Abstract
Purpose
-
This investigation provides a rigorous systematic review of the postoperative outcomes of patients with and without chronic hepatitis C who underwent total hip arthroplasty (THA) and total knee arthroplasty (TKA).
Methods
-
We queried PubMed, Embase, Cochrane Database of Systematic Reviews, Scopus, Web of Science and the ‘gray’ literature, including supplemental materials, conference abstracts and proceedings as well as commentary published in various peer-reviewed journals from 1992 to present to evaluate studies that compared the postoperative outcomes of patients with and without chronic hepatitis C who underwent primary THA or TKA. This investigation was registered in the PROSPERO international prospective register of systematic reviews and follows the guidelines provided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. In our literature search, we identified 14 articles that met our inclusion criteria and were included in our fixed-effects meta-analysis. The postoperative outcomes analyzed included periprosthetic joint infection (PJI), aseptic revision, non-homebound discharge and inpatient mortality.
Results
-
Our statistical analysis demonstrated a statistically significant increase in postoperative complications of patients with chronic hepatitis C who underwent primary THA or TKA including PJI (odds ratio (OR): 1.98, 95% CI: 1.86 – 2.10), aseptic revision (OR: 1.58, 95% CI: 1.50 – 1.67), non-homebound discharge (OR: 1.31, 95% CI: 1.28– 1.34) and inpatient mortality (OR: 9.37, 95% CI: 8.17 – 10.75).
Conclusion
-
This meta-analysis demonstrated a statistically significant increase in adverse postoperative complications in patients with chronic hepatitis C who underwent primary THA or TKA compared to patients without chronic hepatitis C.
Introduction
Recent estimates of the prevalence of patients infected with hepatitis C include at least 5.2 million persons living in the United States. (1) Patients with chronic hepatitis C who underwent total joint arthroplasty (TJA) have been shown to incur some of the highest costs of care as reflected by reimbursements during the 90-day global period (2). With projected annual rates of total hip arthroplasty (THA) and total knee arthroplasty (TKA) expected to increase exponentially to 635 000 and 1.26 million, respectively, by 2030 (3), there is increased interest in investigating the modifiable risk factors that may contribute to postoperative complications. Postoperative complications, including periprosthetic joint infection (PJI), can carry significant morbidity to patients and cost the healthcare system hundreds of millions of dollars per year. (4) Moreover, TJA requires the manual manipulation of sharp boney surfaces and instruments which can lead to high rates of glove perforation (5), leading to the transmission of blood-borne pathogens. As medical innovations have advanced, new medical treatments for chronic hepatitis C have become more affordable and accessible to patients, including ‘curative’ modalities. These advances have stimulated interest into the feasibility, practicality and potential benefits of preoperative treatment of hepatitis C prior to undergoing TJA (6, 7, 8, 9). More recently, a published meta-analysis of observational studies suggests that anti-viral treatment for chronic hepatitis C is associated with a reduced risk of PJI (10).
Although multiple studies have evaluated the incidence of postoperative complications in patients with hepatitis C undergoing TJA, the majority of the literature consists or institutional data which are of insufficient power to form definitive conclusions. This meta-analysis as such sought to provide a rigorous quantitative analysis of the postoperative complications noted of patients with chronic hepatitis C who underwent elective THA and TKA.
Methods
This investigation is registered in the PROSPERO international prospective register of systematic reviews (CRD42019134121). Our investigation followed the guidelines presented by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (11).
This meta-analysis sought to quantitatively investigate the postoperative outcomes of patients with and without chronic hepatitis C who underwent elective THA or TKA. The primary outcome measures used in this analysis include PJI, aseptic revision, non-homebound discharge and inpatient mortality.
Eligibility criteria
The population being studied included patients greater than or equal to 18 years of age with chronic hepatitis C as defined by International Classification of Diseases, Clinical Modification (ICD-9-CM) codes 070.54 or 070.70, or positive hepatitis C virus (HCV) core antigen testing, standard immunoassay testing, rapid immunoassay testing, recombinant immunoblot assay testing, positive HCV viral load (VL) or HCV RNA assay testing who underwent TKA or THA. We compared these patients to patients without chronic hepatitis C who underwent TKA or THA. Exclusion criteria included co-infection with human immunodeficiency virus (HIV) or hepatitis B virus (HBV) at the time of surgery. Additionally, we excluded case reports but included observational (retrospective or prospective) studies, case–control studies and randomized control trial studies in our database search.
