Background

Periprosthetic joint infection (PJI) remains worrisome after total joint arthroplasty. Therefore, methods used to prevent infection should be based on the strongest evidence possible. For instance, perioperative antibiotic prophylaxis [1], skin preparation and draping [2], and some ingenuity in wound closure [3] should be applied. An easy and realistic method is lavage prior to wound closure, particularly methods using antisepsis solutions such as diluted povidone-iodine (PI) or chlorhexidine gluconate (CHG). Indeed, several studies have used antiseptic solutions to prevent PJI.

In terms of preventing surgical site infection (SSI) during general surgery, intraoperative PI is proven to be effective; strong evidence is provided by a meta-analysis of randomized controlled trials conducted approximately 10 years ago [4]. Similarly, another meta-analysis showed that PI lavage significantly reduces SSI after surgery involving spinal instruments [5]. However, PJI is distinct from SSI during general surgery, including abdominal surgery and spine surgery. This is because PJI presents with a particular pathology, including biofilm formation [6] and a specific organism profile [7]; therefore, PI lavage may not have the same effectiveness in preventing PJI as it has in preventing SSI in general surgery. In fact, the results of recent studies on the effectiveness of diluted PI lavage for preventing PJI are controversial [8, 9]. In addition, a recent meta-analysis suggests that diluted PI lavage does not prevent PJI [10]. Thus, a review of the latest evidence is required.

The clinical question of this study is, “Does diluted PI lavage actually reduce the risk of PJI?” The aim of the study was to perform a systematic review and meta-analysis of the current literature concerning the efficacy of diluted PI lavage for preventing PJI in primary and revision surgery.

Methods

This systematic review and meta-analysis was conducted in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) checklist for systematic reviews and meta-analyses (http://prisma-statement.org/PRISMAStatement/Checklist).

Literature search

Multiple comprehensive literature searches of PubMed, Cumulative Index to Nursing and Allied Health Literature (CINAHL), ClinicalTrials.gov, and the Cochrane Library databases were performed on July 13, 2021. Search key words included (“betadine” OR “povidone” OR “povidone-iodine”) AND (“lavage” OR “dilute”) AND (”total hip arthroplasty (THA)” OR “total knee arthroplasty (TKA)” OR “arthroplasty” OR “Periprosthetic joint infection“). An additional manual search was performed to identify other relevant articles or bibliographies.

Study screening and eligibility assessment

After the first extraction of literature, a first screening was performed by two reviewers. During this screening, the title and abstract were reviewed, and inappropriate literature was excluded. Next, eligibility assessment of full manuscripts was performed by the same two reviewers. The inclusion criteria were as follows: direct comparison between the PI and non-PI lavage groups following total joint arthroplasty (TJA), in which primary or aseptic revision arthroplasty was performed; a PI lavage protocol was used, not a combination protocol with other solutions such as chlorhexidine lavage; the overall infection rate was stated, and when a PI regimen was used for lavage, the article included details of the dosing protocols and the duration of PI application. The exclusion criteria were as follows: non-original clinical research articles, including biomechanical or cadaveric studies, technical notes, letters to the editor, expert opinions, review articles, meta-analyses, and case reports; no full text available; duplicate studies from the same investigation group; and reported follow-up < 3 months.

Data extraction

Data were extracted from the full text using a piloted form that included the publication date, the study design, type of surgery, number of patients, follow-up length, type of preoperative prophylaxis, type of postoperative prophylaxis, type of intervention (solution type, application method, and volume used), and type of control. Two investigators performed data extraction and reached agreement in all cases.

Data synthesis and statistical analysis

The results of the systematic review were summarized qualitatively into a meta-analysis of pooled odds ratios with 95% confidence intervals (95% CIs). The analyses were conducted using RevMan 5.3. A P value < 0.05 was considered statistically significant. Heterogeneity of treatment effects among studies was evaluated by calculating I2 and was categorized as low, moderate, or high (I2 < 25%, 25–50%, and > 50%, respectively). A random effects model was applied in cases of high heterogeneity; otherwise, a fixed effects model was applied. Subgroup analyses were conducted to identify potential sources of heterogeneity. All statistical analyses were performed using Review Manager (RevMan, version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) (computer program).

