Recovery balance: a method for estimating losses in a Bacillus anthracis spore sampling protocol


  • S.M. Da Silva,

    Corresponding author
    • Biosystems and Biomaterials Division, Material Measurements Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, USA
    Search for more papers by this author
  • A.A. Urbas,

    1. Biosystems and Biomaterials Division, Material Measurements Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, USA
    Search for more papers by this author
  • J.J. Filliben,

    1. Statistical Engineering Division, National Institute of Standards and Technology, Gaithersburg, MD, USA
    Search for more papers by this author
  • J.B. Morrow

    1. Biosystems and Biomaterials Division, Material Measurements Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, USA
    Search for more papers by this author


Sandra M. Da Silva, National Institute of Standard and Technology, 100 Bureau Dr., Gaithersburg, MD 20899-8315, USA. E-mail:



The aim of this study was to develop a method to calculate the performance, and isolate error contributions occurring in a microbial surface sampling protocol.

Methods and Results

The experiments were conducted using a slip/peel tester to provide consistent pressure during the wipe collection. Fluorescence microscopy was used to count spores deposited on the coupon prior to sampling. The mean recovery efficiency (RE) as well as the efficiency of each step in the process was estimated by a recovery balance (RB), similar to a mass balance. Two studies were conducted in this work. In the first one, the recovery of spores from the solution (REsoln) was 57·7% (SD = 8·0), while spores left on the glass surface after wiping (REb+c) was 2·8% (SD = 2·4). The RE of spores adhered to the tube wall (REtube) and glass surface (REsurf) was 1·2% (SD = 19·6) and 5·8% (SD = 7·1), respectively. From the recovery balance, it was determined that 39·9% (SD = 21·2) of spores were lost to the wipe (REwipe). The applicability of the RB method was demonstrated in a second study by examining the relative impact of parameters affecting spore collection including relative humidity, wipe material, wetting agent and nonporous surfaces.


The approach used in this study pointed out the need for a closer analysis of the complex interaction between spores and wipe material because a substantial percentage of spores were lost to the wipe.

Significance and Impact of the Study

The recovery balance, in association with independent controls, provides an account for error contribution and potential variability on each step of the sampling protocol. The approach is not meant to be a replacement for field or laboratory validation of wipe recoveries but promote the development of new collection methodologies and support protocol optimization in laboratory settings.


Understanding and identifying loss of biological materials during surface sample collection and processing is fundamental to the development of accurate methods to assess surface contamination and decontamination efficacy. Numerous efforts have sought to characterize the efficiency of surface sample collection procedures (Rose et al. 2004; Hodges et al. 2006, 2010; Brown et al. 2007; Edmonds et al. 2009; Lewandowski et al. 2010). However, optimizing method effectiveness is complicated by the abundant environmental parameters encountered in real-world sample collection scenarios (surface materials, humidity, temperature, contaminant properties, etc.) and the multiple steps in sample collection protocols that factor into the overall recovery efficacy of a method. Two recent studies resulted in the validation of two methods for sampling Bacillus anthracis from stainless steel using cellulose sponge-wipes (Rose et al. 2011) and swabs (Hodges et al. 2010). Although a validation study is critically important to establish method reliability, it addresses only whether a given method is adequate for the intended use (BIPM 2008). Surface sample collection is a complex process, and there is a need for a better understanding of the losses contributing to sampling inefficiencies to support biothreat detection technology performance estimates and to develop new and innovative sample collection technologies. In addition, furthering the science of surface sample collection of microbes is important not only for biothreat sample collection mission response (US Department of Homeland and Security (DHS) 2011; US EPA and Technical Brief – Bio-response Operational Testing and Evaluation (BOTE) Project 2012) but also for the food industry (Bower et al. 1996; Moore and Griffith 2002; Gunduz et al. 2008), space exploration (Venkateswaran et al. 2004; La Duc et al. 2007; Schubert et al. 2008; Fajardo-Cavazos et al. 2010) and the clinical laboratory environment (Bakker and Faoagali 1977; Anderson et al. 1982; Babb et al. 1989; Al-Hamad and Maxwell 2008).