We included patients undergoing elective THA or TKA with or without chronic hepatitis C, as defined by Current Procedural Terminology codes 27447 or 27130 or ICD-9-CM procedure code 84.54.
Information sources
The results of our literature search summarized in a PRISMA flow diagram provided in Supplementary Fig. 1 (see section on supplementary materials given at the end of this article) (12). We searched PubMed, EMBASE, Cochrane Database of Systematic Reviews, Scopus, Web of Science and the ‘gray’ literature, including the supplemental materials, conference abstracts and proceedings as well as commentary published in The Journal of Bone and Joint Surgery (British and American Volumes), Clinical Orthopaedics and Related Research and The Journal of Arthroplasty in the English language from 1992 to present. The year 1992 was selected as a start date because it was when ‘hepatitis C’ was introduced as a MeSH term in PubMed. We used MeSH terms and keyword searches that include ‘hepatitis c’, ‘arthroplasty’, ‘total joint arthroplasty’, ‘total knee arthroplasty’, ‘total hip arthroplasty’, ‘outcomes’, ‘total knee replacement’ and ‘total hip replacement’. Please see the supplementary data for a complete list of search terms for each database queried.
Study selection
In order to perform the meta-analysis, we identified 14 studies (6, 7, 8, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23) that met our inclusion criteria (see Table 1). All were retrospective cohort studies.
List of all studies included in the meta-analysis. Note that some studies analyzed TKA and THA separately, so the total sample size reported here does not reflect the sample size of the separate analyses.
| Study | Sample size | Study period |
|---|---|---|
| Bendich et al. (7) | 42 268 | 2014–2018 |
| Best et al. (13) | 8 363 266 | 1990–2007 |
| Cancienne et al. (14) | 218 744 | 2005–2012 |
| Issa et al. (15) | 25 372 | 1998–2010 |
| Issa et al. (16) | 216 | 2002–2011 |
| Kildow et al. (17) | 22 663 | 2005–2012 |
| Mahure et al. (18) | 80 722 | 2010–2014 |
| Novikov et al. (19) | 233 | 2005–2018 |
| Orozco et al. (20) | 149 | 2006–2011 |
| Pour et al. (21) | 215 | 1995–2006 |
| Schwarzkopf et al. (8) | 127 | 2011–2018 |
| Bedair et al. (6) | 105 | 2006–2016 |
| Jergesen et al. (22) | 486 | 2008–2013 |
| Mahure et al. (23) | 137 801 | 2010–2014 |
Some studies included only patients who underwent TKA, others included only patients who underwent THA, and some performed pooled analyses including both surgery types. Since we expect the impact of HCV infection to be heterogeneous across different surgery types, we included results for TKA and THA separately when possible but used the pooled TJA analysis if that was the only result provided by a study. Studies also differed in terms of which patient groups they compared. Most studies compared HCV-infected patients to uninfected patients. However, other studies considered only patients known to be infected with HCV, comparing treated/cured patients to untreated/uncured patients or patients with detectable VL to patients with undetectable VL. We included all of these results in the meta-analysis, since patients who are cured or who have undetectable VL are expected to have similar outcomes to patients with no HCV infection.
Our outcomes of interest were PJI, aseptic revision surgery, non-homebound discharge status and inpatient mortality. We used odds ratios (ORs) to measure the association between HCV status and each outcome. We used the standard asymptotic normal approximation to construct 95% CIs for the logarithm of the OR and then transformed these into CIs for the OR itself. For some outcomes, there was complete separation, meaning that the outcome event was observed zero times in either the infected or the uninfected group. In this case, we added 1/2 to each observed count (infected with outcome, infected without outcome, uninfected with outcome and uninfected without outcome) and then calculated the OR and CI in the same way as above (24).
Data collection
The studies included were selected from a larger group that are the result of a search that utilizes the abovementioned MeSH terms and keyword searches in the previously mentioned databases. Two reviewers determined whether the study evaluates patients with chronic hepatitis C who underwent THA or TKA without co-infection with HIV or HBV by consensus. If there was a disagreement among the two reviewers, a third reviewer adjudicated the disagreement. The data extracted included the outcome measures listed previously.
Statistical analysis
For each outcome of interest, we performed a fixed-effects meta-analysis. Unlike a traditional ‘common-effect’ meta-analysis, the fixed-effects meta-analysis does not assume that all studies estimate a single common association. The fixed-effects approach allows heterogeneity between the studies. We used standard precision weights. We constructed 95% CIs for the pooled estimates using a normal approximation.
