Background: This study provides an updated comparison of the reoperation rates following primary ankle arthrodesis and total ankle replacement on the basis of observational, population-based data from California. We previously reported data from 1995 to 2004, and our current study includes new data from 2005 to 2010. Our hypothesis was that total ankle replacement would demonstrate increasing utilization and lower complication rates given advances in implant design and growth in surgeon experience.
Methods: California’s hospital discharge database was used to identify patients who had undergone primary ankle arthrodesis or total ankle replacement between 1995 and 2010. Short-term outcomes examined were based on Centers for Medicaid & Medicare Services (CMS) surgical quality measures and included readmission for any cause, death, and readmission for 7 common surgical complications (acute myocardial infarction, pneumonia, sepsis, pulmonary embolism, mechanical complications, surgical-site bleeding, and periprosthetic joint infection or wound infection). Long-term outcomes analyzed included rates of major revision surgery (ankle arthrodesis or ankle replacement, adjacent joint procedures (subtalar arthrodesis, triple arthrodesis, tarsometatarsal arthrodesis, and total knee replacement), and below-the-knee amputation. Logistic and proportional hazard regression models were used to estimate the impact of ankle arthrodesis or total ankle replacement on the rates of adverse outcomes, with adjustment for patient factors such as age, sex, race, type of health insurance, and comorbidities. We also compared patients in the 2005 to 2010 cohort with those in the earlier cohort.
Results: In all, 8,491 ankle arthrodesis and 1,280 total ankle replacement cases were identified. Patients managed with ankle replacement were more likely to be female, white, and older and to have Medicare or private health insurance. Short-term complication risk was low for both procedures, and patients managed with total ankle replacement had significantly lower rates of readmission (p < 0.0001) and periprosthetic joint infection/wound infection (p = 0.02) compared with patients managed with ankle arthrodesis.
Conclusions: The inclusion of new data on patients who underwent surgery between 2005 and 2010 demonstrates increasing utilization and lower complication rates for total ankle replacement compared with ankle arthrodesis. These findings suggest that there have been improvements in the clinical safety of total ankle replacement over time.
Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
1University of California Los Angeles, Los Angeles, California
Ankle arthritis results in substantial pain, disability, and diminished quality of life1,2. Conservative treatment options include anti-inflammatory medications, orthotic devices, and operative debridement. Once conservative treatment options fail, ankle arthrodesis has traditionally been the treatment of choice. Arthrodesis has been shown to have good clinical results in terms of pain relief. However, there are ongoing concerns regarding loss of range of motion, potentially increased risk of developing arthritis in the adjacent subtalar and midfoot joints, and risk of nonunion leading to the need for multiple revision surgeries1,3,4. These concerns have led to an ongoing interest in total ankle replacement. Total ankle replacement has the advantages of preserving ankle range of motion and potentially decreasing the risk of developing arthritis of the adjacent subtalar joints. Long-term outcomes comparing the complication and reoperation rates associated with ankle arthrodesis and total ankle replacement as well as the longevity of ankle replacement implants are limited3,5.
We previously used observational, population-based data from California for the years 1995 to 2004 from a statewide inpatient discharge database to compare the complication and reoperation rates following ankle arthrodesis and total ankle replacement2. The purpose of the current study is to provide an updated comparison of the complication and reoperation rates following ankle arthrodesis and total ankle replacement that includes new data on patients for the years 2005 to 2010. Our hypothesis was that total ankle replacement would demonstrate increasing utilization and lower complication rates given advances in implant design and growth in surgeon experience.
Materials and Methods
Data for all hospital discharges in California were obtained for the years 1995 to 2010 from the California Office of Statewide Health Planning and Development (OSHPD) discharge database. This database is compiled annually by OSHPD and includes discharge abstracts from inpatient admissions for all licensed nonfederal hospitals in California 6. Each discharge abstract contains a unique patient identifier, allowing for the tracking of patients over multiple readmissions. It also includes demographic information such as patient age, sex, race and ethnicity, and type of health insurance. Up to 20 inpatient procedures and 24 diagnoses are recorded per hospitalization. All diagnoses and procedures are coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Deaths were identified from the California Death Master Statistical File7.