In many cases, interpretation and comparison of results to better understand the contributions to observed differences in recovery efficiency is very difficult due to variation in experimental conditions employed and error introduced by differences in the approach (e.g. applied force and directional changes during the sampling) taken by the sampler. Estill et al. (2009) reported a recovery performance of c. 5·0% for sampling aerosolized B. anthracis deposited at a concentration of 200 CFU per 100 cm2 on stainless steel with macrofoam swabs and extraction using Butterfield buffer containing Tween 80 (BBT). Hodges et al. (2006) reported a recovery efficiency about 10-fold higher for sample collection of B. anthracis liquid deposited at a concentration of 6 × 104 spores per ml on the same type of substrate with the same adsorptive material (macrofoam swab) but extraction using a phosphate-buffered saline solution containing Tween 80 (PBST). It is difficult to establish whether the dramatic difference in performance between these two studies can be attributed to the extraction solution, micro-organism surface deposition method, surface loading concentration or other factors such as operator variability. Examples of uncertainties inherent to operator variability can be observed in validation studies, in which multiple laboratories are involved in a given study. In a validation study using macrofoam swab for the recovery of viable B. anthracis from steel surfaces involving twelve laboratories, the overall coefficient of variation between laboratories was 16·5% for swabs containing only spores and 25% for swabs containing spores in addition to dust (Hodges et al. 2010). In another validation study, the authors observed a variation of 19·9–31·3% between nine laboratories used to validate a protocol for sampling B. anthracis deposited onto steel surfaces using cellulose sponge-wipe (Rose et al. 2011). In both studies, part of the uncertainty is due to differences in operator wiping a given surface as well as variability in counting colonies (Fowler et al. 1978). A friction/slip tester, a device used to provide reproducible kinetic frictional force (Fkinetic) during the wiping, has been used for evaluating collection efficiency of wipe materials (Verkouteren et al. 2008) and force-sensing resistor (FSR) films for measuring force during wipe sampling (Verkouteren et al. 2012). In both cases, minimizing the operator variability is the goal, in which it can be applied as training guides and in the development of standard practices.

New methods that include direct particle counting provide useful laboratory tools for assessing bacterial contaminant cell counts and potential changes in viability, as well as provide valuable insights into surface sample collection methodologies (Maruyama et al. 2004; Davey 2011; Grand et al. 2011). Direct particle count methods have been developed to assess airborne particles such as pollen grains or pollutants in the environment (Paatero et al. 2005; Mitsumoto et al. 2010), quantifying bacteria by flow cytometry in the aquatic environment and detecting DNA phosphorylation based on single nanoparticle counting (Amann et al. 1990; Ma and Yeung 2010). In microbial surface sampling, direct particle counting could provide information on the interaction (e.g. adhesion) of microbes with the surrounding surface (e.g. topography and chemistry) as well as to assess potential aggregation between particles resulting in a more accurate measurement of the target, especially when dealing with low concentration of organism.

The current study introduces a mass balance concept, hereafter named recovery balance (RB), as a means to identify and estimate sources of loss of biological material during surface sample collection and processing in a sampling protocol. Fluorescence microscopy was used to quantify the spores deposited on the coupon before sampling and the proper use of controls provided a means to assess recovery from each step in the collection to detection process. Our work provides a method to account for errors and system losses as well as contributes to a better understanding of the mechanisms that can affect microbial sample collection efficiency fundamental to the development of more efficient protocols.

Materials and methods

Bacillus anthracis Sterne preparation

Suspensions of green fluorescent protein (GFP)-labelled B. anthracis Sterne pAFp8gfp were prepared by growing a uniform lawn of spores on a modified Schaeffer media as reported previously in the study by Da Silva et al. (2011).

Spore working suspension preparation

Suspensions of Banthracis Sterne were prepared at 1 × 106 spores ml−1 in 95% (v/v) ethanol (200 proof; Sigma-Aldrich, Milwaukee, WI, USA) in a microcentrifuge tube with maximum recovery (MCT-150-L-C; Axygen, Union City, CA, USA) to avoid losses due to spore adherence to the tube wall. Following preparation, the suspension was heated to 65°C for 25 min to kill any vegetative cells or germinated spores. The spore concentration was determined by culturing on LB agar (Fisher, Pittsburg, PA, USA), and plates were incubated at 32°C for 24 h. Finally, two spore suspensions were prepared to be used in the experiments, c. 1 × 105 spores ml−1 and c. 0·5 × 105 spores ml−1, both in 95% (v/v) ethanol.

Surface (coupon) cleaning

Three new coupons were used for each experiment (sacrifice, sampled and receiving) (Fig. 1). Prior to use in the sampling experiment, microscope glass slides (Corning, Lowell, MA, USA) and 304 Annealed no. 4 stainless steel (Online Metals, Seattle, WA, USA) coupons, both 7·62 cm × 5·08 cm, were immersed in hot water containing 2% (v/v) nonionic detergent for 2 h (RBS35; Thermo Scientific, Rockford, IL, USA) during which they were scrubbed with gloved hands to help remove particulate matter from the surface. In the case of stainless steel, before soaking in detergent, the coupons were wiped off with a cloth wetted with hexane to remove grease residue left by the protective film. Once scrubbed, all coupons (glass and steel) were rinsed thoroughly in hot tap water followed by a thorough DI water (resistivity, 18 MΩ cm at 25°C, Milli-Q; Millipore Corp., Bedford, MA, USA) rinse. Cleaned slides were immersed in a 5% (v/v) nitric acid solution for 5 min, rinsed thoroughly in DI water and stored in 95% (v/v) ethanol.

