Effect of chronic kidney disease on total knee arthroplasty outcomes: a meta-analysis of matched control studies

Purpose The purpose of this meta-analysis was to review the current evidence in the literature to find out whether the coexisting chronic kidney disease affected infection, revision, transfusion, readmission, mortality, and the length of hospital stay after total knee arthroplasty. Methods Medline, PubMed, Embase, and the Cochrane Library were searched from their dates of inception to June 30, 2020. The primary outcomes were postoperative infection, revision, and mortality. The secondary outcomes were transfusion, the length of hospital stay, and readmission. A P value of < 0.05 was deemed to be statistically significant. Results A total of 881 articles were identified, and 7 articles that met the inclusion criteria were identified to be eligible. The most important finding of our study was that the chronic kidney disease was associated with increased postoperative transfusion (P < 0.05) and mortality (P < 0.05). Meanwhile, the patients with chronic kidney disease were associated with a higher readmission rate, compared to the patients without chronic kidney disease (P < 0.05). However, chronic kidney disease was not associated with high risks for infection (P > 0.05), revision surgeries (P > 0.05), and a prolonged hospital stay (P > 0.05). Conclusions After total knee arthroplasty, the patients with coexisting chronic kidney disease carry higher risks of transfusion, mortality, and readmission. However, the chronic kidney disease may not be associated with the risk of infection or revision, nor the duration of hospitalization.


Introduction
Total knee arthroplasty (TKA) is a common surgical procedure for the end-stage knee osteoarthritis [1,2], while the prevalence of chronic kidney disease (CKD) is increasing worldwide [3]. Recent studies suggest that the TKA patients with CKD carry higher risks of postoperative complications [4][5][6]. However, a single study compromises the reliability of the conclusions due to its relatively small amount of data and statistical power.
Kidney function is defined using the estimated glomerular filtration rate (GFR). The severity is classified into five stages based on the level of estimated GFR (stage I, > 90 mL/min; stage II, 60-89 mL/min; stage III, 30-59 mL/min; stage IV, 15-29 mL/min; stage V renal failure, < 15 mL/min) [7]. Patients with end-stage CKD must receive dialysis or kidney transplantation, which augment the potential risks of complications after TKA. In addition, the CKD is associated with the comorbidities such as anemia, diabetes mellitus, hypertension, weakened immune system. Each of them is an independent risk factor for postoperative complications [8]. DiMagno et al. [3] found that patients with stage III, IV, or V CKD are at greater risks of complications after TKA.
Gwam et al. [4] found the dialysis-dependent patients are associated with a prolonged hospital stay, but do not carry the risk of 30-day complications. Antonia et al. [5] suggested stage III and IV CKD were associated with the major postoperative complications. Currently, no metaanalysis elucidated the outcomes of TKA in CKD patients.
The purpose of this meta-analysis was to review the current evidence in the literature to find out whether the CKD affects the infection, revision, transfusion, readmission, mortality, and the length of stay (LOS) after TKA. The primary outcomes were infection, revision, and mortality. The secondary outcomes were transfusion rate, the length of hospital stay, and readmission. This is the first meta-analysis to assess the outcomes of TKA in CKD patients.

Search strategy
The systematic literature review was structured to adhere to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement, which included the requirements deemed essential for the transparent reporting of results [9]. Ethical approval was not needed because all the data presented in this study were extracted from the published articles and did not cover any personal data. We searched the online databases including Medline, PubMed, Embase, and the Cochrane Library from their dates of inception to June 30, 2020. We searched the following terms: (

Eligibility criteria
The articles included in the meta-analysis met all of the following inclusion criteria (in the PICOS order): (1) population: patients undergoing primary TKA or minimally invasive TKA; (2) intervention: CKD group; (3) comparison intervention: non-CKD group; (4) outcome measures: at least one of the following outcomes was reported: infection, revision, transfusion, the length of stay (LOS), re-admission, and mortality; (5) study design: prospective study, retrospective study or registry study. The articles that did not assessthe outcomes mentioned above were excluded. The articles evaluating the revision surgery or unicompartmental knee arthroplasty, involving no comparison between CKD and non-CKD or combining outcomes of TKA and total hip arthroplasty were excluded. The conference abstracts, case reports, biochemical trials, letters, and reviews were also excluded.

Data extraction
Two reviewers (CJ and YY) independently checked the full text of the selected articles. Raw information including characteristics of the articles (author, publication year, study design, and follow-up period) and patients' demographic details (the number of patients, average age, gender ratio, and stage of CKD) were extracted. The primary outcomes were the incidence of postoperative complications, including dislocation, infection, deep vein thrombosis (DVT), pulmonary embolism (PE), revision surgeries, and mortality. The secondary outcomes included re-admission, transfusion, and duration of hospitalization. If data were incomplete or could not be extracted directly, we contacted the corresponding authors to ensure the integrity of the data.

Statistical analysis
We used the Review Manager software (v 5.3; Cochrane Collaboration) to perform the meta-analysis. The extracted data were independently entered into the Review Manager by one author and independently checked by another author. The results of the selected studies were pooled for meta-analysis when two or more results were available. We applied the Mantel-Haenszel method to compute the pooled odds ratio (OR). An OR with a 95% confidence interval (CI) or a mean difference with a 95% CI was categorized as dichotomous outcomes or continuous outcomes, respectively. A P value of < 0.05 was deemed to be statistically significant. We used the Q and chi² tests to estimate the P value and I² of the heterogeneity measures. All outcomes were pooled on the random-effect model.