Inclusion and Exclusion Criteria
The sample consisted of all patients who underwent either primary ankle arthrodesis or ankle replacement during the 16-year observation period (from 1995 to 2010) in the state of California. Ankle arthrodesis procedures were identified using ICD-9-CM procedure code 81.11, and ankle replacement procedures were identified using ICD-9-CM procedure code 81.56 (see Appendix). Patients with a non-California ZIP code were excluded. The unit of analysis was hospital discharge for each patient.
Outcomes Studied (Dependent Variables)
The short-term outcomes chosen as dependent variables included death, readmission for any cause following the index procedure, and readmission for surgical complications based on quality measures developed by the Centers for Medicare & Medicaid Services (CMS) for total hip and knee replacement8. These 7 common individual surgical complications were acute myocardial infarction, pneumonia, sepsis, pulmonary embolism, mechanical complications, surgical-site bleeding, and periprosthetic joint infection/wound infection (see Appendix). These complications were identified using ICD-9-CM codes consistent with the occurrence of these outcomes during either the index hospitalization or the readmission8.
Long-term outcomes evaluated were the overall rate of major ankle revision surgery (arthrodesis or replacement), adjacent joint procedures (including subsequent subtalar, triple, and tarsometatarsal arthrodesis), total knee replacement, and below-the-knee amputation. The coding algorithm requires the assumption that the revision code is for the same extremity since laterality is not recorded in the OSHPD database8.
Predictors (Independent Variables)
The primary procedure (ankle arthrodesis or total ankle replacement) was included as a predictive variable. Patient demographic characteristics such as age, sex, race, and type of health insurance were included in the regression models as covariates to adjust for their effect. Patient comorbidities were included in the regression models using the clinical condition categories, which assess comorbid conditions and have been validated for use with the CMS quality measures in studies using administrative databases8.
Hospital volume was also included as an independent variable, and it was defined for each hospital by the mean annual number of ankle arthrodesis or total ankle replacement procedures performed.
The current cohort of patients, who underwent surgery between 2005 and 2010, was compared with the earlier cohort, who underwent surgery between 1995 and 2004.
The patient sample was described with use of bivariate descriptive statistics. Results reported using descriptive statistics include patient demographics and the annual rates of ankle arthrodesis and total ankle replacement. Differences were assessed with use of the t test for continuous variables and the Pearson chi-square test for categorical variables. Raw rates of complications are also reported separately for the ankle arthrodesis and total ankle replacement groups. Multivariate logistic regression was used to identify the independent effect of surgery type on short-term outcomes (death; readmission for any cause; and readmission for acute myocardial infarction, pneumonia, sepsis, pulmonary embolism, mechanical complications, surgical-site bleeding, and periprosthetic joint infection/wound infection) while adjusting for hospital volume and patient factors such as age, sex, race, type of health insurance, and comorbidities. The association between the risk of short-term complications and the selection of primary ankle arthrodesis or ankle replacement is expressed as the odds ratio (OR), along with the 95% confidence interval (CI) and p value, for a complication occurring in the group of patients managed with total ankle replacement with respect to the reference group of patients managed with ankle arthrodesis.
Time-to-event analysis with Cox proportional hazard regression models was used to evaluate long-term complications. Separate analyses of the independent impact of the type of primary procedure (ankle arthrodesis or total ankle replacement) on long-term outcomes (revision surgery [including adjacent joint arthrodesis], below-the-knee amputation, and total knee replacement) were performed by including the same covariates as were used for the logistic regression models. The strength of the association between the risk of long-term complications and the selection of primary ankle arthrodesis or arthroplasty is expressed as the adjusted hazard ratio (HR) along with the 95% CI and p value for complications occurring in the total ankle replacement group with respect to the reference group of patients managed with ankle arthrodesis.
A total of 8,491 patients (86.90%) underwent ankle arthrodesis and 1,280 (13.10%) underwent total ankle replacement during the 16-year observation period. This compares with a total of 4,705 ankle arthrodesis (90.74%) and 480 ankle replacement (9.26%) patients in the prior study (1995 to 2004). Table I lists the annual volume of cases in each group, and Figure 1 shows the trend in this annual volume for each procedure. There is a peak in the annual volume of ankle arthrodesis procedures in 2006, followed by a decline in the volume of these cases. This decrease is complemented by an increase in the annual volume of total ankle replacement cases.
