SEARCH

SEARCH BY CITATION

Keywords:

  • systematic review;
  • orthopedic implant;
  • implant removal;
  • revision;
  • total hip arthroplasty;
  • arthroplasty

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Purpose

To identify studies that have validated administrative and claims database algorithms for identifying patients with orthopedic device revision or removal.

Methods

As a part of the Food and Drug Administration's Mini-Sentinel pilot program, we performed a systematic review to identify algorithms for orthopedic implant removal/revision in administrative and claims databases in the USA or Canada.

Results

Five studies examined the validity of database algorithms against a gold standard of documentation in medical records (n = 3) or codes/documentation in another database (n = 2). The positive predictive values (PPV) of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and/or the Current Procedural Terminology codes for revision total hip arthroplasty (THA) in the US Medicare population compared with medical record review were 92%and 91%, respectively. In another study of the US Medicare population, multiple ICD-9 codes for revision total knee arthroplasty were compared with newly available single ICD-9-CM codes for revision knee arthroplasty; sensitivity was 87% and specificity was 99% (PPV not provided). The fourth study validated the ICD-9-CM codes for revision total knee arthroplasty against Ontario health insurance physician fee service claims as the gold standard and found a PPV of 32%. In the last study in Medicare population, the accuracy of the attribution of revision THA to the same side as the earlier index primary THA was examined; PPV for same laterality of revision THA was 71% (using ICD-9-CM codes).

Conclusions

Validation data, with regard to the ICD-9-CM or the Current Procedural Terminology code algorithms for revision THA in the Medicare population, exist. More validation studies are needed to confirm these findings and examine other large databases. Copyright © 2012 John Wiley & Sons, Ltd.


INTRODUCTION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Joint arthroplasty is one of the most common elective surgeries performed in the elderly. In patients with severe arthritis, it leads to improvement of symptoms including pain and functional impairment.[1, 2] The demand for hip revision arthroplasties is projected to double by 2026.[3] Even greater is the rise of knee revision arthroplasties, which are expected to double by 2015.[3]

The Food and Drug Administration's (FDA) Mini-Sentinel pilot program aims to inform the FDA about the scientific operations needed to conduct active safety surveillance on the medical products it regulates, using existing automated healthcare data. To accomplish this goal, one of the initial steps was to document the established validity of algorithms previously used in administrative and claims data sources to identify various health outcomes of interest. The primary objective of this project was to review studies that have validated algorithms using administrative and claims data from the USA or Canada to identify orthopedic implant removal or implant revision, which were among the 20 health outcomes selected for these reviews. If fewer than five validation studies were identified, a secondary objective was to identify nonvalidated algorithms that have been used to identify the health outcomes of interest using administrative and claims data. This summarized report describes the review process and main findings for the administrative and claims database algorithms for implant removal and revision. The complete report that has been submitted to the FDA is available at http://mini-sentinel.org/foundational_activities/related_projects/default.aspx.

METHODS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Search strategy

The general search strategy was developed on the basis of prior work by Observational Medical Outcomes Partnership and modified slightly for these reports by the Mini-Sentinel investigators, to result in the identification of more citations.[4, 5] A more thorough description of the methods and their rationale can be found in the accompanying manuscript by Carnahan and Moores.[5] The base search strategy was then combined with PubMed terms, including Medical subject heading terms and text words. This included terms such as “arthroplasty,” “orthopedic procedures,” “reoperation,” “device removal,” and “second-look surgery” in clinical studies and registries in humans (details provided in the full report at http://mini-sentinel.org/foundational_activities/related_projects/default.aspx). The workgroup also searched the database of the Iowa Drug Information Service (IDIS) using a similar search strategy to identify other relevant articles that were not found in the PubMed search. The search results were restricted to articles published on or after 1 January 1990. Duplicates were eliminated using a citation manager program. The PubMed search was conducted on 14 May 2010 and updated on 20 July 2010; the IDIS searches on 11 June 2010.

Abstract review

Each abstract was reviewed independently by two investigators (JAK and JAS) to determine whether the full-text article should be reviewed. Exclusion criteria were as follows: (i) not related to implant removal or implant revision, (ii) not an administrative and claims database (insurance claims databases as well as other secondary databases), or (iii) data source not from the USA or Canada. The interrater agreement for abstract inclusion was calculated using a Cohen's kappa statistic. Before starting the full title and abstract review, a trial consensus exercise was undertaken for a random subset of 35 articles. Both reviewers (JAS and JAK) independently selected the same four articles for inclusion with an agreement of 100% and kappa of 1. Mini-Sentinel collaborators were also asked to provide any published or unpublished studies of administrative and claims data algorithms of implant revision or removal.

Full-text review and data abstraction

Once the list of full-text articles had been compiled, each reviewer independently reviewed the full-text and independently abstracted the data (JAS and JAK) by applying the same exclusion criteria as for abstract review. In addition, the following two exclusion criteria were applied: (i) poorly described health outcome of interest identification algorithm that would be difficult to operationalize and (ii) no validation of outcome definition or reporting of validity statistics. Abstractions were compared and any disagreements on data were resolved by discussion. For each algorithm, we abstracted the validity statistics provided in the included studies, including sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).

RESULTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Search strategy and results

The search strategies for the PubMed search were performed on 14 May 2010, updated on 20 July 2010, and the IDIS search was performed on 21 June 2010. The initial PubMed search identified 580 citations; an additional 24 articles were identified from the updated PubMed search, the IDIS search, the reference list of included studies, and e-mails to the authors of included studies and by the Mini-Sentinel collaborators. Of these, we reviewed full text for 55 articles after excluding three duplicate articles. Of these, only five studies had some data on validation, which are the focus of this report.[6-10] Details of the other 31 articles with nonvalidated algorithms are provided in the final evidence report on the Mini-Sentinel Web site (http://mini-sentinel.org/foundational_activities/related_projects/default.aspx).[2, 11-40] The 19 articles excluded were those not related to health outcome of interest (n = 11)[41-51], those that did not use administrative or claims data (n = 7),[52-58] and an editorial article (n = 1).[59] Cohen's kappa for agreement between the two reviewers on inclusion versus exclusion of abstracts for full-text retrieval of articles was 0.95 (SE = 0.028; reviewer 1 selected 31 of 580 articles; reviewer 2 selected 34 of 580 articles, which included all 31 from reviewer 1).

SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Of the five studies that had validated algorithms, one study by Coyte et al.[6] used Canadian administrative and claims data whereas four studies used American administrative and claims data. Of these four studies, three studies (Heck et al.,[7] Katz et al.,[8] and Mahomed et al.,[9]) used the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and/or the Current Procedural Terminology (CPT) codes in Medicare databases (Table 1). The fourth study (Katz et al.,[10]) used ICD-9-CM codes for revision total hip arthroplasty (THA) and ICD-9-CM and CPT codes for primary THA. This study[10] differed from the other four studies because its purpose was to assess whether a revision THA following a primary THA in Medicare recipients can be attributed to the same side, with medical records serving as the gold standard. In essence, this study did not examine the validity of the ICD/CPT codes for revision itself but rather examined the accuracy of attributing the revision to the same side as the primary surgery based on presence of a code. The algorithms across five studies included the use of ICD-9-CM codes; in some instances, CPT-4 codes were used. Patient characteristics and exclusions are listed in Table 1.

Table 1. Algorithm validation studies
Citation/country (US vs Canada)Study population and time periodDescription of outcome studiedAlgorithmValidation/adjudication procedure, operational definition, and validation statistics
  1. NR, not reported.

  2. a

    Additional CPT-4 and ICD-9-CM codes included the following: removal of internal fixation device, mechanical and other complications/infection due to internal prosthetic device, implant, or graft: V54.0, 996.4, 996.6, 996.60, 996.66, 996.67, 996.7, 996.70, 996.77, 996.78; removal of prosthesis/internal fixation device: 78.6, 78.60, 78.65, 80.0, 80.00, 80.05; revision of hip replacement (partial, total): 81.53; CPT-4 procedure codes for removal of implant: 20680, 27090, 27091.