Quality evaluation
Because no randomized trials were included, the Newcastle-Ottawa scale (0 = very poor to 9 = rigorous) was applied to evaluate the quality of non-randomized studies. The Newcastle-Ottawa scale-based methodological quality assessment was conducted in three domains: study selection, intergroup comparability, and exposure (Table 1). A higher score indicated a better quality of an article. Two reviewers (CJ and YY) independently assessed the included articles, and disagreements were resolved by discussion with a third reviewer (WS). The sensitivity analysis was also conducted to evaluate whether any single study had the weight to skew on the whole estimate and data. Publication bias analyses were not performed due to the relatively scant literature.

Search results and study characteristics
A total of 881 articles were initially identified from the online registerry databases, and 858 articles were excluded after primary review of the titles and abstracts. After further full-text evaluation, 16 articles that did not meet the inclusion criteria were excluded. Finally, 7 articles were included in the meta-analysis (Fig. 1) [6,[10][11][12][13][14][15]. The baseline characteristics of the studies and patient demographic details are shown in Table 2. Among them, 6 articles [6, 10-12, 14, 15] were defined as having CKD using the international classification of diseases (ICD) or diagnostic codes on their electronic medical records or registry databases. One article [13] was defined as having CKD by calculating the estimated GFR (< 60 mL/min/1.73 m 2 ; stage II-V). All 7 studies were matched control studies. The Newcastle-Ottawa scales of the included studies ranged from 6 to 9, indicating that the quality of the included studies was at the upper-middle level, and the outcomes of the included studies could be considered to be reliable (

Discussion
Recent studies showed that the prevalence rates of CKD in TKA patients are between 6 % and 27 % [16,17]. In this study, we found the CKD patients carried higher risks of postoperative mortality and readmission, compared to the non-CKD patients. The CKD patients had a higher probability of postoperative transfusion than their non-CKD counterparts. However, the CKD did not affect the risks of postoperative infection, revision or LOS. Ponnusamy et al. [11] found the CKD affected bone volume, mineralization, linear growth, and strength [18]. Tan et al. [8] demonstrated the patients with end-stage CKD carried a higher risk of VTE, because the increased inflammatory state produces the hypercoagulability [19]. The chronic inflammation produces excessive fluid in the soft tissue surrounding the knee and increases the risks of periprosthetic infection and revision surgery [20]. The CKD is often associated with a poor nutritional status, electrolyte disorders, decreased immunity, and anemia [1,8,21,22]. Those factors also increase the risk of postoperative infection. McCleery et al. [10] reported that the CKD increased the risk of early periprosthetic infection by 50 %.
However, there are different conclusions. Chen et al. both CKD and non-CKD patients [1]. Wang et al. [24] did not find an increased infection rate even in the patients who experienced a long period of dialysis vintage.
McCleery et al. [10] found the patients on dialysis had an increased revision rate within one year after TKA, but the result is similar to the revision rate of non-CKD patients. However, Miric et al. [6] found there was no difference between the early and later revision rates. The most common cause of revision TKA is the periprosthetic joint infection, accounted for 62 % of all revision TKAs [25]. We found the similar revision rates in the CKD and the non-CKD patients. However, confirming the actual causes of revision is often difficult, which affects the assessment of the actual revision rates.
Patients with CKD and on long-term dialysis are often associated with renal anemia, metabolic imbalance, elevated risk of bleeding, and poor vascular circulation [26]. Graves et al. [27] showed that the degree of anaemia was more pronounced in patients with more advanced CKD. Moreover, the intra-and postoperative blood loss ranges from 800 mL to 1000 mL [28]. Kaiser et al. [29] found the CKD patients had a higher rate of transfusion (24 %) compared to non-CKD patients (8 %). Our study also supports this conclusion.
Death is a rare and devastating complication of TKA, but the early reported mortality rates range from 17 to 58 % in CKD patients [6,30,31]. Ponnusamy et al. [11] reported athe mortality rate of 1 % in dialysis-dependent patients. Several articles reported that the CKD is the independent risk factor of mortality in the early 90 days, and is the independent risk factor of morbidity in the early 30 days [13,32,33]. Warth et al. [16] demonstrated the incremental mortality was associated with an increased risk of postoperative pulmonary and cardiovascular complications. Our meta-analysis supports these findings.
Our sensitivity analysis was conducted among the outcomes with a high heterogeneity. Removing any single study did not change the statistical results. Therefore, we believe our findings in this meta-analysis are reliable.
Our meta-analysis has limitations. First, the retrospective studies included may lead to potential biases. Second, analyzing all CKD patients together without considering the severity may cause sampling bias, because there is a dosedependent relationship between the severity of CKD and outcomes of TKA [6]. Third, owing to the relatively short duration of follow-up, infection and revision rates were potentially underestimated. Fourth, the presence of publication bias is likely to decrease our confidence in the metaanalytic findings [12]. Fifth, each comorbidity (diabetes, heart failure, peripheral vascular disease, hypertension, etc.) is analysed as an independent risk factor, but overlooking the fact that the patients with multimorbidities may lead to selection bias [3]. Sixth, in the future, more high quality and matched control studies should be conducted to improve the statistical efficiency and precision.

Conclusions
Compared to the non-CKD patients, CKD patients carry higher risks of mortality and re-admission after TKA, as well as a higher incidence of transfusion. However, the CKD may not increase the risk of infection or revision, nor the duration of hospitalization.