Many common morbidities are associated with increased mortality after elective hip and knee replacement including congestive heart failure, cardiovascular disease, anemia, diabetes, and frailty [2, 5, 10, 12], but the impact of chronic kidney disease (CKD) on life expectancy after hip or knee replacement is not well characterized. Most studies [4, 5, 8, 10, 15, 16, 19] have focused on patients with renal failure or short-term mortality, typically within 90 days. As a result, the effects of mild to severe CKD on mortality after elective surgery joint replacement have often been ignored, especially in the long term.
Kidney function is defined by the estimated glomerular filtration rate (eGFR) and is classified by severity into five different CKD stages; normal function is present when the eGFR is 90 mL/min/1.73 m2 followed by mild (89-60), moderate (59-30), severe (29-15), and renal failure (< 15 or dialysis) based on eGFR . Diabetes and hypertension along with use of nonsteroidal antiinflammatory drugs is common among patients undergoing lower extremity joint replacement and they are also associated with CKD [11, 13]. Chronic kidney disease of at least a moderate degree is found in 10% to 20% of patients undergoing elective hip and knee replacement [1, 11, 17]. In previous studies, moderate CKD increased the risk of death approximately twofold at 30 to 90 days postoperatively [4, 5, 10, 19], and patients with end-stage renal failure or renal transplant had 1-year mortality of approximately 4% to 6% . To our knowledge, only one paper describes long-term mortality across all CKD stages, but it focused on patients aged ≥ 75 years and could demonstrate a significant effect only in patients with severely reduced kidney function (eGFR < 30 mL/min/1.73 m2) who had fivefold higher overall mortality . Otherwise, longer followup data exist on patients on dialysis and patients who have undergone kidney transplantation, of whom 50% and 80% to 90%, respectively, had died 5 years postoperatively [8, 15]. Thus, there are little population-based data concerning mortality among patients undergoing joint replacement classified to have mild to severe CKD after the first postoperative months. However, the connection between moderately (eGFR < 60 mL/min/1.73 m2) and mildly (eGFR < 90 mL/min/1.73 m2) reduced kidney function and death has been demonstrated among patients undergoing cardiac surgery and those undergoing carotid endarterectomy [3, 9]. Because mortality plays a key role both in balancing the risks and benefits and determining the cost-effectiveness of surgery, surgeons need more knowledge on the effect of CKD stage on patients’ long-term prognoses. Furthermore, because CKD is usually associated with other comorbidities , knowing the interactions among the most common comorbidities on postoperative mortality is crucial in preoperative evaluation to make the decision whether to operate. Such data are, however, currently lacking.
Therefore, we asked: (1) What is the risk of postoperative mortality in different stages of CKD after elective hip or knee replacement and does the risk increase with mild renal insufficiency? (2) How severe is the risk of death in patients with CKD compared with other major medical comorbidities such as diabetes, cardiovascular disease, and congestive heart failure? (3) Are there risk factor combinations associated with especially poor survival?
Materials and Methods
This study was a retrospective study using longitudinally maintained patient data. The study was performed at the Coxa Hospital for Joint Replacement, a university-affiliated publicly funded orthopaedic hospital specializing in joint replacement surgery with > 3000 joint replacements performed annually representing approximately 15% of all hip and knee replacements performed annually in Finland. Coxa is responsible for joint replacement surgery in Pirkanmaa Hospital District (catchment area population of approximately 550,000 inhabitants) and is also a tertiary referral hospital for one-fifth of Finland. At the end of the study, Coxa had 15 senior orthopaedic surgeons. All joint replacements over the study period were performed by or under direct supervision of these senior surgeons. Since its establishment in 2002, Coxa has maintained an electronic joint replacement database. Comprehensive data of all primary and revision joint replacement surgeries are systematically recorded in the database. This database also produces obligatory reports of all joint replacements to the Finnish Arthroplasty Register that is maintained by the Institute for Health and Welfare (Helsinki, Finland). Compared with the Finnish Hospital Discharge Register, that is also maintained by the same national institute, annual completeness of Coxa’s database has continuously been > 99%. Mortality data are routinely obtained from the Finnish Population Register Centre that covers both all deaths that occur inside Finland and also separately reported deaths abroad. Thanks to these nationwide data sources, that are automatically linked to Coxa’s database, our followup data, in terms of mortality and survival, are practically complete.
This investigation was approved by the Pirkanmaa Hospital District.