Coyte et al.[6]/Canada18 530 TKA patients in Ontario, Canada, from 1 April 1984 to 31 March 1991 (84 months)Revision rate (1301/18 530 underwent revision)Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures code, 93.41 and one of the following ICD-9-CM codes (996.4, 996.6, or 996.7)Ontario Health Insurance physician fee service claims were used as gold standard for validation
 Excluded: hospitalizations were excluded if the patient was not a resident of Ontario, if pertinent data (such as date of birth or place of residence) were missing, or if a knee-replacement procedure either was not performed or was miscoded, as evidenced by a procedure performed in a non–acute care facility or by discharge to home with self-care within 3 days after the procedure (in Ontario during this period, patients were never discharged in this period and procedures were not performed in non–acute care facilities); 1144 patients were excluded (5.8% of population)  Sensitivity of 77.7%, specificity of 97.6%, PPV of 66.9%, and NPV of 98.6% were obtained
 Mean age, 68.9 years old, 63.3% female, 85.2% had osteoarthritis, 80% had a Charlson index of 0   
Heck et al.[7]/USAMedicare Provider Analysis and Review Part A patients who underwent TKA from 1985 to 1990 (60 months). For validation, the 15-month period (1 October 1989 to 31 December 1990) was usedRevision rate (4.2% at 4 years)One or more of the 996.xx complication codes and a previous ICD-9-CM code of 81.41Presence of a specific revision code (81.55) during the 15-month period (1 October 1989 to 31 December 1990) was the gold standard for validation
 Excluded: patients younger than 65 years, infection of the hip, metastatic or bone cancer, conversion of hemiarthroplasty (or other hip surgery) to total hip replacement, fracture of the hip or femur, HMO, not enrolled in both parts of Medicare, and nonresidents of the USA  Sensitivity of algorithm was 87.2% and specificity was 99.0%
 Age was NR, 68.3% female, 88.5% had osteoarthritis   
Katz et al.[8]/USA71 477 THA patients from a Medicare database from July 1995 to June 1996 (12 months)Complications leading to revision (12 956/71 477 underwent revision)Revision THA: presence of single claim including one of the three CPT-4 codes for revision (27134, 27137, or 27138) if the patient was in the hospital on that date; in other cases, the algorithm required the presence of one of the three CPT-4 codes (27134, 27137, or 27138) and an additional CPT-4 or ICD-9-CM codeaA medical record review was performed by trained nurse abstractors. A random sample of 1031 (1.8%) of the primary procedures and 671 (5.2%) revision procedures was used. The PPV was 99% for primary THA and 92% for revision THA
 Excluded: younger than 65 years, infection of the hip, metastatic or bone cancer, conversion of hemiarthroplasty (or other hip surgery) to total hip replacement, fracture of the hip or femur, HMO, not enrolled in both parts of Medicare, and nonresidents of the USA; 18 106 primary patients (24% of the population) were excluded and 961 revision patients were excluded (7%) Primary THA: presence of single claim with an ICD-9 procedure code of 81.51 if the patient was in the hospital on that date; in other cases, the presence of 81.51 and an additional surgical claim with a CPT code of 27130 
 Age, 74.7 ± 6.09 years, 64% female, 94% Caucasian, 94% had osteoarthritis   
Mahomed et al.[9]/USA75 501 THA patients from a Medicare Part A or Part B database from July 1995 to June 1996 (12 months)Complications leading to revision (13 483/75 051 underwent revision)CPT codes for revision (27134, 27137, or 27138) were required to be labeled a revision case; ICD-9-CM code 81.51 or CPT code 27130 to identify primary THA caseA medical record review was performed by trained nurse abstractors; a random sample of 900 primary and 550 revision THA was used; the PPV was 99% for primary THA and 91% for revision THA
 Excluded: younger than 65 years, infection of the hip, metastatic or bone cancer, conversion of hemiarthroplasty (or other hip surgery) to total hip replacement, fracture of the hip or femur, HMO, not enrolled in both parts of Medicare, and nonresidents of the USA; 19.6% of primary and 3.1% of revision patients were excluded; 94% had osteoarthritis   
Katz et al.[10]/USA58 521 THA patients from a Medicare Part A database from July 1995 to June 1996 (12 months)Revision rates (4460/58 521 underwent revision)ICD-9-CM codes for revision (81.53 up to October 2005 and 00.70–00.73 after October); attribution of revision to same side as index primary THA was tested using ICD/CPT codesA medical record review was performed; a sample of 374 THA was selected; the PPV was 71% (95% CI: 66, 76) for revision THA on the same side as the index THA
 Excluded: younger than 65 years, infection of the hip, metastatic or bone cancer, conversion of hemiarthroplasty (or other hip surgery) to total hip replacement, fracture of the hip or femur   
 No demographic characteristics provided   

Coyte et al.[6] tested an algorithm requiring the simultaneous use of Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures code 93.41(total knee replacement, primary or secondary) in any procedure field in the record and any one of the following ICD-9-CM codes in the Canadian Institute for Health Information Abstract Master File (a hospital discharge database) to identify revision arthroplasty after total knee arthroplasty (TKA) in 18 530 patients in Ontario, Canada, from 1984 to 1991. The ICD-9-CM codes were 996.4 (mechanical complication of internal orthopedic device, implant and graft), 996.6 (infection and inflammatory reaction due to internal prosthetic device, implant and graft), or 996.7 (other complications of internal prosthetic device, implant and graft). Procedures with both codes were labeled as revision; all other surgeries were classified as primary. The gold standard was the Ontario Health Insurance physician fee service claims database, which distinguishes between primary and revision cases.[6] Specific fee for service claims that identified revision knee replacements included R244A (revision TKA) or the simultaneous occurrence of E564A (revision of arthroplasty) and any one of the following fee service codes: R248A (total knee replacement with take-down fusion), R441A (total replacement—both compartments), R482A (hemiarthroplasty—single component), or R483A hemiarthroplasty—double component). The revision rate was 7.0% (1301 revision patients/18 530 TKA patients).[6] The algorithm had a sensitivity of 77.7% and a specificity of 97.6%.[6] Because raw numbers were provided, we calculated positive and negative predictive values, which were 66.9% (95%CI = 64.1%, 69.6%) and 98.6% (95%CI = 98.4%, 98.8%), respectively.

Heck et al.[7] validated one set of codes against another code as the gold standard. They used the Health Care Financing Administration Medicare Provider Analysis and Review Part A files to study knee arthroplasties from 1985 to 1990. Starting 1 October 1989, separate ICD-9-CM codes were used for primary surgery (81.54) and revision (81.55), whereas before 1989, all TKAs (primary and revision) were coded as 81.41. Thus, an algorithm using the presence of one or more of 996.xx complication diagnostic codes in combination with previous ICD-9-CM code 81.41 (for TKA) to identify revision TKA was tested during a 15-month period from 1 October 1989 to 31 December 1990. The gold standard for validation of the algorithm was the presence of ICD-9-CM code 81.55 for revision TKA. The rates of revision were 2.2% at 2 years, 3.2% at 3 years, and 4.3% at 4 years. The algorithm showed a sensitivity of 87.2% and a specificity of 99%.[7] PPV and NPV were not provided.

Katz et al.[8],[10] and Mahomed et al.[9] used similar algorithms including ICD-9-CM and/or CPT-4 codes to study complications after THA in Medicare patients, with medical record review serving as the gold standard in each study.