For each analysis, we estimated the between-study heterogeneity in two ways. First, we use the I 2 statistic (25). We also used a re-weighted version of Cochran’s Q, as described by Dersimonian and Laird (26). This re-weighted Q can be interpreted as the estimated standard deviation of the effect sizes around the pooled estimate, weighted by study precision. This is presented on the scale of the logarithm of the OR.
For the analysis of PJI, where there was a mixture of studies that did and did not address confounding, we performed a fixed-effects meta-regression of the log OR estimate on a binary variable indicating whether a study accounted for confounding. This is equivalent to a standard weighted linear regression with inverse variance weights. We constructed a 95% CI for the regression coefficient using the model-based standard error estimate.
Risk of bias in individual studies
In order to investigate the possibility of reporting bias, we produced a funnel plot for each analysis. Asymmetry in the funnel plot can be an indication of reporting bias; in particular, a lack of studies with lower precision and smaller estimated effect sizes may be cause for concern.
In the 14 studies, there were a variety of approaches to addressing the confounding factors. Some studies performed frequency matching of exposed and unexposed patients, others included possible confounders as regression covariates, and a few studies did not account for confounding. For the purposes of the meta-analysis, we considered a study to have accounted for confounding if it provided estimates from either matched or adjusted analyses. Studies varied in terms of which confounders they included, although most used at least age, sex, diabetes, and obesity or BMI.
To assess the relationship between the study-specific effect size estimates and study quality, we performed a fixed-effects meta-regression. We estimate that the difference in effect size comparing studies that controlled for confounding to those that did not is –1.55 (95% CI: -1.70, -1.41) on a logarithmic scale. In other words, we estimate that adjusted/matched ORs are around 79% smaller, on average, than ORs from studies that did not match or adjust for confounders.
Given this substantial difference, we repeated the fixed-effects meta-analysis including only those studies that attempted to control for confounding. The I 2 is now 20%, and the estimated weighted standard deviation of the effect sizes is 0.07. The overall OR estimate in this case is lower than in the previous analysis.
Results
Periprosthetic joint infection
There were 12 studies that analyzed the association between HCV infection and the risk of PJI (see Table 2).
List of all studies included in the PJI analysis. Studies without a listed follow-up time or times did not indicate a pre-established amount of follow-up time.
| Study | Surgery | Comparison groups | Adjusted/matched? | Follow-up |
|---|---|---|---|---|
| Bendich et al. (7) | TKA, THA | Treated, untreated | Yes | NA |
| Best et al. (13) | TJA | Infected, uninfected | No | NA |
| Cancienne et al. (14) | TKA, THA | Infected, uninfected | Yes | 3 months, 6 months, 1 year |
| Issa et al. (16) | THA | Infected, uninfected | Yes | NA |
| Kildow et al. (17) | TKA, THA | Infected, uninfected | Yes | 3 months, 2 years |
| Mahure et al. (18) | THA | Infected, uninfected | No | NA |
| Novikov et al. (19) | TJA | Detectable VL, undetectable VL | Yes | NA |
| Orozco et al. (20) | TKA, THA | Infected, uninfected | Yes | NA |
| Pour et al. (21) | TKA, THA | Infected, uninfected | Yes | NA |
| Schwarzkopf et al. (8) | TKA | Cured, uncured | Yes | NA |
| Bedair et al. (6) | THA | Treated, untreated | No | NA |
| Mahure et al. (23) | TKA | Infected, uninfected | No | NA |
For the meta-analysis, we excluded results comparing treated individuals to uninfected individuals, restricting our analysis to comparisons between treated and untreated individuals. For the study by Bendich et al. (7) and the study by Cancienne et al. (14), where multiple follow-up times were analyzed, we included the 90-day outcomes in order to maximize comparability. Within each study, the OR estimates for different follow-up times were quite similar, so we expect this to have little impact on our final results. The results of the fixed-effects meta-analysis are shown in Fig. 1.
Forest plot for meta-analysis of the association between chronic HCV infection and risk of PJI, including estimates with and without adjustment/matching. The center of each bar is the study-specific point estimate of the OR, and the bar itself is the associated 95% CI. The bar thickness corresponds to how much weight the study received in the pooled estimate. Study 6: Bendich et al. (7); Study 9: Best et al. (13); Study 12: Cancienne et al. (14); Study 33: Issa et al. (16); Study 36: Kildow et al. (17); Study 43: Mahure et al. (18); Study 50: Novikov et al. (19); Study 51: Orozco et al. (20); Study 55: Pour et al. (21); Study 65: Schwarzkopf et al. (8); Study 151: Bedair et al. (6); Study 525: Mahure et al. (23).