The study population included patients undergoing hip or knee replacements at Coxa Hospital between September 2002 and December 2011 (n = 20,575). A total of 2000 patients were excluded from the study as a result of lacking preoperative serum creatine (SCr; unable to calculate eGFR) or having an emergency operation (to maintain homogeneity) (Fig. 1). There were fewer females (58.9% compared with 62.7%; chi square test p = 0.001), fewer knee replacements (30.8% compared with 52.5%; chi square test p < 0.001), and lower body mass index (BMI; medians, 27.5 kg/m2 versus 28.6 kg/m2; independent-sample Mann-Whitney U-test p < 0.001) in the excluded group. The remaining 18,575 patients were included in our analyses. Of the included patients, 22% (4055) died during the followup period, which was until April 9, 2016. The median followup was 7.8 years (interquartile range [IQR], 5.8-10.0 years; range, 0-13.95 years).
The preoperative diagnosis was primary osteoarthritis in 15,467 (83%) of the patients. A total of 2632 (24%) of the operations were revision joint replacements. The posterior approach was used in hip operations and the medial parapatellar approach was used in knee operations in most of the patients. Antibiotic bone cement was used in 3899 (51%) of the hip operations (1190 [14%] missing) and in 9159 (96%) of the knee operations (220 [2%] missing). The median operation time was 115 minutes. Spinal anesthesia was used with few exceptions; only 287 (2%) had general anesthesia.
Demographic data and patient information were obtained from a longitudinally maintained joint replacement database and patient administration database. The following data were collected for analysis: sex, age, BMI, surgically treated joint (hip, knee), and type of operation (primary, revision). A history of diabetes, hypertension, coronary disease, congestive heart failure, and dialysis was obtained from the Finnish national database of reimbursement rights for medication expenses. A table showing reimbursement criteria has been published in our previous study . Reimbursement for medication is a financial benefit that the Finnish government provides for patients with certain diseases and therefore most eligible patients use it. Patients receiving reimbursement for a given medication were regarded as having the relevant disease. The criteria for reimbursement for hypertension medication are higher than diagnostic criteria and therefore some patients having hypertension might not have been found. However, those registered represent patients with true, confirmed diseases. Preoperative SCr and hemoglobin values were obtained from the database of a local laboratory that provides laboratory services to the hospital, the adjacent university hospital, and to most communities in the catchment area. These laboratory tests are routinely obtained as part of the preanesthesia evaluation done 1 to 2 months before surgery, but measurements taken within 6 months before surgery were approved. If multiple preoperative SCr or hemoglobin measurements were recorded, the most recent was used. Anemia was defined as hemoglobin < 117 g/L in women and < 134 g/L in men. SCr value and patient’s age, sex, and race were used to calculate a different eGFR value using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula . All patients were presumed to be white. Patients were classified in different CKD stages according to the eGFR. Normal kidney function (CKD Stage I) was present in 6519 (35%) of the patients and 9917 (53%) had Stage II, 2023 (11%) had Stage III, 81 had Stage IV, and 35 had Stage V CKD (Table 1).
Median age was 69 years (IQR, 61-76). A total of 11,650 (63%) of the patients were female. Median BMI was 28.6 kg/m2 (Table 1).
Kaplan-Meier analysis was performed to draw survival plots and to calculate median survival times. Survival time for certain cutoff points was calculated using life tables. Univariable analysis of patient characteristics and comorbid conditions was made using Cox regression to test which variables have an association with survival. In univariable analysis, age, sex, anemia, diabetes, hypertension, coronary disease, congestive heart failure, CKD stage, and BMI were associated with survival (p < 0.05) and therefore were regarded as confounders for multivariable analysis using Cox regression. Preoperative anemia might be caused by end-stage renal disease and therefore was not included in the multivariable analysis owing to high collinearity (chi square, p < 0.001). CKD disease stage based on the CKD-EPI equation also includes information regarding age and sex and therefore was not included in the multivariable analysis. When Cox regression was used, proportional hazards assumption was tested graphically using log-log plots. For multivariable analysis, Stages IV and V CKD were combined to fulfill the proportional hazards assumption, because the curves crossed several times in the log-log plot (data not shown). In multivariable Cox regression analysis, the CKD classification was the exposure variable, whereas morbidity was the outcome variable. Different combinations of CKD and comorbid conditions were tested with logistic regression for certain cutoff points. Because diabetes, congestive heart failure, and coronary disease were associated with increased and remarkable risk for mortality, they were one at a time combined with CKD. For this analysis, CKD was defined as eGFR < 60 mL/min/1.73 m2.