Katz et al.[8] used a Medicare database to identify 71 477 THA patients from July 1995 to June 1996. The algorithm defined a case as revision THA if they met one of the two criteria (Table 1): (i) presence of one of the three CPT-4 codes (27134, 27137, or 27138) for revision THA and an additional ICD-9-CM or CPT code for revision THA from the following list (removal of internal fixation device, mechanical and other complications/infection due to internal prosthetic device, implant, or graft: V54.0, 996.4, 996.6, 996.60, 996.66, 996.67, 996.7, 996.70, 996.77, 996.78; removal of prosthesis/internal fixation device: 78.6, 78.60, 78.65, 80.0, 80.00, 80.05; revision of hip replacement—partial and total): 81.53, CPT procedure codes for removal of implant: 20680, 27090, and 27091); or (ii) a single claim for one of the three CPT-4 codes (27134, 27137, or 27138) only if the patient was in the hospital on that date. Medical record was the gold standard. The algorithm yielded a revision rate of 18.1% (12 956 revision patients/71 477 THA patients). Using a random sample of 1031 (1.8%) primary THAs and 671 (5.2%) revision THAs, they found that the PPV of the algorithm was 99% for primary THA and 92% for revision THA.[8] NPVs were not provided.

Using a similar algorithm, Mahomed et al.[9] used Medicare Part A (claims submitted by hospitals) and Part B files (claims submitted by physicians and outpatient facilities) to identify 75 501 patients who underwent THA surgery from July 1995 to June 1996.Revision THA was identified with CPT codes 27134, 27137, or 27138 and primary THA by an ICD-9-CM code 81.51 or CPT code 27130. They found that revision THA constituted 18.0% of this sample (12 483 revision patients/75 051 THA patients). A random sample of 900 primary and 550 revision THAs was chosen for validation with medical records as the gold standard. The PPV was 99% for primary THA and 91% for revision THA.[9] NPVs were not provided.

Katz et al.[10] used Medicare part A (hospital) claims database to identify a national cohort of 58 521 THA patients from July 1995 to June 1996 and then identified patients who underwent revision THA between 1995 and 2006 and resided in one of the seven US states. The purpose was to assess the accuracy of the automatic attribution of revision THA to the same side as the previous primary THA and not the accuracy of the code for revision itself. They used ICD-9-CM code 81.51 and/or a CPT code of 27130 to identify primary THAs in Medicare part A claims. They used ICD-9-CM code 81.53 before 1 October 2005 and 00.70 to 00.73 after 1 October 2005 and both codes for 2005–2006 to identify revision THA in Medicare part A claims. The algorithm yielded a revision rate of 7.62% including all states (4460 revision patients/58 521 THA patients). The gold standard was documentation that revision THA had been performed on the side as the index primary THA in the medical records. The validation was performed in a subsample of 374 (29%) of the 1309 primary procedures restricted to seven states. The PPV of the algorithm was 71% for attributing revision THA to the same side as the index THA.[8]

Comparison of patient characteristics between the five included studies

Differences in results across the included studies may, in part, be related to differences in patient/procedure exclusions in each study, validation cohort sample sizes, duration of the validation period, and patient demographics.

There were several differences between the patients included in the five validated studies. The exclusion criteria differed between Coyte et al.[6], Heck et al.[7], and the other three studies from the Katz's group (Table 1). Coyte et al.[6] excluded 1144 patients (5.8% of the population). Heck et al.[7] did not describe how many patients were excluded. Katz et al.[8] excluded 18 106 primary surgery patients (24% of the population) and 961 revision patients (7% of the population). Mahomed et al.[9] excluded 19.6% and 3.1% of primary and revision cohorts, respectively. In 2010, the study by Katz et al.[10] did not report patient exclusions.

The number of patients included in the validation algorithms differed between studies. Katz et al.[8],[10] and Mahomed et al.[9] validated data for a small sample (<5% of the total patient population) using medical records as the gold standard, whereas Coyte et al.[6] and Heck et al.[7] did not perform any chart validation and compared one set of codes to another code (labeled gold standard) in the entire sample. The revision rates somewhat varied—7% in the study by Coyte et al.[6], 4% at 4 years in the study by Heck et al.[7], 18% in the study by Katz et al.[8], 18% in the study by Mahomed et al.[9], and 7.6% in the study by Katz et al.[10] This may partially explain the low PPV in the Coyte et al.[6] article as compared with Katz et al.[8],[10] and Mahomed et al.[9] because PPV depends on prevalence.

Age was not reported in three studies[7, 9, 10], and gender was not reported in two studies.[9, 10] Coyte et al.[6] reported a mean age of 68.9 years with 63.3% female in their cohort. Heck et al.[7] reported that 68.3% of the cohort was female; age was not reported. Katz et al.[8] reported that their cohort was 74.7 years old (SD = 6.1) and 64% female.

In addition to differences in validation methods and statistics, the studies varied in the length of the validation period. Coyte et al.[6] examined cohorts more than 84 months, whereas Katz et al.[8],[10] and Mahomed et al.[9] each examined it more than 12 months. In contrast, the study by Heck et al.[7] had a validation period of 15 months.

Coyte et al.[6] mentioned that diagnosis was searched in any field, whereas other studies did not explicitly mention this detail.