Citation: EFORT Open Reviews 8, 4; 10.1530/EOR-22-0117
The TJA estimate from Best et al. (13) and the TKA/THA estimates from Cancienne et al. (14) have by far the most precision and accordingly receive the heaviest weight in the meta-analysis. The high I2 statistic indicates that the vast majority of the uncertainty in the pooled OR is due to inter-study variability. Intuitively, this is supported by the fact that Best et al.’s (13) estimate is substantially larger in magnitude than the two estimates reported by Cancienne et al. (14) , while each of these estimates has high precision. The estimated weighted standard deviation of the effect sizes is 0.69.
As all the included studies were observational in nature, accounting for confounders (either by matching or regression adjustment) is important. We repeated the fixed-effects meta-analysis including only those studies that attempted to control for confounding. The results of this analysis are shown in Figure 2. The I2 was then 20%, and the estimated weighted standard deviation of the effect sizes was0.07. The overall odds ratio estimate in this case was lower than in the previous analysis. We estimate that the odds of inpatient infection are 1.98 times higher in HCV-infected patients than in uninfected patients (95% CI: 1.86, 2.10).
Forest plot for meta-analysis of the association between chronic HCV infection and risk of PJI, including only estimates with adjustment/matching. The center of each bar is the study-specific point estimate of the OR, and the bar itself is the associated 95% CI. The bar thickness corresponds to how much weight the study received in the pooled estimate. Study 6: Bendich et al. (7); Study 12: Cancienne et al. (14); Study 33: Issa et al. (16); Study 36: Kildow et al. (17); Study 50: Novikov et al. (19); Study 51: Orozco et al. (20); Study 55: Pour et al. (21); Study 65: Schwarzkopf et al. (8).
Citation: EFORT Open Reviews 8, 4; 10.1530/EOR-22-0117
The results of the fixed-effects meta-analysis for the risk of PJI are shown in Fig. 1.
Aseptic revision
There were five studies that analyzed the association between HCV infection and the probability of undergoing an aseptic revision surgery (see Table 3).
List of all studies included in the revision analysis. Studies without a listed follow-up time or times did not indicate a pre-established amount of follow-up time.
| Study | Surgery | Comparison groups | Adjusted/matched? | Follow-up |
|---|---|---|---|---|
| Cancienne et al. (14) | TKA, THA | Infected, uninfected | Yes | 6 months, 1 year, 2 years, 8 years |
| Kildow et al. (17) | TKA, THA | Infected, uninfected | Yes | 3 months, 2 years |
| Novikov et al. (19) | TJA | Detectable VL, undetectable VL | Yes | NA |
| Orozco et al. (20) | TKA | Infected, uninfected | Yes | NA |
| Pour et al. (21) | TJA, TKA, THA | Infected, uninfected | Yes | NA |
VL, viral load.
To maximize comparability, for studies with multiple follow-up times, we used only the results from the 2-year follow-up. As with other outcomes, the results for different follow-up times within a study varied little. Pour et al. (21) included results for both revision and reoperation. We used only revision, since that was the outcome used by all other studies.
The results of the fixed-effects meta-analysis are shown in Fig. 3.
Forest plot for meta-analysis of the association between chronic HCV infection and risk of aseptic revision. The center of each bar is the study-specific point estimate of the OR, and the bar itself is the associated 95% CI. The bar thickness corresponds to how much weight the study received in the pooled estimate. Study 12: Cancienne et al. (14); Study 36: Kildow et al. (17); Study 50: Novikov et al. (19); Study 51: Orozco et al. (20); Study 55: Pour et al. (21).
Citation: EFORT Open Reviews 8, 4; 10.1530/EOR-22-0117
The TKA/THA estimates from Cancienne et al. (14) are the most precise and accordingly receive the heaviest weight in the meta-analysis. The I 2 statistic is relatively low, indicating that heterogeneity is not a major concern. The estimated weighted standard deviation of the effect sizes is also low, around 0.04. The four studies receiving at least 1% weight in the overall OR all provide similar estimates. Overall, we estimate that the odds of revision in HCV-infected patients is 1.58 times as high as in uninfected patients (95% CI: 1.50, 1.67).
Non-homebound discharge
There were five studies that analyzed the association between HCV infection and the risk of non-homebound discharge (see Table 4). Only one of the five studies provided results from an adjusted or matched analysis.
