Secondary Logo

Journal Logo

SELECTED PAPERS FROM THE 9TH INTERNATIONAL CONGRESS OF ARTHROPLASTY REGISTRIES GUEST EDITOR OLA ROLFSON MD, PHD

Does a Prescription-based Comorbidity Index Correlate with the American Society of Anesthesiologists Physical Status Score and Mortality After Joint Arthroplasty? A Registry Study

Kerr, Mhairi M. BSc (Hons), MSc1; Graves, Stephen E. DPhil (Oxon), FAOrthA1,2; Duszynski, Katherine M. BSc, GDPH1; Inacio, Maria C. PhD3; de Steiger, Richard N. FRACS(Orth), PhD2,4; Harris, Ian A. FRACS, PhD2,5; Ackerman, Ilana N. PhD6; Jorm, Louisa R. PhD7; Lorimer, Michelle F. BSc (Maths & Comp Sc), Hons (Stats)2,8; Gulyani, Aarti MPhil (Stats)1; Pratt, Nicole L. PhD1

Author Information
Clinical Orthopaedics and Related Research: October 2021 - Volume 479 - Issue 10 - p 2181-2190
doi: 10.1097/CORR.0000000000001895

Abstract

Introduction

The American Society of Anesthesiologists (ASA) classification system, applied preoperatively by an anesthesiologist, is a simple categorization of a patient’s overall health [12]. It classifies patients into five groups, with the lowest score of 1 assigned to a patient in good health and 5 assigned to a moribund patient [12]. In a study of more than 2 million patients who were hospitalized and undergoing a range of procedures, ASA scores were associated with morbidity and mortality postoperatively [12]. Specifically for joint arthroplasty, ASA scores of ≥ 3 have been associated with a threefold higher risk of readmission [28]. The ASA score has also been found to be associated with postoperative infection, increased mortality, and longer length of hospital stay [24, 25].

An important consideration in the assignment of ASA scores is the presence of comorbidities. Studies have identified that the presence of multiple comorbidities is associated with postoperative complications, longer hospital stay, and increased mortality [4, 18, 33]. In a study of 131,447 patients undergoing hip or knee arthroplasty, more than 60% of these patients had at least one comorbidity [14]. Comorbidity algorithms such as the Charlson comorbidity index and Elixhauser comorbidity scores, which require documented hospital diagnoses [5, 8], have been validated against the risk of mortality [20, 31, 32], and the ASA score has been shown to be correlated with the Charlson comorbidity index [11, 34]. Because patient characteristics such as comorbidities influence the occurrence of outcomes including death, ASA and comorbidity scores can be useful for confounding adjustment when examining outcomes after joint arthroplasty. Unfortunately, such comorbidity scores are not always core components of registry data collections. The Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR), for example, began collecting procedural data in 2002 [3]; however, the registry only commenced collecting ASA scores in 2012. Furthermore, variation in ASA distribution between registries has been identified, which has implications for comparing outcomes between arthroplasty registries [29]. Because not all registries consistently record the ASA score [21] and because there is variability among clinicians when assigning scores [26, 27], other comorbidity scores may be an alternative measure of a patient’s health status. One such score, the Rx-Risk [23], uses only medication-dispensing records to indicate the presence of specific comorbidities that require medicines to treat the condition [6, 23]. Previous studies have found that the Rx-Risk is associated with 1-year and 5-year risk of revision after joint replacement, postoperative infection at 90 days, and 90-day and 1-year mortality in patients undergoing THA or TKA [15-17]. Although the relationship between the ASA score and comorbidities has been examined for selected comorbidity indices [11, 34], to our knowledge, the relationship between ASA and Rx-Risk score has not been assessed. Additionally, we do not know whether a side-by-side comparison of the validity of the ASA and Rx-Risk scores against mortality in the same patient population has been undertaken.