Figure 1.

Illustrative description of the slip/peel tester used for wipe sample collection of spores. Coupon A was defined as sacrificed, coupon B as sampled and coupon C as receiving.

Spore deposition

Two-microlitre aliquots of a suspension containing either 0·5 × 105 spores ml−1 or 1 × 105 spores ml−1 were deposited slowly onto the centre of the coupons and, after ethanol evaporation (less than a minute), the coupons were imaged using fluorescence microscopy with a GFP filter set (excitation 470/40, dichroic 495 and emission 525/50) to quantify the number of spores loaded on the coupon surface. The spores were imaged with an Olympus BX51 microscope equipped with a Retiga 4000R CCD camera (Qimaging, BC, Canada) with an exposure time of 600 ms and 5 × gain. The area (1·2 cm2) containing the deposited spores was imaged in a mosaic mapping procedure using an automated translation stage controlled by the Surveyor software (ver.; Objective imaging, UK). Images were collected using a 12-bit grey-level images with a camera resolution of 2048 × 2048 using a 10× objective, MPlan FL N 10×/0·30BD ∞/-/FN 26·5 (Olympus, Center Valley, PA, USA). The spore counting was performed using a macro written in ImageJ (ver. 1.41o, Wayne Rasband; National Institute of Health, Bethesda, MD, USA) that processed the complete collection of tiles for one mosaic using the procedure described in the image processing for spore counting section. Spores left on the coupon surface after wiping were quantified by plate count method.

Image processing for spore counting

Image analysis of each slide resulted in a mosaic comprised of 169 images, which was processed as a virtual stack in ImageJ using the built-in ‘Analyze Particles’ function. The images were converted to binary by specifying a threshold to eliminate the background (less than 10 counts on an 8-bit scale). The particle (spore) parameters used for analysis were area (pixels squared) and circularity. Analysis of the size distribution of spores imaged through a 10× objective on the microscope system described above found that counting particles with areas ranging from 10 pixels2 to 200 pixels2 and circularity greater than 0·6 (1·0 corresponds to a perfect circle) provided the most accurate spore counts compared to the characterized spore suspension (plate counting). The area corresponding to a single spore averaged 59 pixels2 (SD = 35, n = 2127 particles from 10 replicate samples). Establishing parameter ranges was necessary for spore counts to eliminate spurious particles resulting from noise and hot pixels as well as large particles that were not spores or potential aggregates. This range was determined by depositing about 10 spores on a clean slide and measuring them individually to estimate the area associated with a single spore that could be seen on the microscope in phase-contrast and fluorescence modes. Additional comparisons of particle counts obtained using mosaic images of fluorescent beads (1 μm in diameter) demonstrated no significant statistical difference between 40× and 10× lens magnification (P > 0·05).

Sample collection apparatus operation

The primary motivation in using the slip/peel tester was to minimize operator variability (Verkouteren et al. 2008) by controlling the wiping pressure on the coupons during the collection. The device was accommodated inside an environmental chamber (COY products, Grass Lake, MI, USA) for monitoring the relative humidity and temperature. After imaging the spores deposited on the coupon surface, the coupons were kept under relative humidity (RH) of 45% or 75% at 23 ± 2°C for 5 min in the environmental chamber prior to sampling. The loaded coupons were wiped using the slip/peel tester (TL-2200; Imass Inc., Accord, MA, USA), a device used to provide reproducible kinetic frictional force (Fkinetic) during the wiping (Verkouteren et al. 2008) (Fig. 1). Wipes were secured around a 1200-g sled and wetted with 1 ml wetting agent solution. The sled with attached wipe was placed on a sacrifice coupon (a clean coupon on which the sled rested before initiating wiping across the coupon with deposited spores) adjacent and in contact with the coupon being sampled. After the wipe collection of spores, from the coupon, was complete, the sled was stopped on a third coupon denominated ‘receiving’. A precleaned thin polydimethylsiloxane (PDMS) sheet (250 μm; Rogers Corp., Carol Stream, IL, USA) was placed under the coupons to hold them in place during the wipe collection. The PDMS sheet was wiped with 70% (v/v) ethanol. The total travel distance of the sled was 13·26 cm at 15 cm min−1.

Potential cross-contamination was reduced by disinfecting the environmental chamber before and after each experimental day using 10% spor-klenz (v/v) solution (Steris, Mentor, OH, USA) following a SOP. Briefly, a solution of 10% (v/v) spor-klenz was sprayed on the bench surfaces and slip/peel tester working area inside of the chamber and left overnight. The next day the surfaces of the environmental chamber were cleaned with wipes wetted with DI water and 70% (v/v) ethanol. Quality assurance sampling for the environmental control chamber was conducted periodically to check for background spore contamination. Swab sampling according to the CDC Protocol (Hodges et al. 2010) and streaking on Luria broth agar plus 10 μg ml−1 kanamycin for GFP selection were used to measure the level of background contamination. Surface substrate was characterized by a profilometer (Dektak 6M; Veeco, Plainview, NY, USA) to obtain the roughness. One coupon of each type was used in the characterization.