List of all studies included in the analysis of non-homebound discharge. In all studies, patients were followed until discharge.
| Study | Surgery | Comparison groups | Adjusted/matched? |
|---|---|---|---|
| Best et al. (13) | TKA, THA | Infected, uninfected | No |
| Mahure et al. (18) | THA | Infected, uninfected | Yes |
| Novikov et al. (19) | TJA | Detectable VL, undetectable VL | No |
| Schwarzkopf et al. (8) | TKA | Cured, uncured | No |
| Mahure et al. (23) | TKA | Infected, uninfected | No |
VL, viral load.
The results of the fixed-effects meta-analysis are shown in Fig. 4.
Forest plot for meta-analysis of the association between chronic HCV infection and risk of non-homebound discharge. The center of each bar is the study-specific point estimate of the OR, and the bar itself is the associated 95% CI. The bar thickness corresponds to how much weight the study received in the pooled estimate. Study 9: Best et al. (13); Study 43: Mahure et al. (18); Study 50: Novikov et al. (19); Study 65: Schwarzkopf et al. (8); Study 525: Mahure et al. (23).
Citation: EFORT Open Reviews 8, 4; 10.1530/EOR-22-0117
The estimates from Best et al. (13) have a high level of precision compared to the others and accordingly receive the vast majority of the weight in the overall estimate. The I 2 statistic is high, and the estimated weighted standard deviation of the effect sizes is also large, at around 0.32. These indicate a statistically significant degree of heterogeneity. This could be due, at least in part, to the fact that four of the five studies did not attempt to control for confounding. We would therefore expect the resulting estimates to reflect the characteristics of the study-specific patient population. Overall, we estimate that the odds of non-homebound discharge in HCV-infected patients are 1.31 times as high as in uninfected patients (95% CI: 1.28, 1.34).
Inpatient mortality
There were three studies that analyzed the association between HCV infection and the risk of inpatient mortality (see Table 5). Only one of the three studies provided results from an adjusted or matched analysis.
List of all studies included in the analysis of inpatient mortality. In all studies, patients were followed until discharge or death.
| Study | Surgery | Comparison groups | Adjusted/matched? |
|---|---|---|---|
| Mahure et al. (18) | THA | Infected, uninfected | Yes |
| Schwarzkopf et al. (8) | TKA | Cured, uncured | No |
| Mahure et al. (23) | TKA | Infected, uninfected | No |
The results of the fixed-effects meta-analysis are shown in Fig. 5.
Forest plot for meta-analysis of the association between chronic HCV infection and risk of inpatient mortality. The center of each bar is the study-specific point estimate of the OR, and the bar itself is the associated 95% CI. The bar thickness corresponds to how much weight the study received in the pooled estimate. Study 43: Mahure et al. (18); Study 65: Schwarzkopf et al. (8); Study 525: Mahure et al. (23).
Citation: EFORT Open Reviews 8, 4; 10.1530/EOR-22-0117
The estimates from Mahure et al. (18) have a high level of precision compared to the others and carry almost all of the weight in the overall estimate. The I 2 statistic is low, although the estimated weighted standard deviation of the effect sizes is fairly substantial (0.29). Only one of the three studies give only adjusted estimates. Overall, we estimate that the odds of inpatient mortality in HCV-infected patients are 9.37 times as high as in uninfected patients (95% CI: 8.17, 10.75). We note that the results of Mahure et al. (18) are somewhat unusual; the estimated effect size is extremely large and precise, but the study reports a P-value that is only marginally significant (0.044). Furthermore, in attempting to reconstruct the standard error of the point estimate using the reported CI, we found that the interval is asymmetric about the point estimate on the logarithm scale. We would expect approximate symmetry in this scenario. Due to these concerns, we repeated the mortality analysis removing study 43. The results are shown in Supplementary Fig. 2. As expected, after excluding the study by Mahure et al. (18), the estimated association is substantially lower (OR: 2.62), with low precision.
Supplementary Figure 3 demonstrates the funnel plot for the PJI analysis with only studies that adjusted/matched. The funnel plot suggests there may be some publication bias due to the lack of studies with low precision and small effect size, although there are also few studies with low precision and large effect size. Supplementary Figure 4 demonstrates the funnel plot for the aseptic revision analysis. The asymmetric funnel plot suggests the presence of publication bias or some systematic difference between high- and low-precision studies. Supplementary Figure 5 demonstrates the funnel plot for the non-homebound discharge analysis. Due to the relatively small number of studies, and the general lack of control for confounding, it is difficult to interpret the funnel plot in this instance. Supplementary Figure 5 demonstrates the funnel plot for the inpatient mortality analysis. With only three studies, interpretation is difficult.
