The Journal of Arthroplasty Vol. 25 No. 5 2010
Cryotherapy After Total Knee Arthroplasty A Systematic Review and Meta-Analysis of Randomized Controlled Trials Sam Adie, BSc(Med), MBBS, MSpMed,*y Justine M. Naylor, BAppSc, PhD,*y and Ian A. Harris, MBBS, MMed (Clin Epi), PhD, FRACS*y
Abstract: Cryotherapy has theoretical and practical applications in the reduction of pain, swelling, and blood loss after trauma. We performed a systematic review and meta-analysis of randomized controlled trials on the efficacy of cryotherapy after total knee arthroplasty (TKA). Eleven studies involving 793 TKAs were included. There was considerable clinical and methodological heterogeneity. Cryotherapy resulted in small benefits in blood loss and discharge knee range of motion. There were no benefits in transfusion and analgesia requirements, pain, swelling, length of stay, and gains in knee range of motion after discharge. Despite some early gains, cryotherapy after TKA yields no apparent lasting benefits. Patient-centered outcomes remain underinvestigated. The current evidence does not support the routine use of cryotherapy after TKA. Keywords: total knee arthroplasty, cryotherapy, cold therapy, Cryo/Cuff. © 2010 Elsevier Inc. All rights reserved.
Total knee arthroplasty (TKA) has become the mainstay for surgical management of end-stage arthritis of the knee . Large improvements in pain, mobility, function, and health-related quality of life are reported within 3 to 6 months after surgery and may be evident for 7 to 15 years [2-6]. Despite the significant gains in the subacute phase, TKA is associated with pain , local edema, and significant blood losses [8,9] in the acute postoperative phase. Such outcomes are purported to be important to manage as they have implications for analgesia and transfusion requirements, joint mobility, length of hospital stay, and early functional recovery. Cryotherapy involves the application of cold, using bags of ice or cooled water, to the skin surrounding an injury and is used to minimize the extent of tissue trauma after knee surgery [10,11]. The cold penetrates the soft tissues and when applied over a joint reduces intra-articular temperatures , thereby slowing the conduction of nerve signals  and reducing local blood flow. These changes, in turn, reduce both the transmission of noxious
From the *South West Sydney Clinical School University of New South Wales Sydney, New South Wales, Australia; and yWhitlam Orthopaedic Research Centre Liverpool Hospital Liverpool, New South Wales, Australia. Submitted March 9, 2009; accepted July 6, 2009. No benefits or funds were received in support of the study. Reprint requests: Sam Adie, BSc(Med), MBBS, MSpMed, Orthopaedic Department, Liverpool Hospital Locked Bag 7103, Liverpool BC NSW Australia 1871. © 2010 Elsevier Inc. All rights reserved. 0883-5403/2505-0006$36.00/0 doi:10.1016/j.arth.2009.07.010
signals and the inflammatory response, subsequently affecting the level of blood loss, local swelling, and perceived pain . Cryotherapy has a number of proposed advantages. It is a safe and relatively cheap intervention. Commercially available devices such as the Cryo/Cuff (Aircast, Vista, CA), which combine cooling and compression, cost around US$100 but may be labor intensive because ice exchange is regularly required. More sophisticated automated devices are also available, although these are substantially more expensive. There is also evidence that patient perception of cryotherapy is overwhelmingly positive, although this may not be shared by health professionals . Cryotherapy after TKA has been observed in the United Kingdom and Australia to be standard care in some facilities yet infrequently used in others [16,17]. Conflicting evidence as to the value of cryotherapy from several randomized trials may contribute to the practice disparity. Practice disparity between health care providers for some clinical scenarios, such as the provision of ongoing rehabilitation, may be a legitimate result of disparity in resources or it may be consequent to thoughtful deliberation by providers to meet individual patient needs. However, for interventions such as the application of cryotherapy over a short period in the acute-care period, practice variance is less justifiable. Our aim was to evaluate the effectiveness of cryotherapy after TKA on postoperative blood loss, transfusion requirements, pain, range of motion, swelling, function, and length of stay.
