記住我
Hospital length of stay (LOS) and discharge to skilled nursing facilities are two major driving factors for costs associated with total hip arthroplasty (THA) and total knee arthroplasty (TKA).1–3 Decreasing time spent in the hospital after surgery and encouraging discharge home when possible may help to minimize use of healthcare resources, control costs, and align with patient goals.4,5 These initiatives have resulted in more hip and knee arthroplasty patients being discharged home the same day as their surgery, with a reported increase in same-day discharge (SDD) from 0.95% to 20.5% between 2011 and 2016.6 The effect of the COVID-19 pandemic must also be acknowledged, with one study reporting an increase in SDD from 14.0% in 2019 to 33.9% in 2020 after their hospital was forced to convert many procedures to same-day discharge during the pandemic.7
The ability to predict either LOS or discharge destination based on preoperative factors facilitates management of postoperative care and identifies patients at risk for prolonged, complicated recovery.8 In 2003, Oldmeadow et al9 developed the Risk Assessment and Prediction Tool (RAPT), which consists of six questions (age, sex, walking distance, gait aid, community support, and living with a caretaker after surgery) that patients answer preoperatively (Figure 1). Based on the responses, a final score is calculated out of 12, which places the patient at low (>9), medium (6 to 9), or high (<6) risk for extended LOS and discharge to a rehabilitation facility.
Figure 1: Diagram showing the Risk Assessment and Prediction Tool.9Although RAPT may have once demonstrated utility as a preoperative screening tool for discharge disposition at a time when it was common to use postacute rehabilitation facilities, it may be worth reassessing its applicability to a modern setting where patients are discharged earlier and often directly home. We hypothesized that RAPT would not predict SDD for patients undergoing primary TKA or THA. A secondary goal of our study was to determine whether risk factors such as age, BMI, or other comorbidities are associated with extended hospital stay or discharge to rehabilitation.
MethodsAfter Institutional Review Board approval was obtained, data were retrospectively collected from consecutive patients who had undergone primary, elective TKA or THA between February 2020 and May 2021 by three fellowship-trained arthroplasty surgeons at a single, tertiary academic medical center. Age, sex, BMI, medical comorbidities, and type of anesthesia (short acting spinal or general anesthesia following standard protocols) were recorded. TKAs were done through an anterior incision with a medial parapatellar arthrotomy, and THAs were done through a direct anterior approach. Tranexamic acid was administered preoperatively for all patients if not absolutely contraindicated. For pain control, all patients received standard postoperative pain medication along with local anesthetic injected into soft tissue; peripheral nerve block injections were not regularly used. Intrathecal morphine was not used for any cases, and immediate physical therapy was available to all patients postoperatively. All patients had access to the same preoperative educational resources. Before being discharged, patients had to demonstrate basic capabilities such as walking 100 ft, voiding on own, and tolerating diet.
Patients older than 18 years with BMI less than 45 kg/m2 who had a preoperative RAPT score were included in our analysis. RAPT was administered to patients at one of the preoperative appointments and is available in the patient's electronic medical record. If the RAPT score was unavailable or was not collected, then the patient was excluded from the analysis. The discharge plan for the surgery (ambulatory SDD or not) was retrieved from the note documented by the surgeon at the preoperative visit. These plans are made by the surgeon after discussion with the patient and are made without consideration of the RAPT score. The LOS in nights, discharge destination, and presence of 90-day readmissions and reoperations were also collected.
The total cohort of patients was then divided into TKA and THA cohorts, and their baseline characteristics and discharge outcomes were compared. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were then calculated for several cutoff values to screen for SDD. A receiver operating characteristic (ROC) curve was generated, and the optimal cutoff value was chosen as the point where the Youden Index (sensitivity + specificity −1) was greatest. This was done separately for both the TKA and THA cohorts to generate a cutoff to be used within each cohort. Patients below and above the cutoff were compared to see whether there were any differences about discharge destination, LOS, readmission, and revision surgery rates. Patients were then sorted into SDD and non–same-day discharge (NSDD) cohorts, and logistic regression analysis was done to identify potential factors associated with same-day discharge. These analyses were all done individually for both the TKA and THA cohorts.