Discussion
Based on the meta-analysis we performed, we found a statistically significant increased risk of adverse postoperative outcomes comparing HCV-infected and uninfected patients undergoing TJA. For all outcomes, we see substantial heterogeneity when including both unadjusted estimates and adjusted estimates, as is expected, given that none of the studies were randomized. For the PJI and aseptic revision outcomes, where there are a sufficient number of adjusted estimates to exclude the unadjusted estimates, we observe relatively low heterogeneity. This was somewhat unexpected given that we included analyses of both TKA and THA. There was extreme heterogeneity in the analysis of non-homebound discharge, but a formal analysis of differences between adjusted and unadjusted estimates was not possible as only a single adjusted estimate was available. For mortality, there are only three studies to analyze, one with much more precision than the other two. Since this analysis primarily reflects the results of one study, the estimated heterogeneity is low.
Overall, a few studies dominate the pooled estimates. This is unsurprising given the wide range of sample size, and therefore precision, in the available studies. The validity of the meta-analysis is not affected by this, but it is an important factor in our interpretation of the results. In particular, fixed-effects meta-analysis estimates a precision-weighted average of the effect sizes in the included studies. Because of this, our estimates largely reflect the association between HCV infection and negative postoperative outcomes in the target populations of the large, high-precision studies. If these populations are representative of our population of interest, then there is no issue. For example, in the analysis of PJI (adjusted estimates only), over 99% of the weight is placed on the studies by Cancienne et al. (14) and Kildow et al. (17). Both of these studies sampled individuals from Medicare databases; our pooled OR estimate of 1.98 reflects the increased risk of PJI comparing HCV-infected to uninfected patients in the population of Medicare recipients. If we were primarily interested in younger patients, this could be cause for concern.
The meta-regression in the PJI analysis shows a substantial difference in effect size estimates between studies that did and did not account for confounding via matching or regression adjustment. As with the pooled OR estimates, this result reflects differences in this particular collection of studies; it does not necessarily predict differences between hypothetical future studies. Nonetheless, confounding is always a concern in observational studies, and we expect adjusted estimates to be less susceptible to bias.
There was a wide range of reported sample sizes, from 105 to over 8 million. The study by Best et al. (13) was the largest and used data from the National Hospital Discharge Survey. This data source comprised of a random sample of hospital discharges from non-federal short-stay hospitals in the USA. This random sample represents around 1% of all US hospital discharges. (27) The Survey produces a weighted sample, meant to represent all discharges in the USA. It is possible that Best et al. (13) study’s reported sample size of 8 million represents all TJA surgeries in the weighted sample rather than the unweighted (raw) sample. This is not made clear in the study. Given that the Survey contains around 1% of all hospital discharges, and around 1 million TJA surgeries are performed each year in the USA (28), it is unlikely that the Best et al. (13) study actually has 8 million directly observed surgeries over its 18-year duration. It is more likely that a smaller number of observed surgeries were reweighted to create a representative sample of 8 million surgeries. Consequently, we believe it is unlikely that there is substantial overlap between the Best et al. (13) and other studies.
Conclusion
This meta-analysis is the first quantitative analysis registered in the PROSPERO international database of systematic reviews to evaluate the postoperative outcomes of patients with and without chronic hepatitis C who underwent total hip and total knee arthroplasty. A parallel meta-analysis of adverse outcomes in patients with chronic hepatitis C undergoing total joint arthroplasty was performed by investigators in the People’s Republic of China with access to the China National Knowledge Infrastructure (29). Our quantitative analysis suggests that patients with chronic hepatitis C are more likely to experience adverse postoperative complications including PJI, aseptic revision, non-homebound discharge and inpatient mortality. We found that HCV infection was positively associated with all four outcomes, with estimated ORs of 1.98, 1.58, 1.31 and 9.37, respectively. For non-homebound discharge and inpatient mortality, most of the available studies did not account for potential confounders by patient demographic and health characteristics, which could limit the generalizability of the results. This analysis suggests that future investigations should be undertaken to develop potential preoperative protocols that incorporate treatment of HCV prior to undergoing elective total joint arthroplasty. These investigations will likely need to include a cost–benefit analysis of preoperative HCV treatment and determine its feasibility in everyday clinical practice. Access to healthcare resources and ethical implications of delayed surgical treatment to pursue medical risk modification must also be considered in this discussion.