Therefore, using Australian joint replacement registry data linked with pharmaceutical-dispensing data, our aim was to evaluate the relationship between the ASA score and Rx-Risk score in patients undergoing THA or TKA for osteoarthritis. Specifically, our questions were: (1) Which metric, the Rx-Risk score or the ASA score, correlates more closely with 30- and 90-day mortality after TKA or THA? (2) Is the Rx-Risk score correlated with the ASA score?

Patients and Methods

Study Design and Data Sources

We performed a retrospective study using data from the AOANJRR, which includes data on hip, knee, and shoulder arthroplasty in Australia starting in September 1999 [3]. Registry data collate the prosthesis components used, surgery date, joint laterality, surgery type, surgery reason, as well as patient age and gender. The ASA score has been collected since December 2012, and BMI has been collected since January 2015. The ASA score is determined by anesthesiologists in the operating room during surgery. The ASA score is complete for 92.2% of procedures registered by the AOANJRR since December 2012 [3]. The AOANJRR collects data on more than 98.8% of procedures undertaken nationwide in Australia [3]. Arthroplasty data submitted to the registry are validated against patient-level data provided by Australia’s health departments to maximize the completeness of arthroplasty data capture [3].

The AOANJRR is routinely linked to the National Death Index (NDI) database to determine fact of death and date of death for all patients recorded on the registry. The NDI is administered by the Australian Institute of Health and Welfare (AIHW) and contains death registration information collated from Australia’s eight jurisdictions. Other NDI variables include demographics, date of death, underlying causes of death, and other cause-of-death information. Cause of death is categorized by the Australian Bureau of Statistics using the ICD-10 with nonspecific causes revised up to 2 years after registration of death [1]. Patients registered in the AOANJRR are matched biannually to those in the NDI by the AIHW using a weighted multistage process. The AOANJRR staff then completes a clerical review of AIHW-identified matches, with a conservative approach to confirming matches that might contribute to underreporting of death. Patients who die overseas are not recorded in the NDI, further impacting underreporting of the occurrence of death in the registry.

In 2018, AOANJRR data up to December 2017 (the most recent data available) was also linked to the Australian Government Pharmaceutical Benefit Scheme (PBS) administrative claims database to obtain records of all government-subsidized medicines dispensed to patients. The PBS database is maintained by the Australian Government’s Services Australia department, and contains information on the dispensing of prescription medications and demographic information regarding gender, year of birth, and year of death. Complete pharmaceutical dispensing data captured from July 2002 to December 2017 were obtained for all patients recorded in the registry. Using probabilistic data linkage, the AIHW was able to link 95% of all hip arthroplasty procedures and 96.3% of knee arthroplasty procedures in the AOANJRR to PBS data.

Participants

The study population included all patients registered in the AOANJRR with a primary conventional THA or TKA performed in Australia between June 1, 2013 to December 31, 2017 (134,389 patients who underwent 148,205 THAs and 185,907 patients who underwent 220,028 TKAs). For THA procedures, we included only patients with a conventional total hip; we excluded those with a resurfacing hip procedure. All patients were linked to Medicare Benefits Schedule health services and PBS medicine dispensing information. The June 2013 study date ensured we captured all PBS-subsidized medications among the eligible patients for preoperative comorbidity assessment 1 year prior. We included those with a primary diagnosis of osteoarthritis (the principal surgical diagnosis for most patients who underwent THA [88% {131,336 of 148,205}] and TKA [98% {215,712 of 220,028}]) [3]. Procedures performed for other reasons were excluded (16,209 THA patients and 4008 TKA patients). This yielded 119,076 patients undergoing 131,336 THAs and 182,445 patients undergoing 215,712 TKAs. A further 3% (3575 of 131,336 THAs) and 2% (5211 of 215,712 TKAs) were excluded because of missing ASA information recorded in the AOANJRR. The final cohort included 127,761 THAs in 116,021 patients and 210,501 TKAs in 178,350 patients (Fig. 1). The AOANJRR classifies a THA as a procedure which involves insertion of the following prosthesis types: shell (or cup which is an integrated shell and hip insert/liner), hip insert/liner (and/or use of dual mobility liner), stem, and head. A TKA procedure includes a femoral component, tibial component, and tibial insert/bearing with optional patella button.