Wipe processing

Three different types of wipes were evaluated for recovery efficiency estimates: woven polyester (Value-Tek; cat. no. VTPNWIR-99; Phoenix, AZ), nonwoven polyester–rayon blend (Kendal Versalon; cat. no. 8042; Tyco Healthgroup LP, Mansfiled, MA, USA) and woven cotton (One; cat. no. 9131; Hermitage Hospital Products, Niantic, CT, USA). The same wipe types were previously used by Da Silva et al. (Da Silva et al. 2011) to evaluate the extraction efficiency of B. anthracis Sterne. In the current work, after spore collection, the wipes were removed from the sled using a sterile tweezers and added into 50-ml conical centrifuge tube (cat. no. UP2261; United Lab. Plastics, St Louis, MO, USA) containing 30 ml of PBST (phosphate-buffered saline, PBS, with 0·04% Tween 80). PBST was the optimal solution for dispersing spores in a previous study by Da Silva et al. (2011). The tube was sonicated for 5 min using a sonicator (Fisher Scientific, Pittsburg, PA) with a frequency of 42 kHz and vortexed for 2 min using a vortex mixer (cat. no. 02215370; Fisher Scientific) (Edmonds et al. 2009). The tube was briefly vortexed a second time for 10 sec before a 10-ml aliquot of the extraction solution was withdrawn and passed through a sterile membrane filter (Nalgene analytical filter, 0·45 μm; Thermo Scientific) attached to a vacuum system. The membrane filter was placed onto LB agar for colony growth and incubated at 32°C for 24 h. The ‘sampled’ and ‘receiving’ coupons were inserted into a sterile specimen container (VWR, Bridgeport, NJ, USA) containing 100 ml of PBST, sonicated for 5 min and vortexed for 2 min (Edmonds et al. 2009) to extract any remaining spores and processed in the same condition as the wipe. The reference control was obtained by directly inoculating the spores into PBST (no presence of wipe) and extracting or depositing the spores on the glass, placing the spiked glass in PBST followed by extraction (Eqn (1)). The reference control was processed in the same way as the samples (i.e. submitted to vortexing and sonication, filtration through membrane filter followed by plate count method). In addition, experimental blanks including clean, new coupons and wipes were selected for processing as negative controls. In all cases, no growth was observed.

Calculation of the numbers of spores

The actual numbers of spores deposited onto surface, Ndep, (spores per area), were calculated by direct particle counting as described in image processing for spore counting section by depositing two-microlitre aliquots of a suspension containing either 0·5 × 105 spores ml−1 or 1 × 105 spores ml−1 onto an area equivalent to 1·2 cm2. The spores extracted from the wipe (Nsoln), extracted from the surface substrate (Nsurf) and directly inoculated into PBST extraction solution (Ntube) were each determined separately by plate count method (membrane filter). For each term (Nsoln, Ntube and Nsurf), the spores were extracted in 30 ml of PBST, in which only 10 ml of the extraction solution was filtered to avoid high number of colonies on the filter. The number of colonies obtained on the membrane filter was multiplied by 3 to reach the initial number of spores. To determine the number of spores remaining on the surface after wiping (Nb+c), all 100 ml of PBST used in the extraction was filtered through the membrane and the number of colonies was counted without further calculation. Thus, the number of spores was expressed by area (fluorescence microscopy) or volume (plate count method).