Survivorship decreased with advancing stages of CKD (Stages II to V compared with Stage I) after controlling for BMI, congestive heart failure, coronary disease, diabetes, and hypertension (Table 2). The mean survival time was 13 years (95% confidence interval [CI], 12.5-12.7 years ) in Stage I, 11 years (95% CI, 11.3-11.5 years) in Stage II, 9 years (95% CI, 9.2-9.7 years) in Stage III, 7 years (95% CI, 5.6-7.5 years) in Stage IV, and 6 years (95% CI, 4.9-8.0 years) in Stage V (p < 0.001) (Fig. 2; Table 3).
Compared with other risk factors, Stages III to V CKD had greater hazard ratios (HRs; Stage III 3.76 [95% CI, 3.39-4.19]; Stage IV-V 8.08 [95% CI, 6.33-10.31]; Table 2) for mortality than congestive heart failure (2.11 [95% CI, 1.81-2.45]), coronary disease (1.54 [95% CI, 1.40-1.69]), diabetes (1.71 [95% CI, 1.54-1.90]), or hypertension (1.35 [95% CI, 1.26-1.45]) in a multivariable model including CKD stage, BMI, diabetes, hypertension, coronary disease, and congestive heart failure.
Survival was poorer when CKD was combined with diabetes, coronary disease, or congestive heart failure than CKD alone. The combination of CKD and diabetes had a synergistic effect on mortality, especially at 90 days (odds ratio [OR], 10.8 [95% CI, 4.84-24.3]) and 1 year (OR, 8.15 ;95% CI, 4.90-13.5]; Table 4).
Postoperative mortality of patients with CKD, particularly at long-term followup, is as not well characterized as that related to other comorbidities among joint replacement recipients [2, 5, 10]. Current literature focuses on the most severe extremity of the CKD scale and when the whole scale of CKD is considered, only short-term mortality has been reported [4, 5, 8, 10, 15, 16, 19]. This leaves a gap in our knowledge concerning the most common patient group, ie, patients with Stage II to III CKD, when it comes to followup beyond 90 days or 1 year. In our study, mortality of the whole scale of CKD was reported at a median followup of 8 years.
Our study had some limitations. Our study population is quite homogenous because only 6% of Finnish citizens have a foreign background and most of those are of Nordic origin (Official Statistics of Finland: Population structure [e-publication]. ISSN 1797-5395. Helsinki: Statistics Finland [referred: November 26, 2017]. Available at: http://www.stat.fi/til/vaerak/index_en.html). Our data might be skewed by a large proportion of white patients. In addition, our database does not register patient’s race and therefore we assumed all patients were white. However, the effect of this is probably negligible. Furthermore, the CKD-EPI equation takes race into account  when calculating eGFR and therefore our results can be extrapolated also to other patient groups with different racial distribution. Our data concerning diabetes, hypertension, congestive heart failure, and coronary disease were obtained from a national database, which does not include all patients with a certain disease, but only patients with diagnosed and medically treated conditions who receive reimbursement for medication costs. Because reimbursement right is a remarkable financial benefit, we assume that most patients will apply for it if they fulfill the criteria. Furthermore, use of reimbursement data ensures that the diagnoses are valid. Because the reimbursement criteria for hypertension are very strict , our results can be interpreted concerning patients with difficult hypertension and therefore patients with less severe hypertension do not necessarily have as high mortality as reported in this study. Our consideration of diabetes is limited to patients with medically treated disease, and patients treated with lifestyle interventions only are not taken into account. With these limitations, our result of patients with CKD having greater HRs for mortality compared with patients with difficult hypertension and medically treated patients with diabetes emphasizes the role of CKD as a major risk factor for death.
Severe CKD is associated with increased mortality both at short-term followup and long term [4, 5, 8, 12, 15, 16, 19], but there is a paucity of data about the effects of milder CKD. Deegan et al.  found increased (HR 3.4 with relatively wide 95% CI, 1.25-9.23) long-term mortality among patients with Stage III CKD (compared with Stage I-II CKD) among a similar but smaller group of patients undergoing hip replacements. Although they did not control their multivariable analysis for most relevant comorbid conditions, their results are about the same magnitude with the two-, four-, and eightfold increases in mortality in Stage I versus Stages II through V CKD in this study. However, Deegan et al. were not able to demonstrate any effect of CKD on mortality after knee replacements, probably as a result of a small number of patients. Jämsen et al.  studied an older population and found that patients with Stage IV CKD have a fivefold risk of death compared with patients with Stage I disease, whereas the risk of death in patients with Stages II and III disease did not differ. Smaller number of patients, shorter followup, and obviously different baseline risk of death probably explain the differences between our results (Table 5). Our study extends current knowledge by reporting HRs, with adjustment for comorbidity, separately for every CKD stage in an adequately sized population-based series. Importantly, our findings indicate that survival in patients with Stages I and II CKD is excellent and hence these patients will most likely benefit from the surgery. On the other hand, survival in patients with Stages IV to V CKD is poor and proceeding to surgery should be considered carefully because only half of the patients are alive 5 years postoperatively. Patients with Stage III CKD have a relatively good survival rate postoperatively and they will likely benefit from the operation. However, if the indication for operation is borderline between whether or not to operate or the patient has diabetes, congestive heart failure, or coronary disease, Stage III CKD might guide clinicians to make a decision not to operate. It should be recognized that although clinical benefit in terms of pain and function may be achieved, fewer quality-adjusted life-years are achieved as a result of compromised life expectancy. Therefore, future studies should evaluate the cost-effectiveness of joint replacements in patients with Stage III to V CKD.