F1
Fig. 1:
This flow diagram details participant selection from contributing data sources. The Medicare Benefits Schedule is a list of healthcare services under Australia’s publicly funded healthcare scheme known as Medicare, and it entitles eligible Australian residents to free treatment in all public hospitals and a rebate for medical services including treatment from medical practitioners, eligible midwives, nurse practitioners, and allied health professionals; AOANJRR = Australian Orthopaedic Association National Joint Replacement Registry; MBS = Medicare Benefits Schedule; PBS = Pharmaceutical Benefits Scheme.

Baseline Characteristics of the Study Population

In all, 53% (68,037 of 127,761) of THAs and 56% (117,337 of 210,501) of TKAs were performed in women, and the mean ± SD age at surgery was 68 ± 11 years for THAs and 68 ± 9 years for TKAs (Table 1).

Table 1. - Demographic characteristics of patients undergoing a primary THA or TKA, by ASA category
ASA score % (n) Age (years), mean ± SD Gender, % women RxRisk score, median (IQR)
THA All procedures 100 (127,761) 68 ± 11 53
1 10 (12,932) 59 ± 11 45 1 (0-3)
2 56 (71,199) 67 ± 10 54 3 (2-5)
3 33 (41,801) 71 ± 10 54 5 (4-7)
4 1 (1818) 75 ± 10 47 7 (5-9)
5 < 1 (11) 73 ± 7 36 9 (2-6)
TKA All procedures 100 (210,501) 68 ± 9 56
1 6 (12,893) 63 ± 9 45 2 (1-3)
2 56 (118,430) 68 ± 9 57 4 (2-5)
3 37 (76,886) 70 ± 9 56 5 (4-7)
4 1 (2279) 72 ± 10 47 7 (5-9)
5 < 1 (13) 71 ± 9 69 5 (4-6)

Variables

Measures Used in Correlation Assessment

Rx-Risk Score

Used in studies to estimate utilization of healthcare resources [10, 30] and mortality [9, 17, 19, 23], the Rx-Risk score is calculated as the number of different health conditions for which an individual is treated, with a possible score ranging from 0 to 47, representing the 47 health conditions. A complete list of the medicines and medication codes used in the Rx-Risk index has been published elsewhere [23]. For each Rx-Risk category, medications used to treat that condition are mapped to the category. An individual with at least one dispensing of a medicine in a given category is considered to have been treated using medications for that health condition. The unweighted Rx-Risk score was calculated using all PBS-dispensed medications recorded in the year preceding the admission date for THA or TKA. The day of arthroplasty admission was not used to calculate Rx-Risk scores because medications dispensed on this day could be related to the surgery itself. The validity of the Rx-Risk with 1-year mortality has been assessed in the Australian PBS population [20]. Previous studies in an elderly population of THA patients found that the index was associated with 90-day and 1-year mortality and risk of 1- and 5-year revision [15-17].

ASA Score

The ASA score was identified using the scores recorded at time of surgery and collated for the AOANJRR dataset (Table 1).

Primary and Secondary Study Outcomes

Our primary goal was to determine the concordance between the ASA and Rx-Risk scores and mortality at 30- and 90-days after surgery. All-cause mortality was established using information from the NDI dataset. We calculated the c-statistic from the baseline model, including only age and sex, and determined whether the separate addition of each score to the baseline model improved the concordance with mortality. The secondary goal was to determine the association between the Rx-Risk score and the ASA and whether there were specific conditions identified by Rx-Risk that were most prevalent within each ASA level.