Experimental approach

Two different studies were performed: (i) the development of the recovery balance (RB) method to determine the potential losses of key steps in the sampling protocol of B. anthracis spores used to estimate the total spore recovery and (ii) the application of RB to assess the relative impact of factors, including surface type, relative humidity and wetting agent, affecting spore collection from glass and stainless steel. The first study was conducted by inoculating 10 replicate glass coupons (7·62 cm × 5·08 cm) with 2 μl of spore suspension (1 × 105 spores ml−1) in an experimental design that included 4 factors: 1 wipe (polyester–rayon), 1 wetting agent (0·04% Tween 80–T80), 1 surface (glass) and 1 relative humidity (45%) at 23 ± 2°C. The second study was designed to examine the relative impact of parameters affecting B. anthracis spore collection from nonporous surface rather than method definition and reproducibility as was the aim in the first study. The following 4 factors were considered for the second study: (i) wetting agent: sterile DI water (H2O), T80, phosphate-buffered saline (PBS) and PBST, (ii) wipe material (polyester–rayon, cotton and polyester), (iii) relative humidity (45 and 75%) and (iv) surface (stainless steel and glass). The experimental design consisted of a full-factorial design (4 × 3 × 2 × 2) resulting in 48 runs with additional selected runs to increased confidence in the generated data. For statistical analysis purpose, sample size was based on the maximum numbers of levels involved in the experimental design for each condition. The factor wetting agent containing four different levels (T80, H2O, PBS and PBST) generated six data points (2% r.h. × 3 wipes), relative humidity (45 and 75%) 12 data points (3 wipes × 4 wetting agents) and wipe (polyester, cotton and polyester–rayon) eight data points (2% r.h. × 4 wetting agent). In both studies, the measured response variable was reported as the number of spores per area (spores cm−2) or colony forming unit per volume (CFU per ml) and converted to percentage mean recovery, the number of extracted spores relative to the number of spores initially inoculated on the coupon surface as quantified by fluorescence microscopy. The spores were deposited on an area of 1·2 cm2, and colony forming unit (CFU) was obtained in 30 ml of PBST. The mean recovery efficiency rate was calculated and the uncertainty associated with the mean was reported as standard deviation (SD) for the precision of the measurement, unless otherwise stated. Experiment controls also were performed. In this case, the number of extracted spores was measured relative to a known number of spores inoculated directly into the extraction solution (PBST) and determined by plate count method (membrane filtration) (Ntube) and inoculating spores directly onto coupon followed by extraction in PBST (Nsurf) (Eqn (1)). All generated data were evaluated by one-way analysis of variance, unless otherwise stated, using the free, public-domain software Dataplot (

Recovery balance (RB) for determining losses in the sampling protocol

Adherence of spores to the centrifuge tube walls during wipe extraction and dilution plating, interactions with the substrate surface, viability losses, spore release from the wipe and losses to the wipe were considered in the recovery balance. Equation (1) defines that the number of spores initially deposited (the input) is equivalent to the number of spores recovered from each step of the surface sampling protocol.

display math(1)


Ndep is the number of spores initially deposited on the coupon surface, Nwipe is the number of spores remaining on the wipe, and Ntube is the number of spores adhered to the tube in a given extraction solution (PBST) and was determined by inoculating the spores directly into the extraction solution. Nb+c is the number of spores remaining on the glass slide surface after wiping. The coupons were designated ‘b’ and ‘c’ for sampled and receiving coupons, respectively (Fig. 1). Nsurf is the number of spores extracted from the surface substrate obtained by placing an inoculated coupon in the extraction solution (Fig. 2). Nsoln is the number of spores extracted from the wipe. Plate count method was used to quantify all terms in the Eqn (1), except the term Ndep, which was quantified by fluorescence microscopy. The terms Ntube and Nsurf were obtained through experimental controls. Nsurf was calculated by the number of spores extracted from the surface substrate in PBST minus the number of spores adhered to the tube wall (Ntube). Spores dispersed in solution can adhere to the tube wall (Ntube). Ntube was determined by subtracting the number of spores recovered from the PBST solution from the known inoculum. Nwipe was calculated arithmetically once the other terms were determined.

Figure 2.

Diagram showing the steps comprising the bacterial surface sample collection procedure developed in this study where Ndep is the number of spores deposited initially on the surface, Nsoln is the number of spores released from the wipe into the solution, Nwipe is the number of spores entrapped or adhered to the wipe, Nb+c is the number of spores left on the surface after wiping, Nsurf is the number of spores removed from the glass in a given extraction solution, and Ntube is the number of spores adhered to the tube in a given solution.

The number of spores also can be converted as percentage of recovery efficiency (%RE) as shown in Eqn (2):

display math(2)

In addition to the variables contributing to the recovery efficiency, the difference in absolute cell count between plating counting (viable spores) and microscopy (fluorescent spores) was evaluated and included in the equation as a correction factor, Cp/m, in order to reconcile the bias (systematic error) between the two techniques as described in Eqn (3):

display math(3)

where Cp/m is the number of spores inoculated into extraction solution and recovered by the plate count method divided by the number of spores deposited onto a surface substrate, Nsurf, and quantified with fluorescence microscopy.

In practice, Ntube and Nsurf terms were independent controls measured in replicate experiments. To properly incorporate the average value for each of the terms, the percentage of recovery efficiency (%RE) for each term is independently calculated according to Eqns (4) and (5).

display math(4)
display math(5)

RB can be expressed as in Eqn (6) once each term in Eqn (1) is converted to percentage recovery efficiency (%RE).

display math(6)


The present work was developed in two distinct studies, recovery balance (RB) method development (first study) and application of RB (second study). In the first study, the focus was on the development of a method to measure the recovery efficiency of B. anthracis spores by targeting the main steps composing the sampling protocol while determining the typical variability (precision) in the measurement. In the second study, the focus was on the applicability of the method in assessing the relative impact of factors affecting the performance of the sampling protocol such as surface type, relative humidity, wetting agent and wipe material as reported in this section.