710 The Journal of Arthroplasty Vol. 25 No. 5 August 2010
Materials and Methods We conducted a systematic review and meta-analysis using the guidelines of the Cochrane Collaboration , and our findings were reported according to the Quality of Reporting of Meta-analyses statement . Searching We performed an electronic search of the Cochrane Library (second quarter 2008), MEDLINE (1950-2008), EMBASE (1980-2008), CINAHL (1982-2008), and Web of Science (1900-2008). The electronic search strategy is available from the authors. We searched only published articles and placed no restrictions on time or language of publication. For included articles, reference lists and the “related articles” function on PubMed (www.pubmed. gov) were also assessed for possible inclusions. Study Selection We included articles that (i) had a study population treated with primary TKA; (ii) compared patients who had any form of cryotherapy in the acute phase (defined as the application of cold temperatures to the skin around the operated knee within 48 hours of surgery) to any control; (iii) included any of our outcomes; and (iv) was a randomized trial (where the allocation sequence was truly unpredictable, such as computer generated random sequences) or quasi-randomized trial (where the allocation sequence may be predictable, such as those based on medical record numbers or birth dates). Two authors independently assessed the electronic search results. When an article title seemed relevant, the abstract was reviewed for eligibility. If there was any doubt, the full text of the article was retrieved and appraised for possible inclusion. Any differences among the 2 authors were discussed, and if necessary, the third author was referred to for arbitration. Quality Assessment Assessment for risk of bias of included articles was performed individually by 2 authors according to the Cochrane guidelines . The reporting of individual methodological domains (randomization sequence, allocation concealment, blinding, and conflicts of interest) was graded as (i) adequate (where methods were reported and appropriate), (ii) inadequate (where methods were reported but were inappropriate), (iii) or unclear (where methods were not reported). To avoid inherent problems with scale validity , we did not use a quality scale or checklist. Data Abstraction Two authors independently extracted the data from included articles, and differences were discussed. Data included the general characteristics of each study and the outcomes measured. General characteristics included study design, year of publication, participants, place of study, nature of the cryotherapy intervention (involving the application of cold alone or in combination with a
form of compression dressing), time of application of cryotherapy (whether immediately postoperatively or delayed, and the frequency of application), and the nature of the control (involving the application of a compression dressing or not). Outcomes measured included postoperative blood loss (measured in milliliter of drain output or drop in hemoglobin), transfusion rate, pain (measured using the visual analogue scale, reported out of 10 or 100), analgesia use (measured in milligram per kilogram of opiate analgesic consumed postoperatively), range of motion (measured as the degrees of knee flexion in the operated knee at different stages postoperatively), swelling (measured as the circumference or the change in circumference in millimeter around a fixed point of the knee at different stages postoperatively), function (using a validated score), and length of stay (measured in days spent in an acute care facility postoperatively). We also included reported adverse events as an outcome measure. One included article was in German and was translated by an independent clinician fluent in German. We also attempted to contact the authors of included articles for additional information, which in most cases was regarding an unreported measure of uncertainty such as a standard deviation or P value. Statistical Methods For continuous outcomes, the mean and standard deviations were extracted. Where standard deviations were not reported, we calculated them using the available P value. The most conservative measure was used if this was applied  (eg, for a P value reported as b.05, 0.05 was used to calculate the standard deviation). Where appropriate, we pooled results for comparable outcomes using the inverse variance method for continuous measures and the Mantel-Haenszel method for dichotomous variables. Where outcomes between studies used the same scale, mean differences were used. When no standard deviation or other measure of uncertainty was reported for an outcome, we imputed the standard deviation(s) using the average of the reported standard deviations . This was only performed when most studies had reported a measure of uncertainty. We were also able to calculate standard deviations for one included study when a range and median were provided . Heterogeneity between trials was estimated with the I2 statistic. To investigate heterogeneity, we determined a priori to perform a subgroup analysis of studies comparing different forms of cryotherapy (using the application of cold alone or in combination with compression of the knee) to different controls (with or without compression bandaging) and an analysis according to the frequency of application of the cryotherapy intervention. Fixed effects analysis was the default model used to pool data, but where there was substantial heterogeneity (defined as I 2 N50%), a sensitivity analysis was performed, pooling the results
Cryotherapy After TKA Adie et al
using the random effects model of DerSimonian and Laird . Funnel plots were examined for evidence of publication bias. Where pooling of outcomes was inappropriate, for example, if different times and methods of measurement were used or not all necessary data were reported, we calculated absolute and relative differences to allow a more coherent qualitative interpretation of the data. Review Manager (The Nordic Cochrane Centre, The Cochrane Collaboration, 2008) Version 5.0.0 was used for data analysis.