Supplementary materials
This is linked to the online version of the paper at https://doi.org/10.1530/EOR-22-0117.
ICMJE conflict of interest statement
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
This work did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
References
- 1.↑
Chak E, Talal AH, Sherman KE, Schiff ER, & Saab S. Hepatitis C virus infection in USA: an estimate of true prevalence. Liver International 2011 31 1090–1101. (https://doi.org/10.1111/j.1478-3231.2011.02494.x)
- 2.↑
Rosas S, Sabeh KG, Buller LT, Law TY, Roche MW, & Hernandez VH. Medical comorbidities impact the episode-of-care reimbursements of total hip arthroplasty. Journal of Arthroplasty 2017 32 2082–2087. (https://doi.org/10.1016/j.arth.2017.02.039)
- 3.↑
Sloan M, Premkumar A, & Sheth NP. Projected volume of primary total joint arthroplasty in the U.S., 2014 to 2030. Journal of Bone and Joint Surgery. American Volume 2018 100 1455–1460. (https://doi.org/10.2106/JBJS.17.01617)
- 4.↑
Kurtz SM, Lau E, Watson H, Schmier JK, & Parvizi J. Economic burden of periprosthetic joint infection in the United States. Journal of Arthroplasty 2012 27(8)(Supplement) 61–5.e1. (https://doi.org/10.1016/j.arth.2012.02.022)
- 5.↑
Demircay E, Unay K, Bilgili MG, & Alataca G. Glove perforation in hip and knee arthroplasty. Journal of Orthopaedic Science 2010 15 790–794. (https://doi.org/10.1007/s00776-010-1547-0)
- 6.↑
Bedair HS, Schurko BM, Dwyer MK, Novikov D, Anoushiravani AA, & Schwarzkopf R. Treatment for chronic hepatitis C prior to total hip arthroplasty significantly reduces periprosthetic joint infection. Journal of Arthroplasty 2019 34 132–135. (https://doi.org/10.1016/j.arth.2018.09.036)
- 7.↑
Bendich I, Takemoto S, Patterson JT, Monto A, Barber TC, & Kuo AC. Preoperative treatment of hepatitis C is associated with lower prosthetic joint infection rates in US veterans. Journal of Arthroplasty 2019 34 S319–S326.e1. (https://doi.org/10.1016/j.arth.2019.02.052)
- 8.↑
Schwarzkopf R, Novikov D, Anoushiravani AA, Feng JE, Vigdorchik J, Schurko B, Dwyer MK, & Bedair HS. The preoperative management of hepatitis C may improve the outcome after total knee arthroplasty. Bone and Joint Journal 2019 101–B 667–674. (https://doi.org/10.1302/0301-620X.101B6.BJJ-2018-0723.R3)
- 9.↑
Cornell E, Novikov D, Niu R, Staatz K, Schwarzkopf R, & Smith EL. Hepatitis C antiviral treatment decreases all-cause complications after total joint arthroplasty regardless of the presence of fibrosis. Journal of Arthroplasty 2021 36 1551–1555. (https://doi.org/10.1016/j.arth.2020.12.025)
- 10.↑
Cheng T, Xu D, Yan Z, Ding C, Yang C, & Zhang X. Antiviral treatment for hepatitis C is associated with a reduced risk of periprosthetic joint infection: A meta-analysis of observational studies. Journal of Arthroplasty 2022 37 1870–1878. (https://doi.org/10.1016/j.arth.2022.04.027)
- 11.↑
Moher D, Liberati A, Tetzlaff J, Altman DG & PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA statement. PLOS Medicine 2009 6 e1000097. (https://doi.org/10.1371/journal.pmed.1000097)
- 12.↑
Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, et al.The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021 372 n71. (https://doi.org/10.1136/bmj.n71)
- 13.↑
Best MJ, Buller LT, Klika AK, & Barsoum WK. Increase in perioperative complications following primary total hip and knee arthroplasty in patients with hepatitis C without cirrhosis. Journal of Arthroplasty 2015 30 663–668. (https://doi.org/10.1016/j.arth.2014.11.013)
- 14.↑
Cancienne JM, Kandahari AM, Casp A, Novicoff W, Browne JA, Cui Q, & Werner BC. Complication rates after total hip and knee arthroplasty in patients with hepatitis C compared with matched control patients. Journal of the American Academy of Orthopaedic Surgeons 2017 25 e275–e281. (https://doi.org/10.5435/JAAOS-D-16-00920)