Study 1: recovery balance results

The recovery balance developed in the first study accounted for potential losses during spore deposition, spore collection from the surface and extraction processing (Fig. 2). Control studies to determine a correction factor (systematic error) for the differences in quantification between plate count and microscopy, Cp/m, found a factor of 0·86 (SD = 0·15, n = 5), which was incorporated in eqn (3). In addition, no statistical difference (P = 0·084) between the number of spores counted by fluorescence microscopy and plate count method (paired t-test analysis) was observed. The counts obtained in both cases were 144 (SD = 15, n = 5) and 122 (SD = 16, n = 5), respectively. Once each of the terms in eqn (1) was determined (Table 1), the overall RB could be calculated according to eqn (6). The result associated with each variable was obtained using T80, 45% r.h., polyester–rayon wipe and glass coupon. The results from study 1 indicate that the mean spore recovery efficiency for the extraction solution (REsoln) was 57·7% (SD = 8), while the mean percentage of spores left on the glass after wiping (REb+c) was 2·8% (SD = 2·4) (Table 1). The recovery efficiency of spores left associated with the wipe postextraction processing (REwipe) was 39·9% (SD = 21·2). The other two parameters, REtube and REsurf, were measured as experimental controls resulting in 1·2% (SD = 19·6) and 5·8% (SD = 7·1) of mean recovery efficiency for spore losses to the tube wall and to the glass surface, respectively (Table 1). Thus, the overall RB was 107·4% (SD = 30·9), indicating that all terms identified in eqns (1) and (6) of the recovery balance approach accounted for all of the spores in the system.

Table 1. Summary statistics of the spore recovery balance (RB) in the first studya
 Recovery efficiency %b 
REsolnREB+CREwipeControlsRecovery balance %c
  1. a

    Parameters applied in this study were polyester–rayon wipe, glass coupon, T80, 45% r.h. and PBST as extraction solution.

  2. b

    Extraction efficiency (%RE) of spores extracted from the wipe in solution (REsoln), spores remaining on glass after wiping (REb+c) and spores left on the wipe (entrapped or adhered) after extraction processing (REwipe). Controls are percentage of recovery efficiency for spores adhered to the tube (REtube) and spores adhered to the glass surface (REsurf), both in PBST.

  3. c

    It was calculated as a sum of all terms expressed in recovery efficiency.


Several additional measured parameters were evaluated to support Study 1 including the applied force in the slip-peel tester and the size distribution of spores on the glass surface used to estimate deposited spore number, Ndep. The primary motivation in using the slip/peel tester was to provide a reproducible physical wipe sampling that would minimize potential operator variability and provide for more straightforward evaluation of other parameters affecting sampling efficiency. The mean kinetic force applied during the wiping was 6·25 N (SD = 0·96, n = 10), indicating about 15% of variability across different wiping.

Spore counts obtained from the fluorescence microscopy were 212 (SD = 22, n = 10). A comparison with spore counts obtained from plating an experimental control (direct inoculation in the extraction solution) revealed no statistically significant difference between both techniques (P = 0·109, paired t-test).

Study 2: evaluation of factors that impact recovery

In the second study, the applicability of the method to evaluate the relative impact of parameters on the collection of spores from nonporous surfaces (glass and steel) was demonstrated in a full experimental design with wipe, surface, relative humidity and wetting agent as factors. The overall mean spore recovery efficiency from extraction solution (%REsoln) was higher for glass coupon (40%, SD = 27) than for stainless steel (30%, SD = 23) across all factors involved in the experiment by rank analysis. However, no statistical difference (P < 0·05) was observed between both surfaces (Table 2). It was also observed that on average 1% (SD = 2) and 3% (SD = 3) of the spores were left on the steel and glass surface after wipe sampling (%REb+c). A closer analysis into the mean recovery rate of each factor showed that T80 (56%, SD = 32), 45% r.h. (44%, SD = 30) provided the highest mean recovery for wetting agent and relative humidity, respectively, while cotton (50%, SD = 22) and polyester–rayon (48%, SD = 29) had a similar mean recovery by rank analysis. Among all factors, a statistically significant difference in mean recovery was observed only among the four wetting agents (P = 0·031) (Table 2).

Table 2. Percentage of mean Bacillus anthracis spore recovery efficiency (%REsoln)
 Surface (P = 0·134)
Glass % REsolnaSteel %REsolna
Mean (SD)Sample sizeMean (SD)Sample size
  1. a

    Mean (SD) recovery efficiency for spores released from the wipe into extraction solution. Standard deviation described in parentheses.

  2. b

    Among all factors, an one-way analysis of variance (anova) indicated a statistically significant difference in mean recovery only among the four wetting agents (P = 0·031). In addition, a four-way anova also yielded wetting agent as the only statistical significant factor.

  3. c

    Mean (SD) recovery efficiency for spores remaining onto surface after wiping.

