Background: Propionibacterium acnes is a common pathogen identified in postoperative shoulder infection. It has been shown to be present in culture specimens during primary shoulder arthroplasty; however, recent work has suggested that it is most likely to be a contaminant. Our aim was to identify the potential sources of contamination in shoulder arthroplasty.
Methods: Tissue swabs were obtained for microbiological analysis from consecutive patients undergoing primary shoulder arthroplasty. Routine surgical technique was maintained, and 5 specimens were taken from different sites: (1) the subdermal layer, (2) the tip of the surgeon’s glove, (3) the inside scalpel blade (used for deeper incision), (4) the forceps, and (5) the outside scalpel blade (used for the skin incision).
Results: Forty patients (25 female patients and 15 male patients) were included. Thirteen (33%) of the 40 patients had at least 1 culture specimen positive for P. acnes. Two (8%) of the 25 female patients and 11 (73%) of the 15 male patients had ≥1 culture specimen positive for P. acnes. The most common site of growth of P. acnes was the subdermal layer (12 positive samples), followed by the forceps (7 positive samples), the tip of the surgeon’s glove (7 positive samples), the outside scalpel blade (4 positive samples), and the inside scalpel blade (1 positive sample). There were 27 of 75 swabs that were positive on culture for P. acnes in male patients compared with 4 of 125 swabs in female patients. Male patients had 66 times (95% confidence interval, 6 to 680 times) higher odds of having a positive culture indicating subdermal colonization compared with female patients (p < 0.001).
Conclusions: P. acnes is a common contaminant of the surgical field in primary shoulder arthroplasty. The subdermal layer may be the source of this contamination, and the prevalence of P. acnes in the surgical wound may be due to the surgeon’s manipulation with gloves and instruments. Our findings are consistent with those regarding the increased rates of P. acnes bacterial load and intraoperative growth in male patients compared with female patients.
Clinical Relevance: P. acnes is likely to be spread throughout the surgical field from the subdermal layer via soft-tissue handling by the surgeon and instruments. Strategies need to be utilized to minimize this contact and to reduce the chance of colonization.
Once thought to be a contaminant, Propionibacterium acnes is now recognized as a pathogen following a shoulder surgical procedure1-3. The indolent nature of P. acnes infections can make the diagnosis of P. acnes periprosthetic joint infection challenging4. It is a frequently isolated causative agent in postoperative shoulder infection3; however, the usual prophylactic antibiotics for a shoulder surgical procedure, first-generation cephalosporins, may not be the most appropriate agents for prevention of P. acnes infection5.
P. acnes is an aero-tolerant, anaerobic, gram-positive bacillus that is associated with sebum-producing hair follicles found within the dermis6,7. It colonizes male patients more frequently than female patients because of the increased number of hair follicles, and it more commonly involves the shoulder compared with the hip and knee1,8,9.
A recent study by Levy et al.10 isolated P. acnes from the synovial fluid in 41.5% of shoulders about to undergo shoulder replacement and suggested that P. acnes may be a causative organism for shoulder osteoarthritis. This was further investigated by Maccioni et al. using the Oxford technique for specimen collection11 and demonstrated that only 3.1% of capsule tissue specimens taken at the time of the surgical procedure were culture-positive12. This discrepancy in the rate of culture of P. acnes is unexplained, and the difference may relate to field contamination, contamination and cross-contamination of specimens, or timing of prophylactic antibiotics.
The work performed by Maccioni et al.12 demonstrated that the rate of culture of P. acnes from capsular tissue with strict collection protocols is low. To further this work, we sought to ascertain the likely source of P. acnes contamination by examining commonly used instruments and sites within the surgical field. Presuming that the source of this contamination is the exposed subdermal layer, we hypothesized that this site would have the highest rate of positive culture specimens. We also hypothesized that instruments used to manipulate the skin and subdermal layer would similarly have high rates of P. acnes contamination. Lastly, by identifying any further sites that are frequently contaminated with P. acnes, we could potentially develop strategies to minimize this contamination.
Materials and Methods
In this prospective case series, patients undergoing a primary total shoulder replacement for a primary diagnosis of osteoarthritis were recruited to have 5 swabs taken for microbiological testing from different surgical sites. Two experienced shoulder arthroplasty surgeons performed all procedures. The patients were recruited over a 10-month period (August 2014 to May 2015) with institutional human ethics approval.
The exclusion criteria were a previous shoulder surgical procedure, shoulder infection, previous proximal humeral fracture, use of antibiotics within 2 weeks prior to the surgical procedure, and cortisone injection into the shoulder within 6 months prior to the procedure.