Results The electronic search strategy yielded 396 articles, 25 of which were isolated for further assessment. Eleven studies [24-34] met our inclusion criteria and were included in this systematic review (Fig. 1). Detailed characteristics of the included studies are available from the authors. A total of 793 TKAs in 735 patients (58 patients had bilateral TKAs) were examined in these trials. Clinical Heterogeneity Although patients had similar characteristics, there was considerable clinical heterogeneity between studies. The form of intervention and control differed: 6 studies compared an intervention that combined the application of cold with a form of compression [25,26,29,30,32,34] to a control that applied compression alone to the knee. Two studies compared an intervention combining cold and compression to a control that had neither [24,28]. Four studies compared an intervention that applied only cold (such as cooling pads) to a control that had neither cold Fig. 2. Risk of bias summary: review authors' judgments about each risk of bias item for each included study. + is “yes”, − is “no”, ? is “unclear”.
nor compression [24,27,31,33]. Three studies examined variations of cryotherapy (with or without compression , different frequencies of application  or different temperature settings ) in different intervention groups. Wherever possible, data from these intervention groups were analyzed separately (which explains why some studies appear twice in our analysis). A variety of implants were used, and the use of cement fixation was reported in 5 studies [24,26,27,33,34]. Two studies used continuous passive motion in their postoperative regimen [26,33], whereas 2 others stated that it was not used.
Fig. 1. Flowchart depicting method of trial selection.
Methodological Heterogeneity Fig. 2 provides a summary methodological domain assessment for each included study. Reported scientific methodology was often unclear. Blinding for cryotherapy is difficult due to the nature of the intervention but was adequately achieved in 3 studies [27,30,31] where the temperature settings (eg, cold vs room temperature) on automated devices were set by a third party and concealed from participants and outcome assessors. Allocation concealment was unclear in all but 4 studies
712 The Journal of Arthroplasty Vol. 25 No. 5 August 2010 Table 1. Summary of Pooled Outcomes Outcome Blood loss Transfusion rate Pain day 1 postoperative Pain day 2 postoperative Pain day 3 postoperative Analgesia use Range of motion at discharge Length of stay
Unit of Measurement
95% Confidence Interval
Milliliter of blood Blood transfusion required (yes/no) Visual analogue scale, 0 (best) to 10 (worst) Visual analogue scale, 0 (best) to 10 (worst) Visual analogue scale, 0 (best) to 10 (worst) Milligrams per kilogram of narcotic analgesia Degrees of flexion
MD = −117 OR = 0.56
−216 to −19 0.24 to 1.29
MD = −0.36
−0.83 to 0.11
MD = −1.18
−1.75 to −0.62
MD = −0.19
−0.63 to 0.26
MD = −0.05
−0.16 to 0.06
MD = 6
1.6 to 10.8
MD = −0.07
−1.1 to 0.9
Heterogeneity (I2 Statistic)
MD indicates mean difference; OR, odds ratio.
[28-30,34]. It is also notable that 3 studies [26,29,30] had industry support for their research. Study size ranged from 24 to 98 patients. It is interesting to note that 3 studies [26,29,34] included data from patients who had bilateral procedures. Randomization of the intervention was often not performed in these patients for practical reasons. Instead, the intervention was applied to one knee and the control to the other. There was no evidence of statistical adjustment for these nonindependent results, which are a potential source of bias. A variation in the technique of outcome measurement (such as range of motion and swelling) was also noted, and tests of observer validity were not performed. Outcomes Measured A summary of pooled outcomes are available in Table 1. Small, statistically significant benefits favoring cryotherapy were found for blood loss (Fig. 3) and range of motion at discharge (Fig. 4). There was also a marginal benefit in pain reduction on day 2 postoperatively, but this benefit was not seen on days 1 and 3. No significant benefits were found for transfusion rate, analgesia use, or length of stay.
Variation in the time and/or method of measurement of the following outcomes precluded pooling of results: Four studies measured knee flexion later, between 6 weeks and 3 months postoperatively, and found no difference between groups. Four studies measured postoperative swelling, comparing the same form of intervention and control cryotherapy plus compression to compression alone. All studies showed similar results in the intervention and control groups, and subsequently no significant benefit was found. Function was measured in 1 study , although the outcome was entirely not reported. Most studies monitored adverse events, and these are summarized in Table 2. There was no evidence of a consistent difference in adverse events either in the presence or absence of cryotherapy. Subgroup and Analyses There were no significant subgroup differences in studies measuring blood loss when form (P = .96) or frequency (P = .12) of application of cryotherapy was examined. A sensitivity analysis was performed comparing blood loss in studies that stated use of cemented
Fig. 3. Forest plot of individual and pooled weighted mean differences in blood loss measured in milliliters of drain output. A negative mean difference favors the cryotherapy group. Random effects model used to pool data. SD indicates standard deviation; IV, inverse variance method; CI, confidence interval.