An analysis was then done to assess how the RAPT score compares with the discharge plan documented preoperatively by the surgeon for ability to predict the discharge outcome. Patients were predicted as either SDD or NSDD based on their RAPT score and the cutoffs identified in the ROC curve analysis. The predictive accuracy of RAPT was calculated by taking the number of times the prediction agreed with the discharge outcome and dividing by the total number of patients. A similar calculation was done to generate a predictive accuracy for the preoperatively documented discharge plan.
All statistical analyses were done using JMP (Version 16.0.0, SAS Institute). Data were assessed for normality by the Shapiro-Wilk test and reported as mean ± standard deviation or median (interquartile range) for normally and nonnormally distributed variables, respectively. The independent sample Student t-test and Wilcoxon rank-sum (Mann-Whitney U) test were used to evaluate continuous variables. Categorical variables were compared using the Pearson chi-squared test. P-values < 0.05 were considered statistically significant. Recursive partitioning analysis with classification trees was used to generate a ROC curve and choose a cutoff value that optimizes the sensitivity and specificity. For the logistic regression analysis, univariable analysis was first done, and variables with P-values < 0.20 were chosen to be included in the multivariable analysis.
ResultsA total of 647 patients underwent either TKA or THA between February 2020 and May 2021. Of these, 361 had preoperative RAPT scores recorded in our electronic medical record and thus were included in the analysis. Of note, 199 of these patients were from one surgeon, whereas 76 and 86 patients were from the two other surgeons, the discrepancy due to how consistently RAPT was collected by each surgeon. In total, 147 of 361 patients (42.6%) were discharged the same day as their surgery. The demographic information for the total, TKA, and THA cohorts is listed in Table 1. The TKA cohort was significantly older (P = 0.04) and had both a higher BMI (P < 0.001) and a higher Charlson Comorbidity Index (CCI) (P < 0.001) than the THA cohort.
Table 1 - Demographics and Summary Statistics Variable Total (N = 361) TKA (N = 159) THA (N = 202) P Age, years (range) 66 (58-73) 67 (60-73) 65 (55-72) 0.043 BMI, kg/m2 (range) 28.9 (25.2-32.9) 30.6 (28.1-34.1) 26.8 (23.6-31.0) <0.001 Charlson Comorbidity Index (CCI) 2 (2-4) 3 (2-4) 2 (1-3) <0.001 Male sex, % (n) 44.0% (159) 37.1% (59) 49.5% (100) 0.018 Spinal anesthesia, % (n) 80.1% (289) 79.3% (126) 80.7% (163) 0.733 Preop RAPT (range) 10 (8-11) 9 (8-10) 10 (8-11) <0.001 Discharge home, % (n) 96.1% (347) 96.2% (153) 96.0% (194) 0.927 % Same-day discharge (n) 40.7% (147) 37.1% (59) 43.6% (88) 0.215 90-day readmission rate (n) 3.9% (14) 5.0% (8) 3.0% (6) 0.314 90-day revision surgery rate (n) 1.9% (7) 2.5% (4) 1.5% (3) 0.481RAPT = Risk Assessment and Prediction Tool, TKA = total knee arthroplasty, THA = total hip arthroplasty
Bold represents statistically significant data.
Sensitivity, specificity, PPV, NPV, and the Youden Index across several cutoff values used to screen patients for SDD after TKA are summarized in Table 2. A cutoff of ≥9 had the highest Youden Index and was used to partition TKA patients into low and high RAPT score cohorts for comparison. TKA patients with a RAPT ≥ 9 had lower hospital LOS (P < 0.001), higher SDD rates (P < 0.001), and lower 90-day readmission rates (P = 0.03) compared with patients with RAPT<9 (Table 3). However, there were no differences in discharge to SAR or 90-day revision surgery rates for TKA patients with RAPT ≥9 and RAPT<9.
Table 2 - RAPT Performance Measures at Various Cutoff Levels for Predicting Same-day Discharge After TKA Performance Measure ≥7 ≥8 ≥9 ≥10 ≥11 ≥12 Sensitivity 0.983 0.898 0.847 0.661 0.373 0.220 Specificity 0.190 0.330 0.530 0.640 0.880 0.940 PPV 0.417 0.442 0.515 0.520 0.647 0.684 NPV 0.950 0.846 0.855 0.762 0.704 0.671 Youden Index 0.173 0.228 0.377 0.301 0.253 0.160RAPT = Risk Assessment and Prediction Tool, NPV = negative predictive value, PPV = positive predictive value, Youden Index = sensitivity+specificity −1
Bold represents statistically significant data.