Forty patients with a mean age of 74.3 years (median, 73 years [range, 46 to 92 years]) were included in the study (25 female patients and 15 male patients) (Table I). No patients declined participation. All patients had a primary diagnosis of glenohumeral arthritis, with the cause being primary osteoarthritis in 36 patients and rotator cuff tear arthropathy in 4 patients. Fifteen of the 40 shoulders included were right shoulders and 25 were left shoulders. A reverse total shoulder arthroplasty was performed in 22 patients (those with cuff tear arthropathy or those with idiopathic arthritis and an insufficient rotator cuff), and an anatomical replacement was used in 18 cases. All patients had 1 swab culture obtained for microbiological examination from 5 different sites at the time of the surgical procedure, including (1) the subdermal layer, (2) the tip of the surgeon’s glove, (3) the scalpel blade used for deeper incision (“inside” scalpel blade), (4) the forceps, and (5) the scalpel blade used for the skin incision (“outside” scalpel blade) (Table I).
There were no changes made to the standard surgical technique used in our institution. Patients did not shave or use topical skin treatments prior to the surgical procedure. All patients received a general anesthetic and 2 g of intravenous cefazolin at induction prior to skin preparation and draping. The surgical site was prepared with a Betadine-alcohol skin preparation antiseptic, equivalent to 10% povidone-iodine solution and 30% ethyl alcohol. This preparation was allowed to dry before an adhesive layer was applied. Sterile drapes were applied, and an iodine-impregnated adhesive drape (Ioban; 3M) was used to cover all remaining skin and to isolate the axilla. Once the skin was covered, the surgeon’s gloves and the assistant’s gloves were changed before the incision. A standard deltopectoral approach was used in all patients.
Strict specimen collection protocol was adhered to in all cases. At the time that the glenohumeral joint was opened, swabs (Transwab; Medical Wire & Equipment) were taken from the 5 different sites. Care was taken to hold swabs only by their distal end at the designated cap to avoid contamination by gloves. Swabs were taken from the following sites: the exposed subdermal layer of the incision, the tip of the surgeon’s gloves, the inside scalpel blade, the forceps, and the outside scalpel blade. In cases in which material on the swab site may have dried (e.g., outside scalpel blade), a drop of sterile saline solution was used to moisten the area before taking the swab. Instruments were not changed at any stage during the case.
The culture swabs were transported in a standard sterile culture swab container (Interpath) and were sent for analysis within 2 hours. Microbiology staff processed the specimens using a strict aseptic technique in a Class-II biological safety cabinet. Swabs were inoculated onto blood agar, prereduced blood agar, and chocolate agar plates.
Specimens were incubated at 37°C under 5% carbon dioxide aerobic conditions and anaerobic conditions for 14 days. Aerobic plates were examined at 24 and 48 hours. Anaerobic plates were examined at 48 hours and at 5, 10, and 14 days looking specifically for P. acnes. They were determined to be negative if there was no growth at the 14-day examination. Organisms grown were identified with initial Gram stain and matrix-assisted laser desorption/ionization time-of-flight (MALDI-ToF) mass spectrometry. The minimum follow-up time for a clinical review to exclude infection was 6 months.
A power calculation was performed on the basis of the results of the first 20 patients to ensure sufficient numbers to detect a minimum 10% difference in proportions of contamination from the primary surgical site (subdermal layer). Given that it is well documented that there is a higher bacterial load of P. acnes in men compared with women1, we also evaluated differences between our patients according to sex. A minimum of 35 patients was required to demonstrate a proportional difference of 20% between sexes with 90% power. Therefore, 40 patients were recruited. Statistical analysis was carried out to find differences in proportions, age, and sex using chi-square tests (proportions), Student t tests (age), and logistic regression (odds ratios). Odds ratios are given as a mean with a 95% confidence interval (95% CI).
P. acnes Contamination
Sixteen patients (40%) had a positive culture result on at least 1 of the 5 swabs. Thirteen (33%) of the 40 patients had P. acnes growth on ≥1 of their swabs, 4 had a coagulase-negative Staphylococcus, and 1 had methicillin-sensitive Staphylococcus aureus (MSSA). Of the 200 swabs taken during the study, 31 (15.5%) grew P. acnes, 4 (2%) grew coagulase-negative Staphylococcus, and 1 (0.5%) grew MSSA. The percentage growth of the organisms from positive swabs is demonstrated in Figure 1. The cases of 6 patients who had 3 positive swabs for P. acnes were discussed with a consultant in infectious diseases and clinical microbiology, and these patients were empirically treated with oral amoxicillin (or clindamycin in case of penicillin allergy) for 6 weeks, as they were deemed at high risk of developing a postoperative infection. Although we believed that these results represented contamination, these patients were treated to prevent potential further infection, in keeping with our institutional review board protocol guidelines. There were no significant differences (p > 0.05) in contamination with respect to age, shoulder side (left or right), surgeon, hospital site, reason for the surgical procedure (diagnosis), or prosthesis used. Age was not a significant cofactor (p > 0.05) in any difference between male patients and female patients.