Ethical Approval

Ethical approval for the study was granted by the University of South Australia Human Research Ethics Committee (0000035831) and the Australian Institute of Health and Welfare Ethics Committee (EO2016/4/316).

Statistical Analysis

Separate analyses were conducted for the THA and TKA cohorts. Age, gender, number of comorbidities, ASA score, and Rx-Risk score were analyzed descriptively.

For the first study aim, a logistic regression analysis was used to model the risk of 30-day and 90-day mortality after THA and TKA separately. For each analysis, a baseline model was created including only age and gender as predictors of mortality. Two separate models were then created with the addition of either the ASA or Rx-Risk score, and finally a model was created including age, gender, ASA score, and Rx-Risk score. Model discrimination was compared using the c-statistic. The value of the c-statistic can range from 0 to 1, with 1 indicating perfect concordance and 0.5 indicating chance [13]. Incremental c-statistics for the nested models were compared using the De-Long method [7].

For the second aim, the correlation between the Rx-Risk and ASA scores was estimated using the Spearman correlation coefficient (r), which quantifies the monotonic relationship between the two variables. The Spearman correlation coefficient was used as the ASA and Rx-Risk are ordinal variables and, therefore, we did not expect a linear relationship. The 10 most prevalent Rx-Risk categories within each ASA level were also separately calculated for THA and TKA procedures and are reported descriptively. As fewer than 15 patients with THA or TKA had an ASA score of 5, these patients were excluded from these descriptive analyses.

Results

Is Rx-Risk More Closely Associated with Risk of Death Than the ASA Score?

We found no differences between the ASA and Rx-Risk scores in terms of the association of each of those scores with mortality after THA or TKA (Table 2). The baseline model comprising only age and gender had high concordance with 30-day mortality (THA: c-statistic 0.80 [95% confidence interval 0.76 to 0.84]; TKA: c-statistic 0.71 [95% CI 0.66 to 0.75]) and 90-day mortality (THA: c-statistic 0.77 [95% CI 0.74 to 0.80]; TKA: c-statistic 0.72 [95% CI 0.69 to 0.75]). Including the ASA score to the baseline model increased concordance with 30-day mortality for both THA and TKA (ASA: THA c-statistic 0.83 [95% CI 0.79 to 0.86]; p = 0.005; TKA c-statistic 0.73 [95% CI 0.69 to 0.78]; p = 0.012) (Table 2). Including the Rx-Risk score also increased concordance with 30-day mortality compared to the base model for both THA and TKA (Rx-Risk: THA c-statistic 0.82 [95% CI 0.79 to 0.86]; TKA c-statistic 0.74 [95% CI 0.70 to 0.79]) (Table 2). Results for 90-day mortality were similar.

Table 2. - Associations between ASA and Rx-Risk scores and 30-day and 90-day mortality after THA or TKA
Models 30-day mortality 90-day mortality
C-statistic (95% CI)a p value compared with base modelb C-statistic (95% CI)b p value compared with base modelb
THA
 Base model (age and gender) 0.80 (0.76-0.84) 0.77 (0.74-0.80)
 Base model and ASA score 0.83 (0.79-0.86) 0.005 0.80 (0.77-0.83) < 0.001
 Base model and Rx-Risk score 0.82 (0.79-0.86) 0.003 0.79 (0.76-0.82) < 0.001
 Base model and ASA and Rx-Risk score 0.84 (0.80-0.87) 0.001 0.81 (0.78-0.84) < 0.001
TKA
 Base model (age and gender) 0.71 (0.66-0.75) 0.72 (0.69-0.75)
 Base model and ASA score 0.73 (0.69-0.78) 0.012 0.74 (0.71-0.78) < 0.001
 Base model and Rx-Risk score 0.74 (0.70-0.79) < 0.001 0.75 (0.72-0.78) < 0.001
 Base model and ASA and Rx-Risk score 0.75 (0.71-0.80) < 0.001 0.76 (0.73-0.79) < 0.001
aPossible range 0 to 1, with 1 indicating perfect agreement and 0.5 indicating chance agreement.
bCompared c-statistics (nested models) using the DeLong test.

