Diagnostic Validity of Semiquantitative Swab Cultures

Disclosure: This study was funded by The Department of Veteran’s Affairs, Health Services Research and Development, Nursing Research Initiative (NRI-01-005-1), the Gerontological Nursing Interventions Research Center, the Hartford Center for Geriatric Nursing Excellence, and the John A. Hartford Foundation. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Department of Veteran’s Affairs or the National Institute of Nursing Research.

The swab is the most practical and widely available method for obtaining wound specimen culture. However, the validity of the traditional swab specimen has been questioned, since it fails to capture the level of bacteria beneath the highly colonized wound surface. Recently, it was demonstrated that swab specimens obtained using Levine’s technique¹ and processed in the laboratory using quantitative procedures provided a reasonably accurate culture compared with that of wound tissue specimens processed in the same manner.² Using Levine’s technique, swab specimens were collected by rotating a swab culture over a 1 cm² area of the wound with sufficient pressure to extract fluid from within the wound tissue. The swab specimen was taken from an area over viable wound tissue after being cleansed with saline. These specimens were then processed in the microbiology laboratory using quantitative procedures that allowed quantification of organisms per swab. The results from the swab cultures were then compared to the results from quantitative wound tissue cultures, which are considered the gold standard for assessing wound bioburden.
Although the findings from the present study hold promise of a more practical and valid method of monitoring wound bioburden than obtaining and processing wound tissue specimens, several clinical practice questions remain unanswered. The laboratory procedures necessary to process quantitative swab cultures are more complex and expensive ($150/culture) than procedures to process semiquantitative swab cultures ($50/culture). However, the preciseness of the microbiological information provided by semiquantitative processing is far less than that provided by quantitative processes, and it is unclear how the results of semiquantitative cultures should be interpreted. One study³ examined the correlation between quantitative swab cultures and semiquantitative swab cultures, both obtained using Levine’s technique. The findings of this study indicate a high correlation between the 2 processes in identifying chronic wound infection. A major limitation of this study was that the semiquantitative swab was not compared to the gold standard (ie, quantitative culture of wound tissue). Evaluation of semiquantitative swab cultures in relation to the gold standard would establish the optimum semiquantitative value to use in interpreting these culture findings and provide a reference standard for comparing semiquantitative to quantitative swab cultures with respect to accuracy, practicality, and cost in the clinical setting. The present study examined the diagnostic validity of semiquantitative swab cultures obtained using Levine’s technique and compared semiquantitative and quantitative swab cultures. The research questions were:
1. What is the accuracy of semiquantitative swab cultures obtained using Levine’s technique as compared to quantitative cultures of viable wound tissue (reference standard) among a sample of infected and noninfected, nonarterial, chronic wounds?
2. What is the concordance between semiquantitative swab cultures obtained using Levine’s technique and quantitative tissue cultures with respect to the simultaneous recovery of organisms?
3. Are there differences in the accuracy and concordance between semiquantitative swab cultures obtained using Levine’s technique and quantitative swab cultures obtained using Levine’s technique?
Swab cultures are noninvasive and most laboratories are capable of semiquantitatively processing these specimens. Examining the diagnostic validity of the semiquantitative swab culture would establish their clinical utility as an alternative to quantitative processes. This knowledge would significantly improve the diagnostic tools available to practitioners who treat chronic wounds in settings where quantitative cultures are inaccessible.