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

This report summarized the current body of evidence regarding validated algorithms for the revision of orthopedic implants (namely, revision arthroplasty surgery) from administrative and claims databases. Among the validation studies, two studies (Katz et al.,[8] and Mahomed et al.,[9]) were the most relevant studies because they used medical records as the gold standard on the same Medicare cohort. Both used a rigorous methodology for validation of algorithms for revision arthroplasty using random samples of patients with primary and revision THA. However, the algorithms in the two studies differed somewhat. Katz et al.[8] used the presence of one of the CPT codes for revision (27134, 27137, or 27138) if the patient was hospitalized on the same day or one of these three codes and an additional ICD-9-CM or CPT code for removal of components is present (V54.0, 996.4, 996.6, 996.60, 996.66, 996.67, 996.7, 996.70, 996.77, 996.78, 78.6, 78.60, 78.65, 80.0, 80.00, 80.05, 81.53, 20680, 27090, 27091) to identify revision THA. The second algorithm by Mahomed et al.[9] used the presence of one of the CPT codes for revision (27134, 27137, or 27138) to identify revision THA. Katz et al.[8] and Mahomed et al.[9] provided validation data with regard to algorithms using ICD-9-CM/CPT codes in the US Medicare population. The PPV for revision THA in the Katz et al.[8] study was 92%[8] and in the Mahomed et al.[9] study was 91%.[9] The NPVs were not provided in either study. Both PPV and NPV are important aspects of algorithm validation that can assist readers and researchers in assessing the accuracy of an algorithm. Although we included the Katz et al.[10] study, it differed from other studies in that it did not examine the validity of ICD/CPT codes for revision arthroplasty but rather whether revision following primary THA could be attributed to the same side as the index primary THA.

In this systematic review, we found five studies that provided five unique algorithms for the retrieval of implant and revision from US and Canadian administrative and claims databases. Four studies used Medicare databases (except Coyte et al.[6]), four focused on revision THA (except Heck et al.[7], who focused on revision TKA), and four tested algorithms for revision surgery (except Katz et al.[10], where the algorithm was developed for laterality of revision THA). The study by Heck et al.[7] was not a true validation study because it considered a specific code as the gold standard, which arguably is not an acceptable gold standard. In addition, some codes used in this study are now outdated. Katz et al.[10] used medical record documentation to examine the accuracy for determining the laterality (right versus left) of revision THA; the algorithm was not for identification of revision arthroplasty. Coyte et al.[6] used physician fee service claims database as the gold standard. Katz et al.[8] and Mahomed et al.[9] used medical record documentation as the gold standard, which is an appropriate validation method.

To some extent, findings from Coyte et al.[6] may not be directly comparable with those of the Katz et al.[8] and Mahomed et al.[9] studies because the gold standards were different (physician fee claim database versus medical record documentation). In addition, Coyte et al.[6] excluded 6% of the miscoded revisions by using the algorithm that excluded patients with less than a 3-day stay and discharge to home or with surgeries at non–acute care settings, whereas Katz et al.[8] and Mahomed et al.[9] used different exclusion criteria (hip fracture, pathological fracture, acetabuloplasty, and infection), which likely affected the performance characteristics of these algorithms. The time frame for studies was also different: 1984–1990 for Coyte et al.[6], 1989–1990 for Heck et al.[7], 1995–1996 for Katz et al.[8], 1995–1996 for Mahomed et al.[9] and 1995–2006 for Katz et al.[10] Coyte et al.[6] used the Canadian Institute for Health Information Abstract Master File, whereas the other studies used Medicare data. Thus, several differences between Coyte et al.[6] versus Katz et al.[8]/Mahomed et al.[9] studies can explain the differences in validation statistics, which were better and consistent for the latter studies and slightly lower for Coyte et al.[6]. Another observation was that Katz et al.[8] and Mahomed et al.[9] used the same cohort, and the studies were performed by the same group of investigators, which might explain the consistency of results (validation statistics) between these two studies.

Revision arthroplasty represents a major event and one that accurately reflects major effects on healthcare resource utilization and health-related quality of life in patients with joint replacement. The challenge is the paucity of validated algorithms and the variability in the approaches used. Revisions arthroplasty is largely a consequence of implant failure (loosening, osteolysis, wear, breakage, and dislocation), infection at the bone implant interface, and/or recurrence of major clinical symptoms, pain, and functional decline.[13, 22, 60]

The studies included in this review used the occurrence of diagnostic and/or procedure codes in administrative and claims databases in their algorithms to allow identification of revision cases. In Heck et al.[7] using Medicare data, the revision rate was 4.2% at 4 years. In longitudinal follow-up, the algorithm of Coyte et al.[6] identified 7% hip revision surgeries (Canadian Registry), similar to 7.6% identified by Katz et al.[10]. The two other US studies (Katz et al.[8] and Mahomed et al.[9]) examined a sample of primary and revision THA, where revision THA constituted one fifth of the entire cohort; however, the revision THA were oversampled for the validation study in each case. Most algorithms were tested in Medicare populations (except Coyte et al.[6]), and approximately one third of all arthroplasties are performed in those younger than 65 years. Therefore, validation studies that include younger populations are needed. Future validation studies should include younger populations to avoid systematic bias in algorithms to identify these cohorts.

