamination of surfaces. Food Re-search Int. 23:175185.

3. Bloomfield, S. F. 1991. Methods forassessing antimicrobial activity,p. 122. In S. P. Denyer, and W. B.Hugo (eds.). Mechanisms of actionof chemical biocides: Their studyand exploitation. Society for Ap-plied Bacteriology Technical SeriesNo. 27. Blackwell Scientific Publi-cations, Oxford.

4. Bredholt, S., J. Maukonen, K. Kujan-p, T. Alanko, U. Olofson, U. Hus-mark, A. M. Sjberg, and G. Wirt-anen. 1999. Microbial methods forassessment of cleaning and disinfec-tion of food-processing surfacescleaned in a low-pressure system.Eur. Food Res. Technol. 209:145152.

5. Brown, C. M., and J. R. Hunter.1977. Growth of bacteria at sur-faces: Inf luence of nutrient limita-tions. FEMS Microbiol. Lett. 1:163166.

6. Brown, M. H., C. O. Gill, J. Hollings-worth, R. Nickelson II, S. Seward,J. J. Sheridan, T. Stevenson, J. L.Sumner, D. M. Theno, W. R.Usborne, and D. Zink. 2000. Therole of microbiological testing insystems for assuring the safety ofbeef. Int. J. Food Microbiol. 62:716.

7. Chen, J. 2000. ATP biolumines-cence: A rapid indicator for envi-ronmental hygiene and microbialquality of meats. Dair y FoodEnviron. Sanit. 20:617620.

8. Corbman, B. P. 1985. Textiles. Fi-ber to fabric. 6th ed. McGraw-HillBook Co., Singapore.

9. Davidson, C. A., C. J. Griffith, A. C.Peters, and L. M. Fielding. 1999.Evaluation of two methods for

monitoring surface cleanliness ATP bioluminescence and tradi-tional hygiene swabbing. Lumines-cence 14:3338.

10. Dillon, M., and C. Griffith. 2001How to HACCP: A managementguide. 3rd ed. M.D. Associates,Grimsby.

11. Evans, H. S., P. Madden, C. Douglas,G. K. Adak, S. J. OBrien, T. Djuretic,P. G. Wall, and R. Stanwell-Smith.1998. General outbreaks of infec-tious intestinal disease in Englandand Wales: 1995 and 1996. Comm.Dis. Publ. Health 1:165171.

12. Favero, M. S., J. J. McDade, J. A.Robertsen, R. K. Hoffman, andR. W. Edwards. 1968. Microbiologi-cal sampling of surfaces. J. Appl.Bacteriol. 31:336343.

13. Franz, C. M. A. P., and A. von Holy.1994. Optimization of factors inf lu-encing determination of heat resis-tance of meat-spoilage lactic acidbacteria. S. Afr. J. Sci. 90:535538.

14. Gilbert, R. J. 1970. Comparison ofmaterials used for cleaning equip-ment in retail food premises, andof two methods for the enumera-tion of bacteria on cleaned equip-ment and work surfaces. J. Hyg.Camb. 68:221232.

15. Greene, V. W., and L. G. Herman.1961. Problems associated withsurface sampling techniques andapparatus in the institutional envi-ronment. J. Food Milk Technol.24:262265.

16. Griffith, C., C. Davidson, and L.Fielding. 1997. Casting light on re-producibility. Int Food Hyg. 8:3536.

17. Griffith, C. J., C. A. Davidson, A. C.Peters, and L. M. Fielding. 1997.Towards a strategic cleaning assess-ment programme: hygiene monitor-ing and ATP luminometry, an op-tions appraisal. Food Sci. Technol.Today 11:1524.

18. Griffiths, M. W. 1996. The role ofATP bioluminescence in the foodindustry: New light on old prob-lems. Food Technol. 6:6271.

19. Holah, J. T. 1992. Industrial moni-toring: Hygiene in food processing,p. 645659. In L. F. Melo. (ed).Biofilms science and technology.Kluwer Academic Publishers, TheNetherlands.

20. Hood, S. K., and E. A. Zottola. 1997.Adherence to stainless steel byfoodborne microorganisms duringgrowth in model food systems. Int.J. Food Microbiol. 37:145153.

21. Kassa, H., B. Harrington, M. Bisesi,and S. Khuder. 2001. Comparisonsof microbiological evaluations of se-lected kitchen areas with visual in-spections for preventing potentialrisk of foodborne outbreaks in food


service operations. J. Food Prot.64:509513.