RAPT = Risk Assessment and Prediction Tool, SDD = same-day discharge, SAR = subacute rehabilitation
Bold represents statistically significant data.
TKA patients were then sorted into SDD and NSDD cohorts, and univariable analysis found that lower age (P < 0.001), lower CCI (P < 0.001), spinal anesthesia (P = 0.03), and a RAPT score ≥9 (P < 0.001) were all associated with SDD (Table 4). In multivariable analysis, younger age (OR = 0.92; 95% CI 0.85 to 0.98), spinal anesthesia (OR = 3.25; 95% CI 1.17 to 9.08), and a RAPT score ≥9 (OR = 4.84; 95% CI 2.02 to 11.57) maintained their association with SDD.
Table 4 - Factors Associated With Same-day Discharge After TKA Variable SDD (N = 59) NSDD (N = 100) PBMI = body mass index, CCI = Charlson Comorbidity Index, RAPT = Risk Assessment and Prediction Tool, SDD = same-day discharge, NSDD = non–same-day discharge
Bold represents statistically significant data.
For THA patients, a cutoff of ≥11 was identified as having the highest Youden Index to screen patients for SDD (Table 5). THA patients were sorted into RAPT ≥11 and <11 cohorts, and patients with RAPT ≥11 had shorter hospital stays (P < 0.001), higher SDD rates (P < 0.001), and lower discharge to SAR (P < 0.001) compared with patients with RAPT<11 (Table 6). Univariable analysis found that lower age (P = 0.004), lower CCI (P = 0.003), spinal anesthesia (P = 0.003), and RAPT ≥ 11 (P < 0.001) were associated with SDD for THA patients (Table 7). In multivariable analysis, only a RAPT ≥ 11 (OR = 3.63; 95% CI 1.88 to 7.01) was found to be associated with SDD among THA patients.
Table 5 - RAPT Performance Measures at Various Cutoff Levels for Predicting Same-day Discharge After THA Performance Measure ≥7 ≥8 ≥9 ≥10 ≥11 ≥12 Sensitivity 0.977 0.955 0.932 0.818 0.659 0.250 Specificity 0.228 0.325 0.395 0.526 0.693 0.886 PPV 0.494 0.522 0.543 0.571 0.624 0.629 NPV 0.929 0.902 0.882 0.789 0.725 0.605 Youden Index 0.205 0.280 0.327 0.344 0.352 0.136PPV = positive predictive value, NPV = negative predictive value, Youden Index = sensitivity+specificity −1
Bold represents statistically significant data.
RAPT = Risk Assessment and Prediction Tool, SDD = same-day discharge, SAR = subacute rehabilitation
Bold represents statistically significant data.
BMI = body mass index, CCI = Charlson Comorbidity Index, NSDD = non–same-day discharge, RAPT = Risk Assessment and Prediction Tool, SDD = same-day discharge
Of the 361 patients with RAPT scores, 324 (90%) also had a discharge plan documented preoperatively by the surgeon. Using the previously identified thresholds (≥9 for TKA and ≥11 for THA), RAPT correctly predicted the discharge outcome for 216 of 324 patients (66.7%). By comparison, the plan found in the preoperative note correctly predicted discharge outcome for 296 of 324 patients (91.4%).
DiscussionThe goal of this study was to determine whether preoperative RAPT scores were associated with SDD in a cohort of primary TKA and THA patients at an urban academic hospital. We found that preoperative RAPT scores were associated with SDD. However, the cutoffs of ≥9 and ≥11 for predicting SDD had a predictive accuracy of only 66.7%, whereas the preoperative plan was more than 90% accurate. These findings suggest that although RAPT is associated with same-day discharge, it is not a useful tool to predict SDD for patients undergoing THA and TKA, supporting our initial hypothesis.