Table II lists the demographics of the patient sample. Patients who underwent total ankle replacement were significantly older (a mean age of 68.45 years compared with 65.60 years for those who underwent ankle arthrodesis; p < 0.0001), and they were more likely to be female (51.41% compared with 47.05%; p = 0.01). Patients who underwent total ankle replacement were more likely to be white (86.95% in comparison with 85.10% of those in the ankle arthrodesis group; p < 0.0001). Patients who underwent total ankle replacement were also more likely to have Medicare (49.71%) or private health insurance (44.92%) in comparison with those who underwent ankle arthrodesis (39.11% of whom had Medicare and 44.78% of whom had private health insurance; p < 0.0001). Total ankle replacement procedures were also performed at higher-volume hospitals (with a mean of 179.50 cases per year) compared with ankle arthrodesis procedures (which were performed at hospitals with a mean volume of 84.62 cases per year; p < 0.0001).
Patients managed with total ankle replacement had a decreased rate of readmission (OR, 0.67 [95% CI, 0.661 to 0.671]; p < 0.0001) and periprosthetic joint infection/wound infection (OR, 0.28 [95% CI, 0.10 to 0.80]; p = 0.02) in comparison with patients managed with ankle arthrodesis. There was no significant difference in the rate of acute myocardial infarction, pneumonia, sepsis, pulmonary embolism, mechanical complications, surgical-site bleeding, or death. Table III gives the ORs as well as 95% CIs and p values for the short-term complications occurring in patients managed with total ankle replacement in comparison with those managed with ankle arthrodesis.
Patients managed with total ankle replacement were at a lower risk of requiring subsequent subtalar arthrodesis (HR, 0.45 [95% CI, 0.24 to 0.83]; p = 0.01), requiring ankle arthrodesis as a salvage procedure (HR, 0.22 [95% CI, 0.13 to 0.37]; p < 0.0001), and requiring a below-the-knee amputation (HR, 0.12 [95% CI, 0.03 to 0.49]; p = 0.003). The ankle replacement group was at higher risk of undergoing revision ankle replacement procedures (HR, 11.28 [95% CI, 6.31 to 20.17]; p < 0.0001). There was no significant difference between the 2 groups in terms of the risk of subsequent triple arthrodesis, tarsometatarsal arthrodesis, or total knee replacement. The HRs along with 95% CIs and p values for long-term complications are included in Table IV.
This study presents an updated comparison of the complication rates of total ankle replacement and ankle arthrodesis using California hospital discharge database records. The results from our prior study, which included data from 1995 to 2004 from this database, showed that total ankle replacement was associated with a higher overall short-term complication rate in comparison with ankle arthrodesis. Total ankle replacement was specifically associated with an increased risk of device-related infection as well as major revision surgery following the index procedure; however, it was associated with a decreased risk of subsequent subtalar arthrodesis2. This updated analysis demonstrated that short-term complication rates associated with total ankle replacement relative to ankle arthrodesis were equivalent or lower. These findings suggest that the risks associated with total ankle replacement have decreased over time while the risk of subsequent subtalar arthrodesis following total ankle replacement has remained lower than after ankle arthrodesis over the longer term.
The analyses of this study, which includes additional discharge data from 2005 to 2010, show a lower rate of readmission and periprosthetic joint infection/wound infection in the total ankle replacement group. The findings further show that patients who underwent ankle arthrodesis were more likely than those who underwent total ankle replacement to need subsequent subtalar arthrodesis. These findings suggest that in addition to preserving ankle range of motion, and perhaps as a result of this preservation of motion, total ankle replacement may have the advantage of preventing adjacent joint arthritis. However, it is important to keep in mind that this difference may in part be related to selection bias for the index procedure. For example, total ankle replacement may be less likely to be chosen for patients with complex posttraumatic ankle deformities, who are more likely to develop subtalar arthritis in the future regardless of the type of ankle procedure performed. In this study, including the updated data, there was no significant difference in the rate of major revision surgery between the 2 groups. This finding may reflect the increasing longevity of ankle replacement implants. The patient demographics for each treatment group have not changed over time. Patients undergoing total ankle replacement continue to be older, are more commonly female and white, and are more likely to have Medicare or private health insurance than Medicaid or Workers’ Compensation as the payer.