22. Kvenberg, J. E., and D. J. Schwalm.2000. Use of microbial data for Haz-ard Analysis and Critical ControlPoint verification Food and DrugAdministration perspective. J. FoodProt. 63:810814.

23. Mackintosh, R. D. 1990. Hygienemonitoring in the food processingindustry. Eur. Food Drink Rev.55:5558.

24. Michaels, B., V. Gangar, T. Ayers,E. Meyers, and M. S. Curiale. 2001.The significance of hand drying af-ter handwashing, p. 294301. In J.S. A. Edwards and M. M. Hewedi.(eds.). Culinary arts and sciencesIII: Global and national perspec-tives. Worshipful Company ofCooks Centre for Culinary Re-search, Poole.

25. Miettinen, H., K. Aarnisalo, S. Salo,and A-M. Sjberg. 2001. Evaluationof surface contamination and thepresence of Listeria monocyto-genes in fish processing factories.J. Food Prot. 64:635639.

26. Moore, G., C. Griffith, and L. Field-ing. 2001. A comparison of tradi-tional and recently developed meth-ods for monitoring surface hygienewithin the food industry: A labora-tory study. Dairy Food Environ.Sanit. 21:478488.

27. Mossel, D. A. A., J. E. L. Corry, C. B.Struijk, and R. M. Baird. 1995. Rec-ommended monitoring proce-dures, p. 405446. In Essentials of

the Microbiology of Foods. JohnWiley and Sons, Chichester.

28. Ogden, K. 1993. Practical experi-ences of hygiene control using ATPbioluminescence. J. Inst. Brew.99:389393.

29. Russell, A. D. 1981. Neutralizationprocedures in the evaluation ofbactericidal activity, p. 4559. InC. H. Collins, M. C. Allwood, S. F.Bloomfield, and A. Fox (eds.). Dis-infectants: their use and evaluationof effectiveness. Academic Press,London.

30. Russell, S. M., N. A. Cox, and J. S.Bailey. 1997. Microbiological meth-ods for sampling poultry process-ing plant equipment. J. App. Poul-try Res. 6:229233.

31. Salo, S., and G. Wirtanen. 1999. De-tergent based blends for swabbingand dipslides in improved surfacessampling, p. 121128. In J. Wim-penny, P. Gilbert, J. Walker, M.Brading, and R. Bayston (eds.).Biofilms The good, the bad andthe ugly. Bioline, Cardiff.

32. Salo, S., A. Laine, T. Alanko, A-M.Sjberg, and G. Wirtanen. 2000.Validation of the microbiologicalmethods hygicult dipslide, contactplate, and swabbing in surface hy-giene control: A Nordic collabora-tive study. J. AOAC Int. 83:13571365.

33. Silliker, J. H., and D. A. Gabis. 1975.A cellulose sponge sampling tech-nique for surfaces. J. Milk FoodTechnol. 38:504.

34. Stone, L. S., and E. A. Zottola. 1985.Scanning electron microscopystudy of stainless-steel finishes usedin food processing equipment.Food Technol. 39:110114.

35. te Giffel, M. C., J. Meeuwisse, andP. de Jong. 2001. Control of milkprocessing based on rapid detec-tion of microorganisms. Food Con-trol 12:305309.

36. Tuompo, H., S. Salo, L. Scheinin, T.Mattila-Sandholm, and G. Wirtanen.1999. Chelating agents and deter-gents in sampling biofilms, p. 113-120. In J. Wimpenny, P. Gilbert, J.Walker, M. Brading and R. Bayston(eds.). Biofilms The good, the badand the ugly. Bioline, Cardiff.

37. Vanne, L., M. Karwoski, S. Karppinen,and A-M. Sjberg. 1999. HACCP-based food quality control andrapid detection methods for micro-organisms. Food Control 7:263276.

38. Walter, W. G. 1955. Symposium onmethods for determining bacterialcontamination on surfaces. Bact.Rev. 19:284287.

39. Whyte, W., W. Carson, andA. Hambraeus. 1989. Methods forcalculating the efficiency of bacte-rial surface sampling techniques.J. Hosp. Infect. 13:3341.

40. Yu, F. P., B. H. Pyle, and G. A.McFeters. 1993. A direct viablecount method for the enumerationof attached bacteria and assessmentof biofilm disinfection. J. Microbiol.Methods 17:167180.

Documents

Factors Influencing the Recovery of Microorganisms Using Swabs