Evaluation of bias risk

Risk of bias in non-randomized studies of interventions (ROBINS-I) was graded by two reviewers per study. The Coleman methodology score (CMS) [11] criteria was also evaluated for research methodological quality by two reviewers per study. Its criteria were slightly modified to suit to the purpose of the present systematic review (Table 1). A test for publication bias was not performed because evaluation of publication bias is typically performed only when at least ten studies are included in a meta-analysis.

Table 1 Modified Coleman Methodology Score (CMS) for studies reporting the outcomes of surgery
Full size table

Results

Literature screening and identification

Figure 1 shows the detailed flow and the number of screened publications. The initial electronic database search identified 590 studies. Two additional manual search records were added from other sources. Thus, 490 studies were screened after removing 117 duplicates. After the screening and eligibility assessment process, eight publications [8, 9, 12,13,14,15,16,17] were included in the analysis.

Fig. 1
figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagrams for identification and selection of studies to be included in the meta-analysis

Full size image

Study and treatment characteristics

Table 2 summarizes the study demographics. Only one of the eight studies was a randomized controlled trial; the other seven were retrospective cohort studies. Table 3 summarizes the treatment characteristics in each study. In three studies, we found that the control group was set as “lavage with saline,” whereas chlorhexidine gluconate was used in two studies. No information was provided by three studies.

Table 2 Study demographics and characteristics
Full size table
Table 3 Treatment characteristics in each study
Full size table

Individual study results and synthesis of results

Table 4 summarizes the PJI rates in each study. In total, 10,390 subjects were identified as belonging to the PI lavage group, and 22,623 subjects were identified as belonging to the non-PI lavage group. In the PI lavage group, 91 were identified as having PJI compared with 215 in the non-PI lavage group. Figure 2 shows the results of the meta-analysis. In studies that used a saline control group, the odds ratio for PI lavage was 0.33 (95% CI, 0.16–0.71; P = 0.004); thus, the risk of PJI was significantly reduced. In studies that used a CHG control group, the odds ratio for PI lavage was 2.17 (95% CI, 0.97–4.87; P = 0.06); in this case, the difference was not significant. In studies that provided no detailed information about the control groups, the odds ratio of PI lavage was 1.04 (95% CI, 0.52–2.09; P = 0.92), i.e., no significant difference. Overall, the odds ratio for PI lavage was 0.83 (95% CI, 0.45–1.51; P = 0.54), suggesting that PI lavage has no significant effect on the risk of PJI, although the control group was not unified in this setting.

Table 4 Results of each individual study
Full size table
Fig. 2
figure 2

Forest plot of the random effects model showing the odds ratio and 95% confidence intervals for PI lavage (+) compared with those for PI lavage (−)

Full size image

Risk of bias and quality assessment

Figure 3 summarizes the risk of bias. In one study (Calkins et al. [9]), the overall risk of bias was “moderate.” In another study (Fleischman et al. [14]), the overall risk of bias was “critical.” For the remaining six studies, the overall risk of bias was “serious.” The CMS ranged from 35 to 80 within component studies (mean: 58.25, SD: 17.23, median: 60) (Tables 2 and 5).

Fig. 3
figure 3

Evaluating the risk of bias using risk of bias in non-randomized studies of interventions (ROBINS-I)

Full size image
Table 5 Coleman methodology score (CMS): mean, SD, range, and median values for each component element of CMS
Full size table

Discussion

The most important result in this study is that dilute PI lavage is significantly more effective against PJI than saline lavage. Although the combined results of all studies suggest that PI lavage has no significant effect, careful consideration of the negative control conditions used in each study led us to conclude that dilute PI lavage is significantly more effective in preventing PJI than saline lavage in routine surgical procedures.