Materials and Methods

An observational, cross-sectional design was used. Two swab specimens were obtained from a sample of chronic wounds using Levine’s technique. One was processed using quantitative laboratory procedures and the other using semiquantitative laboratory procedures. The diagnostic validity of the findings from each swab culture process was determined by associating the culture findings of each with quantitative tissue cultures (reference standard).
Setting and sample. The study population comprised patients with nonarterial chronic wounds. A Department of Veteran’s Affairs Medical Center (VAMC) and a university-affiliated tertiary hospital were the study settings. The Institutional Review Board (IRB) that oversees both facilities approved study procedures prior to screening and enrolling subjects. Subjects were screened and enrolled based on the following criteria: 1) age 18 years or older; 2) presence of a full-thickness, nonarterial chronic wound; 3) white blood cell count (WBC) > 1500 cells/mm³ or total lymphocyte count > 800 cells/mm³; 4) platelet count > 125,000/mm³; 5) no coagulopathies and; 6) not receiving anticoagulation therapy. Subjects gave written consent to participate in the study. The sample was restricted to nonarterial chronic wounds in order to decrease the risk of infection imposed by tissue biopsy in wounds with poor perfusion. The sample was restricted to full-thickness chronic wounds in order to justify acquisition of full-thickness tissue specimens. Patients with low white cell counts were excluded to reduce the risk of wound infection related to tissue biopsies. Patients with low platelet counts, coagulopathies, or undergoing anticoagulation therapy were excluded to prevent the risk of bleeding associated with wound biopsy. One wound was randomly selected when subjects had more than 1 eligible wound.
Study variables. The primary study variables were culture findings based on: 1) swab specimens obtained using Levine’s technique and processed using semiquantitative procedures; 2) quantitative swab specimens obtained using Levine’s technique and processed using quantitative procedures; and 3) cultures of wound tissue specimens processed using quantitative procedures (reference standard). Swab and tissue specimens were processed at a single microbiology laboratory dedicated to research studies.
The laboratory procedure for culturing tissue specimens was comparable to that suggested by Krizek and Robson.⁴ Tissue specimens were weighed, homogenized, and serially diluted in tryptic soy broth (TSB, Remel Inc, Lenexa, Kan). Each dilution was plated onto Columbia blood agar (Remel Inc, Lenexa, Kan), BBL™ CHROMagar™ Candida (BD, Sparks, Md), eosin-methylene blue agar (Remel Inc, Lenexa, Kan), and reduced agar media. Reduced agar plates were placed in an anaerobic chamber and incubated in ambient air at 37˚C for 48 hours. CHROMagar Candida plates were placed in an aerobic chamber and incubated at 30˚C for optimal yeast growth. All other plates were incubated under aerobic conditions in 5% CO₂ at 37˚C for 48 hours for optimal growth of Streptococcus and other fastidious organisms. To provide qualitative culture data, organisms were identified using standard microbiologic procedures based on criteria, such as colony morphology and gram stain appearance.⁵ For example, Staphylococcus aureus was identified based on characteristic yellow B-hemolytic colonies on Columbia blood agar, which on stain appeared as gram-positive cocci organized into grape-like clusters and tested catalase- and Staphylococcus- latex positive. Streptococci were identified to Lancefield group (A, B, C, G, and F) by agglutination with appropriate antisera using the PathoDx Kit (Remel Inc, Lenexa, Kan). To provide quantitative culture data, each identified organism was quantified by counting the number of colony forming units (CFU) on each plate. The plate count multiplied by the dilution factor yielded the number of organisms per gram of tissue because tissue specimens were based on tissue weight. The “true” infection status of the wound was based on the quantitative culture findings from tissue specimens. Infected wounds were defined as those with 1,000,000 (ie, 10⁶) or more organisms per gram of viable tissue.⁶ Noninfected wounds were defined as those with less than 1,000,000 organisms per gram of viable wound tissue.