Including both the ASA and Rx-Risk to the base model had the highest concordance in all models tested (Table 2).

Is the Rx-Risk Score Correlated with ASA Score?

The patient comorbidity burden as described by an increasing ASA score was weakly correlated with an increasing Rx-Risk score in both THA (r = 0.45) and TKA (r = 0.38). The median number of comorbidities in the THA cohort, as determined by the Rx-Risk score, was one for patients with an ASA score of 1 (interquartile range 0 to 3) and increased to seven for those with an ASA score of 4 (IQR 5 to 9). In the TKA cohort, the median number of comorbidities was two for patients with an ASA score of 1 (IQR 1 to 3) and seven for those with an ASA score of 4 (IQR 5 to 9) (Table 1).

In patients undergoing THA with an ASA score of 1 or 2, the most prevalent condition for which medicines were used was inflammation pain, identified by the use of NSAIDs (THA: ASA score of 1 = 51%; ASA score of 2 = 54%), and in those with an ASA score of 3 or 4, the most common condition was pain, identified by the use of opioids (62% and 68% for patients with scores of 3 and 4, respectively) (Fig. 2). A similar pattern was observed for patients who underwent TKA; however, those with an ASA score of 3 or 4 also had other common comorbidities including hypertension and hyperlipidemia (Fig. 3).

F2
Fig. 2:
This graph shows the 10 most prevalent Rx-Risk categories within each ASA level for patients who underwent THA; GERD = gastroesophageal reflux disease; IHD hypertension = ischemic heart disease hypertension.
F3
Fig. 3:
This graph shows the 10 most prevalent Rx-Risk categories within each ASA level for patients who underwent TKA; GERD = gastroesophageal reflux disease; IHD hypertension = ischemic heart disease hypertension.

Discussion

The presence of multiple comorbidities in patients undergoing joint replacement procedures has been associated with poor outcomes and increased mortality. The ASA score is used to indicate the overall health of a patient at the time of surgery and takes into consideration some patient comorbidity; however, ASA scores have not always been routinely collected by registries. Efforts to enhance the measurement of patient health are becoming increasingly important, as such information can be useful for confounding adjustment when examining outcomes after joint arthroplasty, and specific comorbidity profiles can also be important to help identify groups of patients that may be at high risk of early death. In this study, we found that both the ASA and Rx-Risk scores had high concordance with 30-day and 90-day postoperative mortality, and when both scores were included together, concordance improved over and above age and gender alone. Despite their strong association with mortality, the ASA and Rx-Risk scores were weakly correlated.

Limitations

The limitations of our study include the recent availability of ASA score data beginning in December 2012. Information on the ASA score before 2013 may demonstrate a different magnitude of correlation that may or may not change the strength of the relationship between these measures. The restriction of more current ASA information in the study does, however, reflect a more contemporary patient profile undergoing THA or TKA. A recent evaluation of change in comorbidities over time by the AOANJRR showed no demonstrable variation in comorbidity burden for patients undergoing arthroplasty between 2002 and 2017 [3]. This consistency in comorbidity burden over a 15-year period suggests that the observed relationship between the ASA and Rx-Risk scores may be similarly maintained for patients undergoing THA or TKA at earlier timepoints. However, the profile of individual comorbidities in those undergoing THA or TKA may change over a longer duration of AOANJRR data collection, thereby contributing to a higher ASA classification assigned at surgery. Another limitation of the Rx-Risk is that it relies on medication-dispensing records. If medicines are purchased over the counter, they will not be captured in healthcare datasets, therefore, some conditions identified by the use of over-the-counter medicines will be underestimated. If those conditions are associated with ASA, it is likely that the correlation will also be underestimated. Additionally, there are some conditions such as mental health disorders (anxiety, depression) and osteoporosis identified by the Rx-Risk that clinicians may not consider when the ASA is assigned. The effect of this will also be an underestimation of the correlation between the scores. Medication dispensing is variable across countries and different healthcare systems and as such, further research is needed to determine whether the results of this study are generalizable to other national registry datasets. Finally, missing or incorrect mortality information for the registry study cohort may contribute to under-reporting of deaths when evaluating the validity of the ASA and Rx-Risk scores in predicting short-term mortality. However, the magnitude of missing mortality information is likely marginal, with a nondifferential impact across the two scoring systems when assessing the predictive validity of mortality risk in patients undergoing THA or TKA.