The time to growth of these specimens was also recorded, with a mean time of 6.2 days (range, 4 to 10.25 days). No specimens collected had any growth before 4 days. Seventeen (55%) of the 31 specimens that grew P. acnes had growth by day 5, 7 specimens (23%) had growth by day 8, and 7 specimens (23%) had growth between 9 and 10.25 days.
Surgical Site of Contamination
The culture specimens obtained from the subdermal layer were almost twice as likely to yield a positive result compared with any other site. In total, P. acnes was cultured from 12 of the subdermal layer swabs, 7 swabs from the tip of the surgeon’s glove, 1 from the inside scalpel blade, 7 from the forceps, and 4 from the outside scalpel blade (Fig. 2). Interestingly, all cases with glove tip contaminations also had a positive result from the subdermal layer (p < 0.001). Six (86%) of the 7 forceps contamination results also had a positive result from the subdermal layer (p < 0.001).
There were 125 swabs taken from the female group and 75 taken from the male group. There were nearly 7 times the total number of swabs positive for P. acnes in the male group (27) compared with the female group (4) (p < 0.001) (Fig. 3).
A significantly higher proportion (p < 0.001) of male patients (73%) had ≥1 positive swabs compared with female patients (8%); a significantly higher proportion (p = 0.001) of male patients (66%) had multiple positive swabs (≥2) compared with female patients (4%). Men also had a significantly higher proportion of P. acnes contamination of the subdermal layer (p < 0.001), the tip of the surgeon’s glove (p < 0.001), and the forceps (p = 0.041) compared with women (Table II). Male patients had 66 times (95% CI, 6 to 680 times) higher odds of having subdermal contamination than female patients (p < 0.001). Male patients had 32 times (95% CI, 5 to 205 times) higher odds of having any contamination than female patients (p < 0.001).
There were no postoperative infections observed in our study. There was one complication within our study group. One patient returned for the routine 4-month follow-up reporting anterior shoulder pain following an injury playing tennis 1 week earlier. He demonstrated point tenderness over the anterior aspect of the shoulder and a painful click on rotation. He had unremarkable radiographs and inflammatory markers and a normal bone scan. These symptoms persisted, and he had a reoperation for exploration at 6 months after the initial procedure. At the time of this surgical procedure, he was found to have suture failure at the upper border of the subscapularis with a prominent suture knot or granuloma. This suture was sent for culture, noting that at his primary procedure he had had swabs from the subdermal layer and the tip of the surgeon’s glove positive for P. acnes. This suture material also grew P. acnes and he was treated with a 6-week regimen of oral amoxicillin (1 g every 6 hours). His symptoms immediately resolved following removal of the suture and he returned to tennis without concern. Thereafter, his clinical follow-up was unremarkable and his C-reactive protein level remained normal.
This study demonstrates that P. acnes is commonly present in the surgical field and on instruments in shoulder arthroplasty. The subdermal layer appears to be the source of contamination, as this is where we found the most positive culture results (39% of all positive results). The high number of positive results highlights the limitations of surgical site disinfection and the effectiveness of antibiotic prophylaxis. As P. acnes colonizes sebaceous glands within the dermis, current surgical site disinfection measures may be suboptimal. Subsequent contamination of the rest of the surgical field likely results from handling the subdermal layer (with gloves or forceps) as is suggested by the correlation between positive swabs from the tip of the surgeon’s glove and surgical instruments and from the subdermal layer. Our study also confirms previous documented findings that male patients carry a higher P. acnes bacterial load and therefore have higher odds of definitively growing P. acnes from a culture specimen obtained during shoulder arthroplasty.
The reported rates of postoperative infection with P. acnes following shoulder surgical procedures vary widely. Sperling et al.13 identified P. acnes as the causative organism in 16% of all primary shoulder arthroplasty infections between 1972 and 1994. More recently, rates of infection with P. acnes have been reported as being as high as 34% (by Richards et al.14) and 56% (by Levy et al.3). A study by Singh et al.15 that examined the causative organisms in periprosthetic shoulder infections over a 33-year period demonstrated increasing rates of P. acnes infection, especially within the last decade of their study. This increase could be accounted for by increased awareness of P. acnes as a causative organism, improved laboratory techniques, and inconsistent definitions and criteria for infection.