Swab specimens to be quantitatively processed were placed in 1 mL of sterile saline and vortexed for 15 seconds. They were then serially diluted in TSB and each dilution was plated onto Columbia blood agar, CHROMagar Candida, and eosin-methylene blue agar. CHROMagar Candida plates were placed in an aerobic chamber and incubated at 30˚C for optimal yeast growth. All other plates were incubated under aerobic conditions in 5% CO₂ at 37˚C for 48 hours for optimal growth of Streptococcus and other fastidious organisms. Swab specimens were not incubated under anaerobic conditions because swab specimens are believed to be unsuitable for anaerobic culture.⁵ To provide qualitative culture data, isolated organisms were identified as described above. Quantification of each organism was accomplished by counting the number of CFU on each plate. Because dilutions are based on 1 swab, the plate count times the dilution factor yielded total number of organisms per swab.
Those swab specimens to be semiquantitatively processed were plated on blood agar using the dilution streak technique. The first quadrant of the plate was streaked using the swab and each successive quadrant was streaked using a new bacteriologic loop in order to dilute the number of bacteria in each quadrant. The plates were incubated under aerobic conditions. Quantification was expressed as 1+, 2+, 3+, or 4+ based on the number of quadrants that demonstrated bacterial growth. Bacterial growth limited to quadrant 1 was categorized as 1+, bacterial growth limited to quadrants 1 and 2 was categorized as 2+, bacterial growth limited to quadrants 1, 2, and 3 was categorized as 3+, and bacterial growth that extended to all 4 quadrants was categorized as 4+. Organisms were identified using standard microbiologic procedures.⁵ Streptococci were identified to Lancefield type (A, B, C, G, and F) by agglutination with appropriate antisera using the PathoDx kit.
Secondary study variables were also measured as a part of study procedures in order to describe the study sample. These variables include age, gender, race, diabetes, systemic antibiotic therapy, red blood cell count (RBC), white blood cell count (WBC), albumin level, Hemoglobin A1c (HbA1c) levels for subjects with diabetes, type of chronic wound, size of the wound, depth of the wound, and duration of study ulcer.
Data collection. Demographic and medical history data were collected from the patient record and patient/caregiver report. Data regarding type and number of chronic wounds were collected from direct observation. Blood samples were drawn for laboratory values.
After removing the dressing, an area near the center of the wound was cleansed with nonbacteriostatic saline and the end of a Culturette® swab was rotated over a 1 cm² area for 5 seconds (Levine’s swab for quantitative procedures). Pressure was applied to the swab for 5 seconds in order to express fluid from within the wound tissue. A second Culturette swab was rotated over the same 1 cm² area as the first swab (Levine’s swab for semiquantitative procedures). Pressure was applied to the second swab for 5 seconds in order to express fluid from within the wound tissue. Cleansing the wound prior to obtaining swab specimens was done in an effort to remove surface organisms immediately prior to obtaining each specimen.
Wound tissue biopsy (tissue specimen) was then performed. A specimen of viable wound tissue was removed from under the area of the wound sampled with the swabs using a 4-mm to 6-mm dermal punch instrument. Sterile technique was employed during the procedure. The specimen was placed in a sterile container labeled as “wound tissue”.
Swab and tissue specimens were then transported to the microbiology research laboratory for processing. Laboratory technicians were kept blind to the study aims and study procedures. Through this, observer bias in laboratory identification/quantification associated with knowing the results of related culture specimens was avoided. All specimens were processed within 2 hours of acquisition in order to minimize alterations in number of organisms related to progression of time. Culture media and isolation criteria were standardized for all specimens in order to decrease bias associated with differences in laboratory methodology.
Wound size was determined by tracing the wound margin on transparent film with an indelible marker. Tracings were labeled with subject identification number and date. The depth of the wound was measured using a cotton tipped swab placed in the deepest portion of the wound. The point on the swab that was level with the periwound skin was marked and measured with a centimeter ruler. A photographic image of the wound was taken with a digital camera.