Cryotherapy After TKA Adie et al
Fig. 4. Forest plot of individual and pooled weighted mean differences in degrees of knee flexion at time of discharge. A positive mean difference favors the cryotherapy group. Fixed effects model used to pool data. SD indicates standard deviation; IV, inverse variance method; CI, confidence interval.
prostheses and those that did not, but no significant difference in these subgroups was found (P = .61). There was no significant subgroup difference in studies measuring analgesia use when form of cryotherapy was examined (P = .97). However, frequency of cryotherapy application appears to confer a benefit when cold application is regularly exchanged rather than when cold is applied continuously (P = .006). There was no significant subgroup difference in studies measuring range of motion when use of continuous passive motion was examined (P = .29). However, when form of intervention/control was examined, cryotherapy plus compression had additive range of motion benefits to compression alone, which was not apparent in studies comparing cryotherapy alone to neither cold nor compression (P = .03). However, inferences based on these analyses are limited by the small study numbers.
Table 2. Summary of Reported Adverse Events Study Albrecht 1997
Kullenberg 2006 Levy 1993
Radkowski 2007 Scarcella 1995 Smith 2002 Walker 1991 Webb 1998
Adverse Events 5 patients reported discomfort requiring discontinuation of cryotherapy; 2 patients had local skin reactions. No direct complications. Two superficial infections occurred in each of the cryotherapy and control groups. Absence of thermal injuries, infection, wound problems, deep venous thromboses, or other complications 1 patient complained of discomfort requiring removal of cryotherapy. Another patient complained of cold but continued with cryotherapy. “No anesthetic, surgical or medical complications.” 1 patient developed a deep venous thrombosis and 1 patient developed a superficial infection in the cryotherapy group. 1 patient in the control group also developed a deep venous thrombosis “No adverse events were noted in either group.” 4 patients complained the cryotherapy device was “too bulky.” Adverse events not reported. “No complications attributable” to cryotherapy. Adverse events not reported.
Discussion This systematic review and meta-analysis included 11 prospective randomized controlled trials. We found that cryotherapy has small benefits for blood loss and early range of motion, whereas no benefits were seen in transfusion requirements, pain, analgesia use, swelling, and length of stay. Included studies were of similar size and quality. Reported methodology was often unclear, and only 1 study  conformed to the Consolidated Reporting of Trials statement . We anticipated the problem of different forms of cryotherapy (which included an aspect of compression) being compared to different controls. We also anticipated that cryotherapy may be applied at different temperatures and frequencies. These sources of heterogeneity were explored using subgroup and sensitivity analyses. A visual examination of funnel plots showed no evidence of asymmetry. We did not perform statistical tests for publication bias since there were few studies, the studies were all of similar size and methodology, and the funnel plots were visually symmetrical. Blood loss is a significant problem after TKA, with estimates of up to 1500 mL of total blood loss reported . Blood transfusion may be considered necessary in some patients to avoid symptomatic anemia and subsequent delays in postoperative rehabilitation and discharge. This review showed a small reduction in postoperative blood loss in the cryotherapy group. However, the decreased blood loss is of uncertain significance because the more important clinical correlate—the need for blood transfusion—was only measured in 2 studies and was not significantly improved with cryotherapy. A reduction of allogeneic blood transfusion with cryotherapy use would be an important finding because donor blood transfusion, in addition to the financial costs, is associated with delayed wound healing, increased length of stay , and higher rates of infection . We pooled data on pain relief using cryotherapy after TKA according to time postoperatively. Although a benefit was found on day 2, this finding was not shared on days 1 and 3. Pain after TKA is treated with potent techniques such as regional nerve blocks and narcotic analgesia, and therefore, a more useful clinical correlate is