Is Rx-Risk More Closely Associated with Risk of Death Than the ASA Score?

Both the ASA and Rx-Risk were comparable in terms of their individual association with death after arthroplasty and both improved the concordance over and above age and sex alone. The inclusion of both ASA and Rx-Risk was associated with the highest discrimination in all models tested, suggesting that both scores are useful and may capture different information. Although many national arthroplasty registries collect the ASA score, including the Norwegian Arthroplasty Registry (since 2005), Finnish Arthroplasty Registry (since May 2014), Swiss National Joint Registry (since 2015), and the AOANJRR [29], full-population capture of the score has only recently been implemented in these registries. For example, the ASA was first collected by the AOANJRR in 2012; however, 10 years of data are available before this date. Additionally, large variation in ASA class distribution has been demonstrated across global arthroplasty registries, likely due to differences in comorbidity profile and access to arthroplasty [29]. Large-scale administrative datasets including entire-population hospital morbidity data and dispensing data are becoming increasingly available and, when linked to registry datasets, can be used to supplement registry data where gaps in clinical information exist.

Is the Rx-Risk Score Correlated with the ASA Score?

In this study, we identified that the ASA score, a subjective measure of health at the time of surgery, was weakly correlated with Rx-Risk, an objective measure of comorbidity based on prescriptions dispensed preoperatively. This finding is likely due to the different scales used by the scores. The ASA score is an ordered scoring system; that is, 1 indicates normal health and 5 represents moribund health, and the Rx-Risk score summarizes all comorbidity categories for which a medicine was dispensed, and an increasing Rx-Risk score does not always correspond to decreasing health. The Rx-Risk includes medications for active treatment of disease as well as disease prevention. For example, bisphosphonates are indicated to prevent fractures owing to osteoporosis, and statins are indicated to prevent cardiac outcomes associated with hypercholesterolemia. Although the correlation between ASA and Rx-Risk was weak, the advantage of the Rx-Risk measure is that it may provide more specific information about individual comorbidities that are actively managed, and may help to identify specific conditions or clinical features outside those broadly considered when making an ASA assessment that place patients at increased risk of poor outcomes after surgery. For example, in this study we found that for patients undergoing THA or TKA who had an ASA score of 1 or 2, the most common comorbidity was pain that was treated using anti-inflammatory medications, and patients with an ASA score of 3 or 4 had higher prevalence of opioid use to treat pain; hence, patients with a higher ASA score may experience more severe pain. In patients with an ASA score of 3 or 4, common comorbidities were related to existing cardiovascular comorbidities, and there was a high prevalence of the use of antihypertensive medicines and anticoagulants. Another study has also identified that patients with heart disease have a threefold increased likelihood of mortality within 90 days after arthroplasty (pooled odds ratio 2.96 [95% CI 1.95 to 4.48]) [22]. Work by the AOANJRR is ongoing regarding the collection of pre- and postoperative patient-reported outcome measures (PROMs) to identify patients who may experience adverse surgical outcomes [2]. However, collection of medication information is not integrated with the PROMs process. Future studies investigating the association of PROMS with poor surgical outcomes could incorporate medication profile and usage to further identify arthroplasty patients who would benefit from enhanced postoperative care.