CONCLUSIONS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

Although the PPVs of the two algorithms in Medicare data were high, additional data are needed before we feel confident in recommending the use of these algorithms to the FDA to query databases for implant removal or revision. These two algorithms with high validation statistics were tested in one database (Medicare) for revision THA by one team of investigators. These findings need to be replicated in another database by other groups to increase confidence in its generalizability. Furthermore, algorithms need to be developed for other arthroplasties (knee and shoulder) and tested in populations with differing revision rates. Future research endeavors need to provide validation of these existing algorithms in Medicare databases, to develop new approaches/algorithms to improve prediction even further, and to test these algorithms in multiple large databases and populations.

ETHICAL APPROVAL

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

This study did not involve any human subjects or animals. Because it was a systematic review of published data, no ethical committee approval was required.

CONFLICT OF INTEREST

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

There are no financial conflicts related to this work. J.A.S. has received speaker honoraria from Abbott; research and travel grants from Allergan, Takeda, Savient, Wyeth, and Amgen; and consultant fees from Savient, URL pharmaceuticals, and Novartis. M.B. has received research funding from Stryker, Smith, and Nephew, Zimmer, Pfizer, Amgen, and DePuy. J.A.K. has no financial relationships.

This study did not need any approval from Human Studies Committee because it was a review of published data in the public domain.

KEY POINTS

  • An algorithm for claims data-based definition for revision total hip arthroplasty in the Medicare population has been developed.
  • Additional validation in other claims and administrative databases are needed.
  • Similar algorithms should be developed and validated for revision and implant removal in other joints.