From a clinical standpoint, it is crucial to differentiate between contamination and infection. This study examined the rate and site of contamination. Time to growth of P. acnes from all swabs was relatively high, with a mean time of 6.2 days. There were no patients who met clinical criteria for infection intraoperatively or met clinical or microbiological criteria at the follow-up of 6 to 18 months. We recognize the limitations of this follow-up period, particularly with respect to the often late-onset indolent infections seen with P. acnes.
It was recently proposed by Levy et al. that P. acnes could be a causative organism for osteoarthritis, because of the high rate of growth of P. acnes (41.8%) from intraoperative specimens at the primary arthroplasty10. When Maccioni et al.12 optimized specimen handling, using a modified Oxford protocol, the rate of growth from intraoperative specimens was dramatically reduced (3.1%), and that same protocol was used in the current study.
The correlation between the growth of P. acnes from the skin and subdermal layer and both the tip of the surgeon’s glove and the forceps was significant. The correlation demonstrates that it is likely that surgeon handling of the skin and subdermal layer contaminates the rest of the surgical field. The high rate of growth from the swabs is despite appropriate skin preparation as P. acnes remained prevalent within the field. This is presumably because P. acnes resides in the subdermal layer and, once the skin is opened, the bacteria are then exposed to the rest of the field.
Further research could therefore be focused upon methods to reduce the exposure of P. acnes to the rest of the field. Techniques such as minimizing handling of the subdermal layer, changing gloves again after the dermis is cut, and making sure that implants never touch the subdermal layer should be considered first and foremost. Repeat preparation of the exposed subdermal layer with an antibacterial agent once the wound is opened and throughout the case may also help to reduce the bacterial load. It appears that the type of antibacterial agent used is not important. Saltzman et al.16 showed no significant difference among three different commonly used agents in their ability to eliminate P. acnes. Antiseptic-soaked wound towels to cover all skin edges and subdermis may also reduce the chance of further contamination throughout the case.
Other authors have previously reported high rates of P. acnes isolated in culture specimens obtained at the time of revision shoulder arthroplasty and have suggested that this may represent an indolent infection and/or even a possible cause of prosthetic loosening17,18. As a result of this study and the previous work conducted by Maccioni et al.12, we recommend caution in interpreting results from previous studies that have shown high rates of P. acnes infection, and we emphasize using strict aseptic collection techniques for fluid and tissue in future studies. As we have demonstrated, the P. acnes cultured may represent a surgical field contamination rather than a preexisting infection.
Our study had a number of limitations. First, as demonstrated in the study by Mook et al.18, it would have been useful during specimen collection to perform a control swab (e.g., sterile sponge) to determine the accuracy of the microbiology laboratory processing. Thus, we do not know whether laboratory contamination occurs. It might also have been better to obtain >1 swab from each site to determine the accuracy of the microbiology laboratory. The fact that all samples were taken at one point during the operation might impact the culture results, as delay in culturing might affect bacterial growth. Our results are based on our techniques of skin preparation and tissue handling, and a single microbiology laboratory was used to process specimens and thus the results may not be generalizable. Antibiotics that were given at the time of induction of anesthesia may inhibit bacterial growth and may affect the rate of positive cultures found with our specimens. It has been demonstrated by Pottinger et al.8 that withholding antibiotics before specimens are taken (in revision arthroplasty) results in a higher culture yield of P. acnes. This, coupled with the fact that cefazolin may be less effective than the preferred antibiotics (e.g., penicillins or clindamycin) for treating P. acnes infections, suggests that the rates of contamination in our study are unlikely to be an underestimation.
In summary, the current study suggests that P. acnes identified in the surgical field of a shoulder arthroplasty is a contaminant that derives from the edges of the surgical incision and is spread throughout the field because of contact with the surgeon’s gloves and surgical instruments. Our findings are consistent with those in the literature regarding the high rates of P. acnes bacterial load and intraoperative growth in male patients compared with female patients. The high rate of surgical field contamination has relevance to perioperative antibiotic prophylaxis, skin preparation, development of periprosthetic shoulder infections, and interpretation of wound cultures at the time of revision shoulder arthroplasty. We recommend reinforcement in the current surgical practice of strict skin handling; avoidance of the subdermal layer by gloved fingers, instruments, and implants; and a focus in future studies on limiting iatrogenic contamination of the surgical wound with P. acnes.
Investigation performed at the Sydney Shoulder Research Institute, St. Leonards, Australia
Disclosure: The authors did not receive any outside funding or grants in support of their research or for preparation of this work. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work and “yes” to indicate that the author had other relationships or activities that could be perceived to influence, or have the potential to influence, what was written in this work.
- Copyright © 2016 by The Journal of Bone and Joint Surgery, Incorporated