The oldest study, conducted by Brown et al. [12], was a retrospective cohort study with a total of 2540 consecutive patients receiving total joint arthroplasty; they reported a significantly lower rate of infection in the dilute PI lavage group than in the control saline lavage group. Also, a recent large retrospective cohort study reported a lower rate of infection in a PI group than in a saline lavage group when the groups were propensity-matched [17]. Similarly, a study by Calkin revealed that dilute PI lavage significantly reduced occurrence of acute PJI after aseptic revision of TKA and THA [9]. This study had the lowest risk of bias due to its randomized controlled trial design and the inclusion of a defined negative control, saline lavage. By contrast, two another retrospective cohort studies conducted by the same group using a relatively large number of samples concluded that dilute PI lavage does not reduce the risk of reoperation for infection after both primary and revision THA and TKA [8, 16]. Thus, there was no consensus regarding the routine use of dilute PI lavage for the prevention of PJI.

We identified only one previous meta-analysis study examining the effectiveness of dilute PI lavage for the prevention of PJI; this was a systematic review study by Kim et al. [10]. They found no difference in the overall postoperative infection rates between the PI and non-PI lavage groups. While our overall results agree with these results, our sub-analysis of studies that reported saline solution controls clearly shows that PI lavage reduces rates of PJI significantly. There are several possible reasons for this discrepancy. First, we included very recent publications from 2020, which showed positive results for PI lavage. Second, we performed sub-analysis excluding studies that used CHG as a control or that provided no detailed information about the control solution. A retrospective study by Hart et al. showed negative effect of PI lavage for preventing PJI in large cohort of revision arthroplasty. This study did not clearly define control group protocol that was just “no use of PI lavage.” In addition, the use of PI lavage was decided only by surgeon’s discretion that should arise severe selection bias. Studies that compared PI and CHG failed to show an advantage of PI. Indeed, the study by Driesman et al. compared PI and CHG lavage as preventive measures for PJI [13], but found no differences in their effectiveness. Because the study was conducted to show the “non-inferiority” of CHG compared with PI, we could not use this study to investigate the effectiveness of PI. Indeed, we found a significant negative effect in our sub-analysis using CHG control groups. Thus, our overall analysis identified high heterogeneity: I2 = 73%. Similarly, we excluded studies that did not provide detailed information about control groups. This was the case for three retrospective cohort studies, which were excluded from subgroup analysis. Thus, subgroup analysis of studies that included a saline control showed low heterogeneity I2 = 7%.

Several recent studies were excluded from this systematic review during eligibility assessment, some of which reported evidence supporting PI utility. Nazal et al. reported that treating sterile water splash basins with dilute PI (0.02% solution) eliminates intraoperative contamination of such splash basins during total joint arthoplasty [18]. This may contribute indirectly to reducing the risk of PJI. Cichos et al. conducted an in vitro study to compare the effectiveness of PI, CHG gluconate, and vancomycin with respect to minimal inhibitory concentrations (MIC) and time to death of multiple bacteria [19]. They showed that all bacterial isolates tested were killed only by PI and that PI killed all bacteria tested immediately on contact; exposure time was not the key factor. On the other hand, a negative finding was that PI had a chondrotoxic effect on the superficial cartilage layer [20].

It should be noted that all except one of the publications analyzed herein were retrospective in design; therefore, the risk of bias was high (“serious”) in seven of the eight studies. Further studies with a lower risk of bias (i.e., a prospective randomized design with strict negative controls) are needed to support the evidence that PI is effective in preventing PJI. In addition, alternatives to PI solution lavage, such as CHG solution lavage [21] or vancomycin powder [22], should be investigated for their ability to eradicate PJI.

Conclusion

This systematic review and meta-analysis of the current literature demonstrates that diluted PI lavage is significantly better than saline solution lavage for preventing PJI. We recommend diluted PI lavage (0.35%) be used to prevent PJI rather than saline solution lavage.