714 The Journal of Arthroplasty Vol. 25 No. 5 August 2010 analgesia use. However, no statistically significant benefit was found for postoperative narcotic consumption here. Significant reductions in postoperative pain using cryotherapy after anterior cruciate ligament reconstruction have been reported . Similar to our findings, pain reductions were reported in a few studies and only in the early postoperative period. Restricted knee range of motion and knee swelling are potential impediments to early rehabilitation and discharge. We found only a small, transient benefit from cryotherapy on knee flexion in the early postoperative phase. Because early gains in knee motion were not maintained, other factors downstream (for example, habitual activity or posturing) may have overridden the early benefits of cryotherapy. Likewise, no benefit on knee swelling was observed. The integrity of the wound after TKA is paramount, and swelling has long been postulated to impair wound healing secondary to its effects on perfusion and/or diffusion distances for oxygen . All 4 studies that measured this outcome compared cryotherapy plus compression to compression bandaging alone and suggest that the application of cold yields no additive benefits to swelling reduction when compression is also applied. Most studies in this review did not report any adverse events directly related to cryotherapy. A few patients required the cessation of the intervention due to discomfort, and there were only few reports of superficial wound infections, which also occurred in corresponding control groups. Notably, patient self-reported function, health-related quality of life, and satisfaction outcomes were not reported in included studies. These outcomes measure aspects that are of greater importance to patients. Therefore, although benefits in blood loss and range of motion may be demonstrated, these are of uncertain clinical significance if they do not result in improved function, general health, or satisfaction. Our findings are mainly limited by the quality and number of included studies. Trial methods and required data were not always reported, which required calculation or imputation of values. However, we did not find any evidence of systematic differences in effect estimates when studies were adequately reported. This study has several strengths. First, we attempted to pool findings from 11 randomized controlled trials that were likely to be underpowered to detect differences in outcomes. Second, we used stringent methods and performed a thorough assessment for the different forms of bias likely to exist in small studies. Third, this study synthesizes the strength of the evidence on a topic that is of high clinical relevance to the practice of TKA. Our main finding is that cryotherapy after TKA has uncertain clinical benefits. Considering the frequency of TKA surgery, the potential convenience and advantages of cryotherapy, and the lack of large randomized trials
with adequate methodology, these highlight the need for further research. Ideally, a future investigation would compare a cryotherapy intervention to a placebo control (or use simultaneous bilateral TKAs) and be adequately powered to detect differences in total body blood loss, transfusion rate, pain, equivalent narcotic analgesia use, and patient-centered outcomes after discharge.
Acknowledgments The authors would like to acknowledge Ms Karen Andrews of the Ken Merten Medical Library, Liverpool Hospital, for her assistance in developing the electronic search strategy.
References 1. Felson DT, Lawrence RC, Hochberg MC, et al. Osteoarthritis: new insights. Part 2: treatment approaches. Ann Intern Med 2000;133:726. 2. Naylor JM, Harmer AR, Heard RC, et al. Patterns of recovery following knee and hip replacement in an Australian cohort. Aust Health Rev 2009;33:124. 3. Bourne RB, McCalden RW, MacDonald SJ, et al. Influence of patient factors on TKA outcomes at 5 to 11 years followup. Clin Orthop 2007;464:27. 4. Harris WH, Sledge CB. Total hip and total knee replacement (2). N Engl J Med 1990;323:801. 5. Hawker G, et al. Health-related quality of life after knee replacement. J Bone Joint Surg Am 1998;80:163. 6. Sorrells RB, Voorhorst PE, Murphy JA, et al. Uncemented rotating-platform total knee replacement: a five to twelveyear follow-up study. J Bone Joint Surg Am 2004;86-A:2156. 7. Jamison RN, Ross MJ, Hoopman P, et al. Assessment of postoperative pain management: patient satisfaction and perceived helpfulness. Clin J Pain 1997;13:229. 8. Berman AT, Geissele AE, Bosacco SJ. Blood loss with total knee arthroplasty. Clin Orthop 1988;137. 9. Sehat KR, Evans RL, Newman JH. Hidden blood loss following hip and knee arthroplasty. Correct management of blood loss should take hidden loss into account. J Bone Joint Surg Br 2004;86:561. 10. Bert JM, Stark JG, Maschka K, et al. The effect of cold therapy on morbidity subsequent to arthroscopic lateral retinacular release. Orthop Rev 1991;20:755. 11. Raynor MC, Pietrobon R, Guller U, et al. Cryotherapy after ACL reconstruction: a meta-analysis. J Knee Surg 2005;18:123. 12. Martin SS, Spindler KP, Tarter JW, et al. Does cryotherapy affect intraarticular temperature after knee arthroscopy? Clin Orthop 2002;184. 13. Abramson DI, Chu LS, Tuck Jr S, et al. Effect of tissue temperatures and blood flow on motor nerve conduction velocity. JAMA 1966;198:1082. 14. Matsen III FA, Questad K, Matsen AL. The effect of local cooling on postfracture swelling. A controlled study. Clin Orthop 1975;201. 15. Lombardi A, Drouillard PD, Mallory TH, et al. The efficacy of continuous cold therapy in the postoperative management of total knee arthroplasty patients. Orthop Trans 1993;17: 157.