Conclusion

Monitoring joint arthroplasty outcomes is a critical function of arthroplasty registries; however, confounding is a particular problem in this assessment when the treatment choice is influenced by the severity of the underlying condition or the health of the patient. Although ASA and Rx-Risk were weakly to moderately correlated, they did demonstrate a monotonic relationship, and each score was highly associated with the probability of mortality shortly after undergoing a procedure. The ASA score is not always routinely collected by registries, hence the Rx-Risk score may be a useful additional measure of overall patient health, with the advantage of defining conditions for which a patient receives pharmaceutical treatment. Medication information may be useful to further understand risk factors in patients with potential for poor outcomes after joint replacement surgery.

Acknowledgments

We thank the Commonwealth Department of Health, assisted by the Commonwealth Department of Human Services, for access to the national Medicare Benefits Schedule and Pharmaceutical Benefits Scheme datasets. We thank the services of the Australian Institute of Health and Welfare Data Integration Services Centre for facilitating provision and linkage of the National Death Index, Medicare Benefits Schedule, and Pharmaceutical Benefits Scheme datasets with the Australian Orthopaedic Association National Joint Replacement Registry.

References

1. Australian Bureau of Statistics. Causes of death, Australia methodology – technical note: causes of death revisions, 2017 final data. Available at: https://www.abs.gov.au/methodologies/causes-death-australia-methodology/2019#technical-note-causes-of-death-revisions-2017-final-data. Accessed May 30, 2021.
2. Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR). AOA PROMs pilot project: final report. Available at: https://aoanjrr.sahmri.com/documents/10180/681914/AOANJRR+PROMs+Pilot+Final+Report. Accessed June 1, 2021.
3. Australian Orthopaedic Association National Joint Replacement Registry. Hip, Knee and Shoulder Arthroplasty: 2020 Annual Report. AOA; 2020.
4. Bozic KJ, Lau E, Kurtz S, Ong K, Berry DJ. Patient-related risk factors for postoperative mortality and periprosthetic joint infection in Medicare patients undergoing TKA. Clin Orthop Relat Res. 2012;470:130-137.
5. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383.
6. Clark DO, Von Korff M, Saunders K, Baluch WM, Simon GE. A chronic disease score with empirically derived weights. Med Care. 1995;33:783-795.
7. DeLong ER. Comparing the areas under two or more correlated receiver operator characteristic curves: a non-parametric approach. Biometrics. 1988;44:837-845.
8. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8-27.
9. Fan VS, Maciejewski ML, Liu C-F, McDonell MB, Fihn SD. Comparison of risk adjustment measures based on self-report, administrative data, and pharmacy records to predict clinical outcomes. Health Serv Outcomes Res Method. 2006;6:21-36.
10. Fishman PA, Goodman MJ, Hornbrook MC, Meenan RT, Bachman DJ, O'Keeffe Rosetti MC. Risk adjustment using automated ambulatory pharmacy data: the RxRisk model. Med Care. 2003;41:84-99.
11. Guo R, Yu W, Meng Y, et al. Correlation of ASA grade and the Charlson comorbidity index with complications in patients after transurethral resection of prostate. Urology. 2016;98:120-125.
12. Hackett NJ, De Oliveira GS, Jain UK, Kim JY. ASA class is a reliable independent predictor of medical complications and mortality following surgery. Int J Surg. 2015;18:184-190.
13. Hosmer DW. Applied Logistic Regression. Wiley; 2000.
14. Hutchings A, Neuburger J, Grosse Frie K, Black N, van der Meulen J. Factors associated with non-response in routine use of patient reported outcome measures after elective surgery in England. Health Qual Life Outcomes. 2012;10:34.