Wetting agentb (P = 0·031)
T8056 (32)836 (20)6
PBST45 (14)538 (21)8
PBS30 (25)820 (34)5
H2O24 (16)422 (18)7
%r.h. (P = 0·766)
4544 (30)1327 (23)12
7536 (22)1232 (25)14
Wipe (P = 0·174)
Cotton50 (22)1023 (25)9
Polyester–rayon48 (29)639 (21)7
Polyester24 (25)929 (23)10
Overall recovery efficiency (%RE) – surface
Mean %REsolna40 (27)2530 (23)26
Mean %REb+cc3 (3)151 (2)18


The work described in this manuscript provides a mean to account for spore losses to the sample collection and processing while evaluating the robustness of a wipe method performance. The RB method is meant to be applicable only in a laboratory setting during the development of a protocol and not meant to be a replacement for field or laboratory validation of wipe recoveries. The recovery balance (RB) method developed in Study 1 utilizes fluorescence microscopy and traditional microbiological techniques to better understand losses to the analytical process that cannot be easily accounted for otherwise. Furthermore, application of a slip/peel tester and relative humidity control during spore recovery from glass and stainless steel in Study 2 is a demonstration of the wipe recovery balance for different experimental factors of interest to the field collections community. In the first study, the conditions applied for sample collection from glass coupons (T80, 45% r.h. and polyester–rayon wipe) led to a high percentage of spores (REwipe of 39·9%, SD = 21·2) remaining on the wipe postextraction processing (Table 1). Losses of spores to the tube, REtube, and glass coupon, REsurf, were low at 1·2% (SD = 19·6) and 5·8% (SD = 7·1), respectively (Table 1), resulting in a combined loss of c. 47% when these three terms were added. An evaluation of a swab protocol for the recovery of B. anthracis from stainless steel, reported in the literature, involving five analysts from two laboratories confirmed that 40% of spores were not recovered (Hodges et al. 2006). Also consistent with this report, Hodges et al. found that additional extraction/vortexing processing recovered an additional 6·2% spores released from the swab, while about 6·1% were left on the coupon surface after wiping. Based on Hodges et al.'s results, extraction from the wipe may be improved by the addition of a second extraction step in the current study. Hodges et al. postulated upon plate count analysis that the remaining spores may be lost during the sample processing step, trapped inside the swab or remaining on the coupon surface. This current work provides a means to attribute the lost spores with the adsorptive material and surface substrate.

The contribution of REsurf and REtube are important controls for characterizing and interpreting the recovery efficiency results. Both measurements are directly influenced by the characteristics of the extraction solution (e.g. ionic strength, adhesion and pH) and the coupon topography, respectively (Wilson et al. 2001; Ubbink and Schar-Zammaretti 2007; Seale et al. 2008; Verran et al. 2008). In the current study, both terms, spores extracted directly from glass slides (REsurf) and also inoculated into the extraction solution (REtube), both as experiment control, had a small impact (5·8%, SD = 7·1 and 1·2%, SD = 19·6, respectively) on the overall RB, and observed losses were attributed to entrapment or adherence to the wipe material (REwipe) (Table 1). The high recoveries from the tube and surface substrate (REsurf and REtube) were likely due to the use of PBST, a highly efficient extraction solution (Da Silva et al. 2011; Rose et al. 2011) and a smooth glass coupon (Edmonds et al. 2009). Increasing vortexing and sonication processing times may improve the release of spores from the wipe as demonstrated previously by Hodges et al., in which a second vortexing processing increased the recovery of B. anthracis from macrofoam swabs in 6%. (Hodges et al. 2006). To elaborate an effective sampling protocol, an evaluation of the impact of experimental conditions on the overall recovery should be considered. Downey et al. (2012) studying the relative impact of variables on the extraction processing of vegetative cells (Bacillus cereus, Bacillus thailandensis and Escherichia coli) observed that the organism type provided the largest impact on the extraction efficiency performance across all three organisms. Moreover, the effect of the extraction solution was organism dependent. Da Silva et al. (2011) observed that the most important factor driving the extraction performance of B. anthracis Sterne inoculated directly on wipes was extraction solution.