Cryotherapy After TKA Adie et al 16. Barry S, Wallace L, Lamb S. Cryotherapy after total knee replacement: a survey of current practice. Physiother Res Int 2003;8:111. 17. Naylor J, Harmer A, Fransen M, et al. Status of physiotherapy rehabilitation after total knee replacement in Australia. Physiother Res Int 2006;11:35. 18. Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0. The Cochrane Collaboration. Available from www.cochrane-handbook. org, 2008. 19. Moher D, Cook DJ, Eastwood S, et al. Improving the quality of reports of meta-analyses of randomized controlled trials: the QUOROM statement. Quality of reporting of metaanalyses. Lancet 1999;354:1896. 20. Higgins J, Altman D. Assessing risk of bias in included studies. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0. The Cochrane Collaboration. Accessed June 2008 at www.cochrane-handbook.org, 2008. 21. Higgins J, Deeks J. Selecting studies and collecting data. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0. The Cochrane Collaboration. Accessed June 2008 at www.cochrane-handbook.org, 2008. 22. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol 2005;5:13. 23. Deeks J, Higgins J, Altman D. Analysing data and undertaking meta-analyses. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0. The Cochrane Collaboration. Accessed June 2008 at www. cochrane-handbook.org, 2008. 24. Albrecht S, LeBlond R, Kohler V, et al. Cryotherapy as an analgetic technique in direct postoperative management of elective joint replacement. Z Orthop Ihre Grenzgeb 1997; 135:45. 25. Gibbons CE, Solan MC, Ricketts DM, et al. Cryotherapy compared with Robert Jones bandage after total knee replacement: a prospective randomized trial. Int Orthop 2001;25:250. 26. Healy WL, Seidman J, Pfeifer BA, et al. Cold compressive dressing after total knee arthroplasty. Clin Orthop 1994; 143.
27. Ivey M, Johnston RV, Uchida T. Cryotherapy for postoperative pain relief following knee arthroplasty. J Arthroplasty 1994;9:285. 28. Kullenberg B, Ylipaa S, Soderlund K, et al. Postoperative cryotherapy after total knee arthroplasty—a prospective study of 86 patients. J Arthroplasty 2006;21:1175. 29. Levy AS, Marmar E. The role of cold compression dressings in the postoperative treatment of total knee arthroplasty. Clin Orthop 1993;174. 30. Radkowski CA, Pietrobon R, Vail TP, et al. Cryotherapy temperature differences after total knee arthroplasty: a prospective randomized trial. J Surg Orthop Adv 2007;16:67. 31. Scarcella JB, Cohn BT. The effect of cold therapy on the postoperative course of total hip and knee arthroplasty patients. Am J Orthop 1995;24:847. 32. Smith J, Stevens J, Taylor M, et al. A randomized, controlled trial comparing compression bandaging and cold therapy in postoperative total knee replacement surgery. Orthop Nurs 2002;21:61. 33. Walker RH, Morris BA, Angulo DL, et al. Postoperative use of continuous passive motion, transcutaneous electrical nerve stimulation, and continuous cooling pad following total knee arthroplasty. J Arthroplasty 1991;6:151. 34. Webb JM, Williams D, Ivory JP, Day S, et al. The use of cold compression dressings after total knee replacement: a randomized controlled trial. Orthopedics 1998;21:59. 35. Moher D, Schulz KF, Altman DG. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. Lancet 2001; 357:1191. 36. Weber EWG, Slappendel R, Prins MH, et al. Perioperative blood transfusions and delayed wound healing after hip replacement surgery: effects on duration of hospitalization. Anesth Analg 2005;100:1416. 37. Davies L, Brown TJ, Haynes S, Payne K, Elliott RA, McCollum C. Cost-effectiveness of cell salvage and alternative methods of minimising perioperative allogeneic blood transfusion: a systematic review and economic model. Health Technol Assess. 2006;10:iii-iv, ix-x, 1-210. 38. Hunt TK, Rabkin J, von Smitten K. Effects if edema and anemia on wound healing and infection. Curr Stud Hematol Blood Transf 1986;53:101.