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES

This work was supported by the FDA through the Department of Health and Human Services (HHS) contract number HHSF223200910006I. The study sponsors had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. The views expressed in this document do not necessarily reflect the official policies of the Department of Health and Human Services, nor does the mention of trade names, commercial practices, or organizations imply endorsement by the US government. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. SUMMARY AND DISCUSSION OF ALGORITHMS USED IN STUDIES WITH VALIDATION
  7. DISCUSSION
  8. CONCLUSIONS
  9. ETHICAL APPROVAL
  10. CONFLICT OF INTEREST
  11. ACKNOWLEDGEMENTS
  12. REFERENCES
  • 1
    NIH Consensus Statement on total knee replacement December 8–10, 2003. J Bone Joint Surg Am 2004; 86-A(6): 13281335.
  • 2
    Kane RL, Saleh KJ, Wilt TJ, Bershadsky B. The functional outcomes of total knee arthroplasty. J Bone Joint Surg Am Aug 2005; 87(8): 17191724.
  • 3
    Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am 2007; 89(4): 780785.
  • 4
    Jarrett N, Lux L, West S. Systematic evaluation of health outcome of interest definitions in observational studies and clinical definitions for the Observational Medical Outcomes Partnership: angioedema: report. Available at: http://omop.fnih.org/sites/default/files/RTI%20Angioedema%20Final%20Report%20110509.pdf [7 June 2010].
  • 5
    Carnahan RM, Moores KG. Mini-Sentinel's systematic reviews of validated methods for identifying health outcomes using administrative and claims data: methods and lessons learned. Pharmacoepidemiol Drug Saf 2012; 21(S1): 8289.
  • 6
    Coyte PC, Hawker G, Croxford R, Wright JG. Rates of revision knee replacement in Ontario, Canada. J Bone Joint Surg Am 1999; 81(6): 773782.
  • 7
    Heck DA, Melfi CA, Mamlin LA, et al. Revision rates after knee replacement in the United States. Med Care 1998; 36(5): 661669.
  • 8
    Katz J, Losina E, Barrett J, et al. Association between hospital and surgeon procedure volume and outcomes of total hip replacement in the United States Medicare population. J Bone Joint Surg Am 2001; 83-A(11): 1622–1629.
  • 9
    Mahomed NN, Barrett JA, Katz JN, et al. Rates and outcomes of primary and revision total hip replacement in the United States Medicare population. J Bone Joint Surg Am 2003; 85-A(1): 2732.
  • 10
    Katz JN, Wright EA, Baron JA, et al. Predictive value of Medicare claims data for identifying revision of index hip replacement was modest. J Clin Epidemiol 2010; 64(5): 543546.
  • 11
    Abularrage CJ, Weiswasser JM, Dezee KJ, Slidell MB, Henderson WG, Sidawy AN. Predictors of lower extremity arterial injury after total knee or total hip arthroplasty. J Vasc Surg 2008; 47(4): 803807; discussion 807–808.
  • 12
    Bolognesi MP, Marchant MH, Jr., Viens NA, Cook C, Pietrobon R, Vail TP. The impact of diabetes on perioperative patient outcomes after total hip and total knee arthroplasty in the United States. J Arthroplasty 2008; 23(6 Suppl 1): 9298.
  • 13
    Bozic KJ, Kurtz SM, Lau E, et al. The epidemiology of revision total knee arthroplasty in the United States. Clin Orthop Relat Res 2010; 468(1): 4551.
  • 14
    Hervey SL, Purves HR, Guller U, Toth AP, Vail TP, Pietrobon R. Provider Volume of Total Knee Arthroplasties and Patient Outcomes in the HCUP-Nationwide Inpatient Sample. J Bone Joint Surg Am 2003; 85-A(9): 17751783.
  • 15
    Koval KJ, Zhou W, Sparks MJ, Cantu RV, Hecht P, Lurie J. Complications after ankle fracture in elderly patients. Foot Ankle Int 2007; 28(12): 12491255.
  • 16
    Manley M, Ong K, Lau E, Kurtz SM. Total knee arthroplasty survivorship in the United States Medicare population: effect of hospital and surgeon procedure volume. J Arthroplasty 2009; 24(7): 10611067.
  • 17
    Ong KL, Lau E, Manley M, Kurtz SM. Effect of procedure duration on total hip arthroplasty and total knee arthroplasty survivorship in the United States Medicare population. J Arthroplasty 2008; 23(6 Suppl 1): 127132.
  • 18
    Rastogi A, Mohr BA, Williams JO, Soobader MJ, de Brantes F. Prometheus payment model: application to hip and knee replacement surgery. Clin Orthop Relat Res 2009; 467(10): 25872597.
  • 19
    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): 21432149.
  • 20
    Zhan C, Kaczmarek R, Loyo-Berrios N, Sangl J, Bright RA. Incidence and short-term outcomes of primary and revision hip replacement in the United States. J Bone Joint Surg Am. 2007; 89(3): 526533.
  • 21
    Aynardi M, Pulido L, Parvizi J, Sharkey PF, Rothman RH. Early mortality after modern total hip arthroplasty. Clin Orthop Relat Res 2009; 467(1): 213218.
  • 22
    Bozic KJ, Kurtz SM, Lau E, Ong K, Vail TP, Berry DJ. The epidemiology of revision total hip arthroplasty in the United States. J Bone Joint Surg Am 2009; 91(1): 128133.
  • 23
    Cram P, Vaughan-Sarrazin MS, Wolf B, Katz JN, Rosenthal GE. A comparison of total hip and knee replacement in specialty and general hospitals. J Bone Joint Surg Am 2007; 89(8): 16751684.
  • 24
    Hawker G, Wright J, Coyte P, et al. Health-related quality of life after knee replacement. J Bone Joint Surg Am. 1998; 80(2): 163173.
  • 25
    Hudson JI, Kenzora JE, Hebel JR, et al. Eight-year outcome associated with clinical options in the management of femoral neck fractures. Clin Orthop Relat Res 1998; 348: 5966.
  • 26
    Kurtz SM, Ong KL, Schmier J, et al. Future clinical and economic impact of revision total hip and knee arthroplasty. J Bone Joint Surg Am. 2007; 89(Suppl 3): 144151.
  • 27
    Kurtz SM, Lau E, Ong K, Zhao K, Kelly M, Bozic KJ. Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030. Clin Orthop Relat Res 2009; 467(10): 26062612.
  • 28
    Kurtz SM, Ong KL, Schmier J, Zhao K, Mowat F, Lau E. Primary and revision arthroplasty surgery caseloads in the United States from 1990 to 2004. J Arthroplasty 2009; 24(2): 195203.
  • 29
    Kurtz SM, Ong KL, Lau E, Bozic KJ, Berry D, Parvizi J. Prosthetic joint infection risk after TKA in the Medicare population. Clin Orthop Relat Res 2010; 468(1): 5256.