Study 2 provided the applicability of RB by evaluating the relative impact of factors affecting the overall wipe collection performance. The factors investigated were surface substrate (glass and steel), wipe material (polyester–rayon, polyester and cotton), relative humidity (45 and 75%) and wetting agent (T80, H2O, PBS and PBST). The mean recovery from the wipe (REsoln) obtained with T80 wetting agent using glass coupons was consistent in both studies: Study 1 RB method development, 58%, SD = 8, and Study 2 RB method application, 56%, SD = 32 (Tables 1 and 2). Considering the full-factorial experiment design, T80, 45% RH, polyester–rayon and cotton wipes had the largest performance by rank analysis. However, statistically significant differences were observed only among the four wetting agents (P = 0·031) (Fig. 2). Due to the number of samples and factors involved for each condition in the experimental design, more replicates would be required in order to draw a robust conclusion about the contribution of each factor to the overall recovery of spores. However, it is known that some of these variables affect microbial surface sampling performance (Da Silva et al. 2011; Downey et al. 2012). Variability imposed by surface type (Rose et al. 2004; Vorst et al. 2004; Seale et al. 2008; Simoes et al. 2008) can be attributed to potential adhesion of spores to surfaces by van der Waals, electrostatic and capillary forces (Van Oss 1994; Hermansson 1999; Wilson et al. 2001; Ubbink and Schar-Zammaretti 2007) as well as a potential physical interaction due to the degree of surface roughness (Verran et al. 2008). Relative humidity (RH) is an important factor in successful microbial inactivation (e.g. decontamination building) (Hubbard et al. 2009) and strongly influences airborne cultivability (Rule et al. 2009). Therefore, the potential impact of RH on sample collection should be investigated further.

Recommended wetting agents and wipe materials common to environmental sampling protocols are a result of measurements to optimize for sampling efficiencies in a range of diverse environments. The results of both Studies 1 and 2 indicate that a closer analysis of the interaction between adsorbent material (wipe) and spores should be conducted to assess the contribution to losses inherent to the fibre characteristics. All wetting agents utilized in this work containing Tween 80 (T80 and PBST) resulted in the best recovery efficiency (REsoln) when compared with the other wetting agents by rank analysis (Table 2). Improved recovery due to the addition of Tween 80 in wetting agent and extraction solutions has been associated with high environmental sampling performance (Kim et al. 2008; Da Silva et al. 2011). In the current study, spores were likely entrapped in the wipe fibres contributing to the low release of spores from the wipes. The potential interaction between adsorptive material and spores, and the impact on sampling performance, has been investigated with other types of materials such as cotton fibres, polyester, macrofoam (Rose et al. 2004; Probst et al. 2010). Adhesion of bacteria and spores in the adsorbent material (e.g. swab and wipe) fibres is a result of physical and chemical interactions including wettability, organism size and fibre chemical composition, and assembly (Hasan et al. 2008). It has been reported extraction efficiencies close to 100% when B. anthracis spores were directly inoculated on wipes including polyester–rayon (Rose et al. 2004; Da Silva et al. 2011). In contrast, the additional physical forces applied during collection from the surface in this study may have contributed to entrapment and entanglement of the spores in the wipes.

Physical and chemical interactions as well as surface roughness can impact mean recoveries from surfaces (Rose et al. 2003; Brown and Jaffe 2006; Krauter et al. 2012). The mean recovery of spores (REsoln) obtained from glass and steel was found to be 40% (SD = 27) and 30% (SD = 23), respectively (Table 2). The proper use of controls enables accurate determination that the majority of spores were not recovered from the wipe during extraction processing. Furthermore, the number of spores left on the surface after wiping, REb+c, was low with mean values of 4%, SD = 3 (glass) and 1%, SD = 2 (steel). Both surfaces are considered nonporous; however, the average roughness (Ra) was nearly 100 times greater for stainless steel (Ra = 0·16 μm) when compared to glass (Ra = 0·0018 μm). Spores were observed to accumulate in the grooves of the steel surface (data not shown) during deposition. In a recent study by Krauter et al. (2012), a linear dependence of recovery efficiency of B. atrophaeus on surface roughness was observed. Because roughness and other physical or chemical parameters clearly impact recovery efficiencies, it would be more representative to report independently measured surface properties as well as a description of the macroscopic characteristic (e.g. porous, nonporous, wallboard, ceramic tile, steel) in validation and performance evaluation studies.

Comparisons of recovery efficiencies in the literature is not a straightforward task as many of the studies utilize different experimental conditions and vary the factors that may impact performance but do not report out controls for those factors. The novelty of this work is it provides independent estimates of the performance of each step of the surface sample collection process and is a means to assess how factors contribute to the overall method robustness. The use of digital imaging constitutes a valuable tool to gain insight into how biological materials adhere to surfaces while providing a better understanding of the parameters affecting sampling efficiency independent of operator contributions. Enabling independent process evaluation steps promotes the development of new methodologies and supports method optimization in a laboratory setting that can be more effectively applied in field study applications. Additionally, knowing the recovery capability of the sampling method before use in the field can improve confidence in the data generated by the sample collection and extraction process in real-case situations.


The Department of Homeland and Security (DHS) Science and Technology Directorate sponsored the production of this material under Interagency Agreement HSHQDC-09-X-00457 with the National Institute of Standards and Technology (NIST). We thank Greg Gillen and Jennifer Verkouteren for providing insightful information and help on the development of the slip/peel tester method for biological application. We thank Barb Jones for supporting the experiment to be envisioned a number of years ago.


Certain commercial equipment, instruments or materials are identified in this paper in order to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.