Bozic et al.  and Hunt et al.  reported an approximately twofold increase in risk of death at 90 days among patients with renal disease and among patients with congestive heart failure. The risk with diabetes was 1.3- to twofold. Hypertension and coronary disease were not associated with 90-day mortality in the first study  and not investigated in the latter . In our multivariable analysis, Stages III to V CKD led to a four- to eightfold increase in the overall risk of death compared with Stage I disease, whereas congestive heart failure and diabetes increased the risk of death two- and 1.7-fold, respectively. Therefore, Stages III to V CKD seem to have more of an effect on mortality than diabetes, congestive heart failure, hypertension, and coronary disease, which differs from the results reported by others [5, 10]. Our followup was longer, which probably explains the differences in our study compared with Bozic et al.  and Hunt et al. . Because they did not provide information regarding patient demographics, further comparisons cannot be made. Nevertheless, our study suggests that CKD is definitely among the conditions that should be considered when planning surgery and care should be taken to recognize and grade CKD in preoperative assessment .
Our results indicating that stage of CKD is the strongest factor associated with mortality can be explained by considering this disease as a manifestation of long-lasting comorbid conditions such as diabetes and hypertension . This view is supported by the novel finding that the risk of death in patients with CKD accompanied by coronary disease or congestive heart failure is additional. However, CKD and diabetes have a synergistic effect on mortality during the first postoperative year and an additional effect at 5 years. This synergistic effect might be explained by diabetes increasing morbidity burden more than congestive heart failure or coronary disease. Previously, it has been shown that CKD increases mortality in patients with coronary disease  and diabetes  even three- to ninefold. These studies, however, lacked a comparison group with either condition. Such risk factor combinations have not been studied in joint replacement recipients either. Given the considerable additional effects on mortality, this study emphasizes the importance of paying attention to individual patients’ morbidity profiles rather than the presence of separate conditions. However, the synergistic effect of diabetes and CKD should be investigated more because our results present relatively wide CIs.
With this study, we found that all stages of CKD are associated with postoperative mortality in the long term, which was not known before. Furthermore, underlining the importance of these findings, CKD seems to be a stronger predictor of mortality than other comorbidities. Importantly, patients with CKD combined with diabetes, coronary disease, or congestive heart failure was found to have especially poor survival. We recommend surgeons consider CKD as an important risk factor among others and to take CKD stage into account when determining when it is reasonable to operate. In our opinion, in clear indications, patients having Stages I to III CKD will likely benefit from the surgery because they have a long life expectancy. However, in Stages IV and V and in patients having both diabetes and Stage III CKD, these should be considered before proceeding to surgery. Because patient survival affects the overall benefit of surgery, future studies should evaluate whether hip and knee replacements are cost-effective in patients with progressive CKD and also how the high risk of death among patients having both CKD and diabetes should be acknowledged in clinical practice. In our clinic, the information from this study helps us in counsel patients and their family members about the increased risks associated with joint replacement surgery in patients with renal insufficiency.
1. Ackland GL, Moran N, Cone S, Grocott MP, Mythen MG. Chronic kidney disease and postoperative morbidity after elective orthopedic surgery. Anesth Analg. 2011;112:1375–1381.
2. Bellamy JL, Runner RP, Vu CCL, Schenker ML, Bradbury TL, Roberson JR. Modified Frailty Index is an effective risk assessment tool in primary total hip arthroplasty. J Arthroplasty. 2017;32:2963–2968.