RAPT has been well studied over the years since its inception in 2003 as a preoperative screening tool by Australian researchers who demonstrated 75.2% accuracy for predicting whether patients would be discharged to inpatient rehabilitation facilities, with 89.2% accuracy specifically among high-risk patients.9 RAPT's predictiveness for discharge destination was validated in French patient populations by Dauty et al10 and Coudeyre et al.11 In the United States, Hansen et al12 found the RAPT to be predictive for the low-risk and high-risk patient cohorts and Dibra et al13 found that it accurately predicted discharge destination for patients undergoing revision TKA and THA. However, Cizmic et al14 found that among a cohort of extended LOS patients, defined as greater than 3 days, the RAPT was less accurate at predicting discharge disposition, especially for medium-risk patients. Another study conducted in the United States by Cohen et al15 found that a RAPT with modified cutoffs may be more accurate at predicting discharge disposition. Alshawani et al found RAPT to be somewhat predictive of LOS, but not of discharge destination among a patient population in the United Kingdom.16 Most recently, Oeding et al17 found the RAPT score to be associated with inpatient versus outpatient status after TKA and THA. RAPT's ability to predict discharge deserves additional investigation, especially as fewer patients are being discharged to rehab facilities than in the past.18 However, in recent years, there has been a rise in ambulatory surgery, and RAPT may be a potentially useful tool for predicting same-day discharge. Whether RAPT can reliably predict whether patients undergo SDD after surgery was not examined by any of these previous studies.
Our study identified a cutoff of ≥9 to be used for screening TKA patients, whereas a cutoff of ≥11 was identified for THA. This difference may be explained by the difference in these two cohorts' baseline characteristics. The TKA patients were significantly older, had a greater BMI, and had more medical comorbidities than those undergoing THA. The preoperative RAPT scores were higher for the THA cohort compared with TKA. Thus, a higher cutoff is necessary for the THA cohort to distinguish those likely to be SDD versus those at risk for NSDD, given that the THA cohort is, as a whole, healthier and has higher RAPT scores compared with the TKA cohort.
Although there is a positive association between RAPT score and SDD, RAPT correctly predicted discharge only 66.7% of the time and is less accurate than the preoperatively documented discharge plan which was 91.7% accurate. A closer inspection of data reveals that only a little over 60% of patients with a RAPT score of 12, the highest score possible, underwent SDD in both TKA and THA groups. It is also worth noting that RAPT has a better NPV than PPV, with 85.5% of TKA and 72.5% of THA patients below their cutoff not being discharged same day. Thus, it may be used as a NPV tool to predict NSDD but should not be used as a PPV tool to predict SDD. The overall predictive accuracy of RAPT is limited by the low PPV because many patients with RAPT above the cutoff fail to be discharged same day. Other factors may be involved, as suggested by our finding that the use of spinal versus general anesthesia was associated with SDD in the TKA cohort.
We would like to acknowledge that this study is not without limitations. As a retrospective study, there are inherent biases associated with data collection and the study design. Only 361 of 647 patients had RAPT scores collected, and many patients were missing scores possibly due to the effect of the COVID-19 pandemic and that only one of the three surgeons routinely collected RAPT scores. The surgical approach for THA and whether which surgeon performed the surgery had an effect on SDD was not assessed and is another potential limitation. Since this study was conducted at a single academic hospital in a major metropolitan city in the United States, the results may not necessarily be generalized to other healthcare settings or other geographies. Similarly, these data represent arthroplasty care delivered over a specific period and may not continue to be applicable as perioperative protocols continue to evolve. Along these same lines, the influence of the pandemic on a patient's willingness to consider SDD during this study period deserves additional study. In addition, since this study only included primary hip and knee joint arthroplasties, the findings may not be applicable to conversion or revision procedures. Finally, time of day of surgery affects SDD because later cases have shorter windows to meet discharge milestones, and this was not included as a variable.
The ability to plan postoperative care has become increasingly important to several stakeholders including patients, physicians, and healthcare institutions. In particular, patients who are not candidates for SDD may benefit from targeted initiatives including more extensive home support. The preoperatively collected RAPT score has initially shown promise at predicting whether patients would be discharged home or to a rehabilitation facility after undergoing hip and knee arthroplasty surgery. However, with the recent increase in same-day discharge, having a reliable predictor would be very valuable. Our findings suggest that the RAPT score is not a useful tool to predict whether a patient would be discharged home directly after undergoing primary hip and knee joint arthroplasty. Whether RAPT scores can be used to predict SDD after revision and other procedures may be the focus of future studies. In addition, more work should be done to assess RAPT's utility in other healthcare settings.