The main weaknesses of this study relate to the types of information available in administrative databases such as the one used in this study. There is no information on preoperative severity of arthritis or other factors that may complicate surgery, such as bone loss or instability. Additionally, there is no information available to describe postoperative pain relief and functional outcomes. Complications may be underrepresented in this study as only complications that required readmission were identified. Moreover, the decision to recommend ankle arthrodesis may be made to patients with higher degrees of medical comorbidity and therefore those with a greater risk of complications.
This database includes only inpatient admissions; thus, any ankle arthrodesis or total ankle replacement surgeries performed on an outpatient basis are not included in these analyses. In addition, there was a slight discrepancy in the annual volume of each procedure between the 2 studies, with a slightly higher volume of annual cases in each group in the earlier cohort (1995 to 2004) included in this study. This discrepancy reflects refinement of the coding methods and exclusion criteria that were used in the current study.
One of the key weaknesses of this observational study design is selection bias between the groups. A propensity score model was created by calculating a propensity probability for patients undergoing total ankle replacement. The statistical analyses were repeated excluding patients in the top and bottom 1% of propensity probability for total ankle replacement. Significant differences in the demographics of the patient sample were unchanged compared with the model that included all patients. Comparisons of the outcomes were also similar, with both models demonstrating a significant difference in the rate of postoperative periprosthetic joint infection/wound infection between the 2 groups. Overall readmission rates also had a consistent trend in both analyses, with a lower rate for the total ankle replacement group; however, this failed to reach the threshold for significance, with p = 0.059, in the propensity model.
Another weakness of this study design is the difference in length of the observational period for patients depending on their procedure date. The mean follow-up was 6.6 years, with a standard deviation of 4.6 years. The bottom quartile for the observation period was 2.7 years or less. Ongoing observation would be helpful to identify differences in the reoperation rates over longer postoperative periods.
One additional weakness of the study is the reliance on administratively coded data. For example, the diagnoses of degenerative joint disease, rheumatoid arthritis, and osteonecrosis exceed a total of 100%, as some patients had codes for multiple diagnoses. The sensitivity and specificity of this coding have not been established, and caution should be used when interpreting the predictive effect of these diagnoses in the current study.
Despite these weaknesses, this study provides useful population-based information on complications following ankle arthrodesis and total ankle replacement. These new data suggest that the risks of short-term complications after total ankle replacement have lessened over time and that it is a safe alternative to ankle arthrodesis for the treatment of ankle arthritis.
The specific procedure codes used to identify the cohort sample are available with the online version of this article as a data supplement at jbjs.org.
Investigation performed at the Department of Orthopaedic Surgery, University of California Los Angeles, Los Angeles, California
Disclosure: The authors indicated that there was no external source of funding for this study. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
1. Coester LM, Saltzman CL, Leupold J, Pontarelli W. Long-term results following ankle arthrodesis for post-traumatic arthritis. J Bone Joint Surg Am. 2001 ;83(2):219–28.
2. SooHoo NF, Zingmond DS, Ko CY. Comparison of reoperation rates following ankle arthrodesis and total ankle arthroplasty. J Bone Joint Surg Am. 2007 ;89(10):2143–9.
3. SooHoo NF, Kominski G. Cost-effectiveness analysis of total ankle arthroplasty. J Bone Joint Surg Am. 2004 ;86(11):2446–55.
4. Jiang JJ, Schipper ON, Whyte N, Koh JL, Toolan BC. Comparison of perioperative complications and hospitalization outcomes after ankle arthrodesis versus total ankle arthroplasty from 2002 to 2011. Foot Ankle Int. 2015 ;36(4):360–8. Epub 2014 Oct 30.
5. Saltzman C. Editorial: why ankle replacement? Clin Orthop Relat Res. 2004 ;424:2.
6. Office of Statewide Health Planning and Development. 1996 discharge data file format documentation—confidential layout. Sacramento, CA; 1997.
Copyright 2016 by The Journal of Bone and Joint Surgery, Incorporated
7. Zingmond DS, Ye Z, Ettner SL, Liu H. Linking hospital discharge and death records—accuracy and sources of bias. J Clin Epidemiol. 2004 ;57(1):21–9.