Statistical Analysis

Subject and wound characteristics. Subject and wound data were entered into electronic databases. Data were double entered to reduce entry errors and cleaned using range and consistency checks. The subject and wound databases were used to compute summary statistics of the subject and wound sample and to compute inferential statistics for examining differences between the infected and noninfected groups. Nominal variables were statistically described with percentages and statistically compared between the infected and noninfected groups using a chi-squared test for independence or Fisher’s exact test. Continuous variables were described with means and standard deviations and statistically compared between the infected and noninfected groups using the nonparametric Wilcoxon rank-sum test. A 2-tailed alpha level of .05 was employed to identify significant differences between the high and low microbial load wounds.
Accuracy of semiquantitative swab cultures. To address the first research question, a receiver operating characteristic (ROC) curve computed using the nonparametric trapezoidal approach⁷ was used to evaluate the semiquantitative culture. An ROC curve is a plot of sensitivity versus 1-specificity for each value of the semiquantitative swab and hence, is a graphical way of displaying the complete performance of the swab culture at all possible cut values (ie, +1, +2, +3, and +4).
The area under the ROC curve (AUC) provides a summary measure of accuracy for the semiquantitative swab culture. It estimates the accuracy rate for discriminating between infected and noninfected wounds,⁷ defined as the probability that a randomly selected patient with an infected wound (ie, < 1,000,000 organisms per gram of tissue) will have a higher outcome (ie, +1, +2, +3, or +4) than a randomly selected patient with a noninfected wound. The AUC for an ROC curve is always between 0 and 1. A noninformative test will have an ROC curve that lies along the diagonal and hence, have an AUC of approximately 0.5, while a very informative test will have a ROC curve close to the upper left corner and an AUC close to 1. Confidence intervals (95%) were computed for the AUC and the null hypothesis (HO: AUC = 0.5) was statistically tested (alpha level .05, 2-tailed) using the algorithm described by DeLong et al.⁸ Computations were performed using STATA® 8.1 for Windows®.
Concordance of semiquantitative swab cultures. To address the second research question, recovery of specific organisms was compared between the tissue and swab specimens. First, recovery of organisms was described for each specimen by calculating the mean number of different species isolated per wound and the frequency with which the organisms, Staphylococcus aureus, Pseudomonas aeruginosa, and Group A and Group B Streptococci were recovered. The frequency with which anaerobes were recovered from tissue specimens was also tabulated. The recovery of anaerobes from swab specimens could not be tabulated because swab specimens were not incubated under anaerobic conditions.
Second, concordance between the tissue and swab specimen was examined. For each study wound, the concordance in recovering all organisms was determined using the following equation⁹:

number of specific isolates
simultaneously recovered from
the tissue and the swab specimen x100
total number of isolates collected by either specimen

The mean concordance was then calculated. In addition, the concordance between the tissue and swab specimen in recovering specifically S aureus, P aeruginosa, and Group A and Group B Streptococci was computed as total (ie, total number of concordant observations/total number of paired observations), occurrence (ie, number of positive concordant observations/total number of wounds in which the specific organism was recovered from either the tissue or swab specimen), and nonoccurrence (ie, number of negative concordant observations/total number of wounds in which the specific organism was absent from either the tissue or swab specimen) concordance. Three measures of concordance were calculated because estimates are influenced by the manner in which concordance is defined and total concordance can be inflated by a high percentage of nonoccurrence agreements when few occurrence agreements occur.
Comparison of semiquantitative and quantitative swab cultures. To address the third research question, the statistical procedures detailed under research question 1 and 2 were repeated for the quantitative swab cultures. The equality of the AUC was then tested from the semiquantitative swab cultures with the AUC from the quantitative swab cultures using a pairwise comparison. These tests were performed using the algorithm described by DeLong et al.⁸ The test was 2-tailed with alpha = .05. Differences in concordance between semiquantitative swab cultures and quantitative swab cultures were examined through direct comparison.