- 15.↑
Issa K, Boylan MR, Naziri Q, Perfetti DC, Maheshwari AV, & Mont MA. The impact of hepatitis C on short-term outcomes of total joint arthroplasty. Journal of Bone and Joint Surgery. American Volume 2015 97 1952–1957. (https://doi.org/10.2106/JBJS.O.00183)
- 16.↑
Issa K, Pierce TP, Harwin SF, Scillia AJ, McInerney VK, & Mont MA. Does hepatitis C affect the clinical and patient-reported outcomes of primary total hip arthroplasty at midterm follow-up? Journal of Arthroplasty 2017 32 2779–2782. (https://doi.org/10.1016/j.arth.2017.04.001)
- 17.↑
Kildow BJ, Politzer CS, DiLallo M, Bolognesi MP, & Seyler TM. Short and long-term postoperative complications following total joint arthroplasty in patients with human immunodeficiency virus, hepatitis B, or hepatitis C. Journal of Arthroplasty 2018 33 S86–S92.e1. (https://doi.org/10.1016/j.arth.2017.10.061)
- 18.↑
Mahure SA, Bosco JA, Slover JD, Vigdorchik J, Iorio R, & Schwarzkopf R. Risk of complications after THA increases among patients who are coinfected with HIV and hepatitis C. Clinical Orthopaedics and Related Research 2018 476 356–369. (https://doi.org/10.1007/s11999.0000000000000025)
- 19.↑
Novikov D, Feng JE, Anoushiravani AA, Vigdorchik JM, Lajam CM, Seyler TM, & Schwarzkopf R. Undetectable hepatitis C viral load is associated with improved outcomes following total joint arthroplasty. Journal of Arthroplasty 2019 34 2890–2897. (https://doi.org/10.1016/j.arth.2019.06.058)
- 20.↑
Orozco F, Post ZD, Baxi O, Miller A, & Ong A. Fibrosis in hepatitis C patients predicts complications after elective total joint arthroplasty. Journal of Arthroplasty 2014 29 7–10. (https://doi.org/10.1016/j.arth.2013.03.023)
- 21.↑
Pour AE, Matar WY, Jafari SM, Purtill JJ, Austin MS, & Parvizi J. Total joint arthroplasty in patients with hepatitis C. Journal of Bone and Joint Surgery. American Volume 2011 93 1448–1454. (https://doi.org/10.2106/JBJS.J.00219)
- 22.↑
Jergesen HE, Thielen ZP, Roever JA, Vashon TT, Wu HH, & Yi PH. Primary hip and knee arthroplasty in a safety net hospital: substance abuse and other factors affecting short-term complications. Journal of Arthroplasty 2018 33 3003–3008. (https://doi.org/10.1016/j.arth.2018.05.007)
- 23.↑
Mahure SA, Bosco JA, Slover JD, Vigdorchik JM, Iorio R, & Schwarzkopf R. Coinfection with hepatitis C and HIV is a risk factor for poor outcomes after total knee arthroplasty. JB & JS Open Access 2017 2 e0009. (https://doi.org/10.2106/JBJS.OA.17.00009)
- 24.↑
Sweeting MJ, Sutton AJ, & Lambert PC. What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data. Statistics in Medicine 2004 23 1351–1375. (https://doi.org/10.1002/sim.1761)
- 25.↑
Higgins JPT, & Thompson SG. Quantifying heterogeneity in a meta-analysis. Statistics in Medicine 2002 21 1539–1558. (https://doi.org/10.1002/sim.1186)
- 26.↑
DerSimonian R, & Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials 1986 7 177–188. (https://doi.org/10.1016/0197-2456(8690046-2)
- 27.↑
Bekkers S, Bot AGJ, Makarawung D, Neuhaus V, & Ring D. The National Hospital Discharge Survey and Nationwide Inpatient Sample: the databases used affect results in THA research. Clinical Orthopaedics and Related Research 2014 472 3441–3449. (https://doi.org/10.1007/s11999-014-3836-y)
- 28.↑
Williams SN, Wolford ML, & Bercovitz A. Hospitalization for total knee replacement among inpatients aged 45 and over: United States, 2000–2010. NCHS Data Brief 2015 210 1–8.
- 29.↑
Cheng T, Yang C, Hao L, Cheng X, Hu J, Ren W, & Zhang X. Hepatitis C virus infection increases the risk of adverse outcomes following joint arthroplasty: A meta-analysis of observational studies. Orthopaedics and Traumatology, Surgery and Research 2022 108 102947. (https://doi.org/10.1016/j.otsr.2021.102947)