References 1. Gkagkalis G, Pereira LC, Fleury N, Luthi F, Lécureux E, Jolles BM: Are the Cumulated Ambulation Score and Risk Assessment and Prediction Tool useful for predicting discharge destination and length of stay following total knee arthroplasty? Eur J Phys Rehabil Med 2019;55:816-823. 2. Haghverdian BA, Wright DJ, Schwarzkopf R: Length of stay in skilled nursing facilities following total joint arthroplasty. J Arthroplasty 2017;32:367-374. 3. Bozic KJ, Ward L, Vail TP, Maze M: Bundled Payments in total joint arthroplasty: Targeting opportunities for quality improvement and cost reduction. Clin Orthop Relat Res 2014;472:188-193. 4. Molloy IB, Martin BI, Moschetti WE, Jevsevar DS: Effects of the length of stay on the cost of total knee and total hip arthroplasty from 2002 to 2013. J Bone Joint Surg Am Vol 2017;99:402-407. 5. Gwam CU, Mohamed NS, Dávila Castrodad IM, et al.: Factors associated with non-home discharge after total knee arthroplasty: Potential for cost savings? Knee 2020;27:1176-1181. 6. Clouette J, Agarwalla A, Ravi B, et al.: Same-day discharge following total joint arthroplasty: Examining trends, discharge dispositions, and complications over time. Orthopedics 2020;43:204-208. 7. Cherry A, Montgomery S, Brillantes J, et al.: Converting hip and knee arthroplasty cases to same-day surgery due to COVID-19. Bone Joint Open 2021;2:545-551. 8. Sconza C, Respizzi S, Grappiolo G, Monticone M: The risk assessment and prediction tool (RAPT) after hip and knee replacement: A systematic review. Joints 2019;07:041-045. 9. Oldmeadow LB, McBurney H, Robertson VJ: Predicting risk of extended inpatient rehabilitation after hip or knee arthroplasty. J Arthroplasty 2003;18:775-779. 10. Dauty M, Schmitt X, Menu P, Rousseau B, Dubois C: Using the Risk Assessment and Predictor Tool (RAPT) for patients after total knee replacement surgery. Ann Phys Rehabil Med 2012;55:4-15. 11. Coudeyre E, Eschalier B, Descamps S, et al.: Transcultural validation of the Risk Assessment and Predictor Tool (RAPT) to predict discharge outcomes after total hip replacement. Ann Phys Rehabil Med 2014;57:169-184. 12. Hansen VJ, Gromov K, Lebrun LM, Rubash HE, Malchau H, Freiberg AA: Does the risk assessment and prediction tool predict discharge disposition after joint replacement? Clin Orthop Relat Res 2015;473:597-601. 13. Dibra FF, Parvataneni HK, Gray CF, Vasilopoulos T, Prieto HA: The risk assessment and prediction tool accurately predicts discharge destination after revision hip and knee arthroplasty. J Arthroplasty 2020;35:2972-2976. 14. Cizmic Z, Feng JE, Anoushiravani AA, Borzio RW, Schwarzkopf R, Slover JD: The risk assessment and prediction tool is less accurate in extended length of stay patients following total joint arthroplasty. J Arthroplasty 2019;34:418-421. 15. Cohen E, Reid DBC, Quinn M, et al.: Modifying the RAPT score to reflect discharge destination in current Practice. Arthroplasty Today 2021;7:17-21. 16. Alshahwani AA, Dungey M, Lillie C, Krikler S, Plakogiannis C: Predictive value of the risk assessment and prediction tool (RAPT) score for primary hip and knee arthroplasty patients: A single-center study. Cureus 2021;13:e14112. 17. Oeding JF, Bosco JA, Carmody M, Lajam CM: RAPT scores predict inpatient versus outpatient status and readmission rates after IPO changes for total joint arthroplasty: An analysis of 12,348 cases. J Arthroplasty 2022;37:2140-2148. 18. Tarity TD, Swall MM: Current trends in discharge disposition and post-discharge care after total joint arthroplasty. Curr Rev Musculoskelet Med 2017;10:397-403.
留言 (0)