Results

Subject and wound characteristics. Forty-four subjects participated in the study. Thirty-five (79.5%) of the subjects were men and 43 (97.7%) were white. Descriptive statistics of subject characteristics are presented in Table 1 for the total of infected and noninfected groups. Comparison of the infected and noninfected groups revealed no significant differences in age, gender, race, diabetes status, systemic antibiotic treatment, RBC, WBC, albumin level, and HbA1C levels between the infected and noninfected groups.
Wound characteristics for the total group, infected, and noninfected group are summarized in Table 2. Of the 44 wounds, 15 (34%) were infected. The diabetic foot ulcer was the most common type of wound. Comparisons of the infected and noninfected groups showed no significant difference in wound type, wound size, wound depth, or wound duration between the 2 groups.
Accuracy of semiquantitative swab cultures. The ROC curve for the semiquantitative swab cultures is presented in Figure 1. The AUC is presented in Table 3. The AUC of the semiquantitative swab culture was 0.0639, which was not significantly higher than the diagonal chance line of .50 (ie, diagonal line on Figure 1; P = 0.0501), suggesting a noninformative test.
Concordance of semiquantitative swab cultures. The mean number of organisms recovered per specimen and the frequency with which specific organisms (ie, S aureus, P aeruginosa, Group A and B Streptococci, and anaerobes) were recovered are presented in Table 4. The mean number of organisms recovered from the semiquantitative swab cultures was less than the mean number recovered from tissue cultures. Concordance between semiquantitative swab and tissue cultures with regard to recovering organisms is presented in Table 5. The mean concordance of semiquantitative cultures in recovering all organisms was only 57%. For S aureus, total occurrence and nonoccurrence concordance ranged from 25 to 45%. Occurrence concordance for P aeruginosa and Group A and B Streptococci was 0% while nonoccurrence concordance was 84% and 95%, respectively.
Comparison of semiquantitative and quantitative swab cultures. The AUC for quantitative swab cultures was .821. The ROC curve for quantitative swab cultures is also presented in Figure 1. The AUC of the quantitative swab culture was significantly higher than the diagonal chance line of 0.5 (P = 0.0002) and is shown in Table 3. In addition, it was significantly higher than the AUC of the semiquantitative swab culture (P = 0.0128). The mean number of species per specimen and the frequency with which specific organisms were recovered from the quantitative swab cultures is presented in Table 4. The mean number of organisms recovered from quantitative swab cultures was more comparable to tissue cultures than semiquantitative swab cultures were. The concordance of quantitative swab cultures with tissue cultures is presented in Table 5. The mean concordance of quantitative swab cultures in recovering all organisms was 72% and substantially higher than the mean concordance of semiquantitative swabs. Concordance of quantitative swab cultures in recovering S aureus ranged from 80% to 93%, which is much higher than semiquantitative swab cultures. Like semiquantitative swab cultures, occurrence concordance in recovering P aeruginosa and Group A and Group B Streptococci was 0%. However, nonoccurrence concordance in recovering these organisms was 100%, again higher than semiquantitative swab cultures.

Discussion

The findings of this study suggest that swab specimens processed using semiquantitative processes do not provide culture findings that correlate well with culture findings from tissue specimens. This finding is true for both the semiquantitative results (ie, 1+, 2+, 3+, or 4+) and the qualitative results (ie, type of organisms recovered). Moreover, semiquantitative swab specimens did not perform as well as swab specimens processed using quantitative procedures when both are compared to cultures of tissue specimens. The reason for these findings may be that semiquantitative processing is not able to isolate and grow all organisms since all organisms are competing on the same media plate. Also, semiquantitative methods do not provide precise information regarding number of organisms. These limitations do not occur when specimens are processed using quantitative methods.
The findings of this study have important implications for practice. Use of a semiquantitative swab culture is of less value in guiding wound care decisions than a quantitative swab culture, since the semiquantitative swab provides less accurate information regarding the true bacterial burden of the wound tissue. However, care must be taken to collect swab specimens in a manner that insures the acquisition of microbes from within the wound tissue such as that provided with Levine’s technique.

Clinical Editor’s Note:

The reader is referred to the article, (Berg JO, Mossner BK, Skov MN, Lauridsen J, Gottrup F, Kolmos HJ. Antibacterial properties of EMLA and lidocaine in wound tissue biopsies for culturing. Wound Repair Regen. 2006;14(5):581–585), which outlines the antibacterial effect of anesthetics commonly used by wound care practitioners. EMLA, a topical anesthetic, has been found to have a high antibacterial effect against most common wound bacteria and is not recommended for use prior to collecting swab or biopsy culture specimens from wounds. Preservative-free 1% lidocaine also has an antibacterial effect on wound bacteria, but it appears to be minimal when the specimen is collected within 2 hours of use of the anesthetic. It is for this reason that the authors recommend preservative-free 1% lidocaine for wound anesthesia if the specimen for culture is collected within 2 hours of anesthetic use. It appears that wound culture results may be misleading if these recommendations are not followed.