15. Inacio MC, Pratt NL, Roughead EE, Graves SE. Predicting infections after total joint arthroplasty using a prescription based comorbidity measure. J Arthroplasty. 2015;30:1692-1698.
16. Inacio MC, Pratt NL, Roughead EE, Graves SE. Using medications for prediction of revision after total joint arthroplasty. J Arthroplasty. 2015;30:2061-2070.
17. Inacio MCS, Pratt NL, Roughead EE, Graves SE. Evaluation of three co-morbidity measures to predict mortality in patients undergoing total joint arthroplasty. Osteoarthritis Cartilage. 2016;24:1718-1726.
18. Jain NB, Guller U, Pietrobon R, Bond TK, Higgins LD. Comorbidities increase complication rates in patients having arthroplasty. Clin Orthop Relat Res. 2005;435:232-238.
19. Johnson ML, El-Serag HB, Tran TT, Hartman C, Richardson P, Abraham NS. Adapting the Rx-Risk-V for mortality prediction in outpatient populations. Med Care. 2006;44:793-797.
20. Lu CY, Barratt J, Vitry A, Roughead E. Charlson and Rx-Risk comorbidity indices were predictive of mortality in the Australian health care setting. J Clin Epidemiol. 2011;64:223-228.
21. Lujic S, Watson DE, Randall DA, Simpson JM, Jorm LR. Variation in the recording of common health conditions in routine hospital data: study using linked survey and administrative data in New South Wales, Australia. BMJ Open. 2014;4:e005768.
22. Podmore B, Hutchings A, van der Meulen J, Aggarwal A, Konan S. Impact of comorbid conditions on outcomes of hip and knee replacement surgery: a systematic review and meta-analysis. BMJ Open. 2018;8:e021784.
23. Pratt NL, Kerr M, Barratt JD, et al. The validity of the Rx-Risk Comorbidity Index using medicines mapped to the Anatomical Therapeutic Chemical (ATC) Classification System. BMJ Open. 2018;8:e021122.
24. Rauh MA, Krackow KA. In-hospital deaths following elective total joint arthroplasty. Orthopedics. 2004;27:407-411.
25. Ridgeway S, Wilson J, Charlet A, Kafatos G, Pearson A, Coello R. Infection of the surgical site after arthroplasty of the hip. J Bone Joint Surg Br. 2005;87:844-850.
26. Riley R, Holman C, Fletcher D. Inter-rater reliability of the ASA physical status classification in a sample of anaesthetists in Western Australia. Anaesth Intensive Care. 2014;42:614-618.
27. Sankar A, Johnson SR, Beattie WS, Tait G, Wijeysundera DN. Reliability of the American Society of Anesthesiologists physical status scale in clinical practice. Br J Anaesth. 2014;113:424-432.
28. Schaeffer JF, Scott DJ, Godin JA, Attarian DE, Wellman SS, Mather RC 3rd. The association of ASA class on total knee and total hip arthroplasty readmission rates in an academic hospital. J Arthroplasty. 2015;30:723-727.
29. Silman AJ, Combescure C, Ferguson RJ, et al. International variation in distribution of ASA class in patients undergoing total hip arthroplasty and its influence on mortality: data from an international consortium of arthroplasty registries. Acta Orthop. 2021;92:304-310.
30. Sloan KL, Sales AE, Liu CF, et al. Construction and characteristics of the RxRisk-V: a VA-adapted pharmacy-based case-mix instrument. Med Care. 2003;41:761-774.
31. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57:1288-1294.
32. Thompson NR, Fan Y, Dalton JE, et al. A new Elixhauser-based comorbidity summary measure to predict in-hospital mortality. Med Care. 2015;53:374-379.
33. Wang T, Ackland T, Hall S, Gilbey H, Parsons R. Functional recovery and timing of hospital discharge after primary total hip arthroplasty. Aust N Z J Surg. 1998;68:580-583.
34. Whitmore RG, Stephen JH, Vernick C, et al. ASA grade and Charlson Comorbidity Index of spinal surgery patients: correlation with complications and societal costs. Spine J. 2014;14:31-38.
© 2021 by the Association of Bone and Joint Surgeons