  • 30
    Losina E, Barrett J, Baron JA, Levy M, Phillips CB, Katz JN. Utilization of low-volume hospitals for total hip replacement. Arthritis Rheum 2004; 51(5): 836842.
  • 31
    Losina E, Barrett J, Mahomed NN, Baron JA, Katz JN. Early failures of total hip replacement: effect of surgeon volume. Arthritis Rheum 2004; 50(4): 13381343.
  • 32
    Lyman S, Jones EC, Bach PB, Peterson MG, Marx RG. The association between hospital volume and total shoulder arthroplasty outcomes. Clin Orthop Relat Res 2005; 432: 132137.
  • 33
    Mahomed NN, Barrett J, Katz JN, Baron JA, Wright J, Losina E. Epidemiology of total knee replacement in the United States Medicare population. J Bone Joint Surg Am 2005; 87(6): 12221228.
  • 34
    Manley M, Ong K, Lau E, Kurtz SM. Effect of volume on total hip arthroplasty revision rates in the United States Medicare population. J Bone Joint Surg Am 2008; 90(11): 24462451.
  • 35
    Ong KL, Mowat FS, Chan N, Lau E, Halpern MT, Kurtz SM. Economic burden of revision hip and knee arthroplasty in Medicare enrollees. Clin Orthop Relat Res 2006; 446: 2228.
  • 36
    Kreder HJ, Williams JI, Jaglal S, Axcell T, Stephen D. A population study in the Province of Ontario of the complications after conversion of hip or knee arthrodesis to total joint replacement. Can J Surg 1999; 42(6): 433439.
  • 37
    Paterson JM, Williams JI, Kreder HJ, et al. Provider volumes and early outcomes of primary total joint replacement in Ontario. Can J Surg 2010; 53(3): 175183.
  • 38
    Bozic KJ, Ong K, Lau E, et al. Risk of Complication and Revision Total Hip Arthroplasty Among Medicare Patients with Different Bearing Surfaces. Clin Orthop Relat Res 2010; 468(9): 23572362.
  • 39
    Hagen TP, Vaughan-Sarrazin MS, Cram P. Relation between hospital orthopaedic specialisation and outcomes in patients aged 65 and older: retrospective analysis of US Medicare data. BMJ 2010; 340: c165.
  • 40
    SooHoo NF, Lieberman JR, Farng E, Chambers L, Zingmond DS. Factors That Predict Short-term Complication Rates After Total Hip Arthroplasty. Clin Orthop Relat Res 2010; 468(9): 23632371.
  • 41
    Keating JF, Orfaly R, O'Brien PJ. Knee pain after tibial nailing. J Orthop Trauma 1997; 11(1): 1013.
  • 42
    Khuri SF, Daley J, Henderson W, et al. Risk adjustment of the postoperative mortality rate for the comparative assessment of the quality of surgical care: results of the National Veterans Affairs Surgical Risk Study. J Am Coll Surg 1997; 185(4): 315327.
  • 43
    Kurtz SM, Lau E, Schmier J, Ong KL, Zhao K, Parvizi J. Infection burden for hip and knee arthroplasty in the United States. J Arthroplasty 2008; 23(7): 984991.
  • 44
    Lagaay PM, Hamilton GA, Ford LA, Williams ME, Rush SM, Schuberth JM. Rates of revision surgery using Chevron-Austin osteotomy, Lapidus arthrodesis, and closing base wedge osteotomy for correction of hallux valgus deformity. J Foot Ankle Surg 2008; 47(4): 267272.
  • 45
    Lu X, Hagen TP, Vaughan-Sarrazin MS, Cram P. The impact of physician-owned specialty orthopaedic hospitals on surgical volume and case complexity in competing hospitals. Clin Orthop Relat Res 2009; 467(10): 25772586.
  • 46
    Lu X, Hagen TP, Vaughan-Sarrazin MS, Cram P. The impact of new hospital orthopaedic surgery programs on total joint arthroplasty utilization. J Bone Joint Surg Am 2010; 92(6): 13531361.
  • 47
    Rashiq S, Finegan BA. The effect of spinal anesthesia on blood transfusion rate in total joint arthroplasty. Can J Surg 2006; 49(6): 391396.
  • 48
    SooHoo NF, Lieberman JR, Ko CY, Zingmond DS. Factors predicting complication rates following total knee replacement. J Bone Joint Surg Am 2006; 88(3): 480485.
  • 49
    Soohoo NF, Zingmond DS, Lieberman JR, Ko CY. Primary total knee arthroplasty in California 1991 to 2001: does hospital volume affect outcomes? J Arthroplasty 2006; 21(2): 199205.
  • 50
    Soohoo NF, Zingmond DS, Lieberman JR, Ko CY. Optimal timeframe for reporting short-term complication rates after total knee arthroplasty. J Arthroplasty 2006; 21(5): 705711.
  • 51
    SooHoo NF, Farng E, Zingmond DS. Incidence of pulmonary embolism following surgical treatment of metatarsal fractures. Foot Ankle Int 2010; 31(7): 600603.
  • 52
    Bozic KJ, Durbhakula S, Berry DJ, et al. Differences in patient and procedure characteristics and hospital resource use in primary and revision total joint arthroplasty: a multicenter study. J Arthroplasty 2005; 20(7 Suppl 3): 1725.
  • 53
    Deshmukh AV, Koris M, Zurakowski D, Thornhill TS. Total shoulder arthroplasty: long-term survivorship, functional outcome, and quality of life. J Shoulder Elbow Surg 2005; 14(5): 471479.
  • 54
    Gioe TJ, Killeen KK, Hoeffel DP, et al. Analysis of unicompartmental knee arthroplasty in a community-based implant registry. Clin Orthop Relat Res 2003; 416: 111119.
  • 55
    Gioe TJ, Killeen KK, Grimm K, Mehle S, Scheltema K. Why are total knee replacements revised?: analysis of early revision in a community knee implant registry. Clin Orthop Relat Res 2004(428): 100106.
  • 56
    Lyman S, Dunn WR, Spock C, Bach PB, Mandl LA, Marx RG. Validity of same-side reoperation after total hip and knee arthroplasty using administrative databases. J Knee Surg 2009; 22(1): 1720.
  • 57
    Singh JA, Ayub S. Accuracy of VA Databases for Diagnoses of Knee Replacement and Hip Replacement. Osteoarthritis Cartilage 2010; 18(12): 16391642.
  • 58
    Wright RJ, Sledge CB, Poss R, Ewald FC, Walsh ME, Lingard EA. Patient-reported outcome and survivorship after Kinemax total knee arthroplasty. J Bone Joint Surg Am 2004; 86-A(11): 24642470.
  • 59
    Khuri SF, Daley J, Henderson W, et al. The National Veterans Administration Surgical Risk Study: risk adjustment for the comparative assessment of the quality of surgical care. J Am Coll Surg 1995; 180(5): 519531.
  • 60
    Jafari SM, Coyle C, Mortazavi SM, Sharkey PF, Parvizi J. Revision hip arthroplasty: infection is the most common cause of failure. Clin Orthop Relat Res 2010; 468(8): 20462051.