3. Bissacco D, Catanese V, Fossati A, Salvati S, Zanella G, Settembrini PG. Role of preoperative serum creatinine and estimated glomerular filtration rate values in asymptomatic patients undergoing carotid endarterectomy. J Cardiovasc Surg (Torino). 2015 Sep 1. [Epub ahead of print]
4. Boniello AJ, Simon MS, Emenari CC, Courtney PM. Complications and mortality following total hip arthroplasty in the octogenarians: an analysis of a national database. J Arthroplasty. 2017 Aug 31. [Epub ahead of print]
5. Bozic KJ, Lau E, Kurtz S, Ong K, Rubash H, Vail TP, Berry DJ. Patient-related risk factors for periprosthetic joint infection and postoperative mortality following total hip arthroplasty in Medicare patients. J Bone Joint Surg Am. 2012;94:794–800.
6. Deegan BF, Richard RD, Bowen TR, Perkins RM, Graham JH, Foltzer MA. Impact of chronic kidney disease stage on lower-extremity arthroplasty. Orthopedics. 2014;37:e613–618.
7. Engelbertz C, Reinecke H, Breithardt G, Schmieder RE, Fobker M, Fischer D, Schmitz B, Pinnschmidt HO, Wegscheider K, Pavenstädt H, Brand E. Two-year outcome and risk factors for mortality in patients with coronary artery disease and renal failure: the prospective, observational CAD-REF Registry. Int J Cardiol. 2017;243:65–72.
8. Goffin E, Baertz G, Rombouts JJ. Long-term survivorship analysis of cemented total hip replacement (THR) after avascular necrosis of the femoral head in renal transplant recipients. Nephrol Dial Transplant. 2006;21:784–788.
9. Howell NJ, Keogh BE, Bonser RS, Graham TR, Mascaro J, Rooney SJ, Wilson IC, Pagano D. Mild renal dysfunction predicts in-hospital mortality and post-discharge survival following cardiac surgery. Eur J Cardiothorac Surg. 2008;34:390–395; discussion 395.
10. Hunt LP, Ben-Shlomo Y, Clark EM, Dieppe P, Judge A, MacGregor AJ, Tobias JH, Vernon K, Blom AW. 90-day mortality after 409,096 total hip replacements for osteoarthritis, from the National Joint Registry for England and Wales: a retrospective analysis. Lancet. 2013;382:1097–1104.
11. Jämsä PP, Oksala NKJ, Eskelinen AP, Jämsen ER. Chronic kidney diseases among patients undergoing elective arthroplasty: risk groups and the value of serum creatinine. J Arthroplasty. 2017 Aug 9. [Epub ahead of print]
12. Jämsen E, Puolakka T, Eskelinen A, Jäntti P, Kalliovalkama K, Nieminen J, Valvanne J. Predictors of mortality following primary hip and knee replacement in the aged: a single-center analysis of 1,998 primary hip and knee replacements for primary osteoarthritis. Acta Orthop. 2013;84:44–53.
13. Levey AS, Coresh J, Balk E, Kausz AT, Levin A, Steffes MW, Hogg RJ, Perrone RD, Lau J, Eknoyan G; National Kidney Foundation. National Kidney Foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Ann Intern Med. 2003;139:137–147. Erratum in Ann Intern Med. 2003;139:605.
14. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612.
15. Lieberman JR, Fuchs MD, Haas SB, Garvin KL, Goldstock L, Gupta R, Pellicci PM, Salvati EA. Hip arthroplasty in patients with chronic renal failure. J Arthroplasty. 1995;10:191–195.
16. Lieu D, Harris IA, Naylor JM, Mittal R. Review article: Total hip replacement in haemodialysis or renal transplant patients. J Orthop Surg (Hong Kong). 2014;22:393–398.
17. Perregaard H, Damholt MB, Solgaard S, Petersen MB. Renal function after elective total hip replacement. Incidence of acute kidney injury and prevalence of chronic kidney disease. Acta Orthop. 2016;87:235–238.
18. Salinero-Fort MÁ, San Andrés-Rebollo FJ, de Burgos-Lunar C, Abánades-Herranz JC, Carrillo-de-Santa-Pau E, Chico-Moraleja RM, Jiménez-García R, López-de-Andrés A, Gómez-Campelo P. Cardiovascular and all-cause mortality in patients with type 2 diabetes mellitus in the MADIABETES Cohort Study: association with chronic kidney disease. J Diabetes Complications. 2016;30:227–236.
19. Warth LC, Pugely AJ, Martin CT, Gao Y, Callaghan JJ. Total joint arthroplasty in patients with chronic renal disease: is it worth the risk? J Arthroplasty. 2015;30(Suppl):51–54.