Abstract
Introduction:
Predicting the length of stay (LOS) after total joint arthroplasty (TJA) has become more important with their recent removal from inpatientonly designation. The American College of Surgeons (ACS) NSQIP Surgical Risk Calculator and the CMS’ Diagnosis Related Groups (DRGs) calculator are two common LOS predictors. The aim of our study was to determine how our actual LOS compared to those predicted by both the ACS and DRG.
Methods:
99 consecutive TJA (49 hips and 50 knee procedures) were reviewed in Medicare-eligible patients from four fellowship-trained arthroplasty surgeons. Predicted LOS was calculated using the DRG and ACS risk calculators for each patient using demographics, medical histories and comorbidities. LOS was compared between the predicted and the actual LOS for both total hip and total knee arthroplasty (THA, TKA) using paired t-tests.
Results:
Actual LOS was shorter in the THA group versus the TKA group (1.29 days vs 1.46 days, p < 0.05). The actual LOS of patients at our institution was significantly shorter than both DRG and ACS predictions for both THA and TKA (p<0.05). In both the THA and TKA patients, the actual LOS (1.29 and 1.46 day) was significantly shorter than the DRG-predicted LOS (2.15 and 2.15 days) which was significantly shorter than the ACS-predicted LOS (2.9 and 3.14 days).
Discussion:
We found the actual LOS was significantly shorter than that predicted by both the DRG and ACS risk calculators. Current risk calculators may not be accurate for contemporary fast-track protocols and newer tools should be developed.
Introduction:
Predictive analytics have become increasingly important in our current healthcare economy. There are multiple tools which use large amounts of demographic data to attempt to predict certain events, such as post-operative complications, length of stay and discharge disposition. Length of stay (LOS), defined as the start time for surgery to the time of patient discharge from the hospital, is an important metric for hospitals and surgeons to monitor in terms of costs, outcomes, and supply chain management. The average patient cost of a day in the hospital has been reported between $2,000 and $3,000.1 Hospital costs can be better managed if LOS is more accurately predicted, which has been further complicated by recent inpatient vs. outpatient designations2. More recently, the coronavirus has also led to an increase in outpatient procedures as hospitals may be too full to admit patients overnight3.
Both the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) Surgical Risk Calculator and CMS’ Diagnosis Related Groups (DRGs )are used to predict hospital LOS for various orthopedic and non-orthopedic surgical procedures. Our primary aim was to determine how accurately the LOS predictions produced by these calculators was for patients treated at our joint replacement center. Secondarily, we also sought to compare the ACS outlier predictions to the patients with LOS >1 standard deviation (SD) above our mean to determine its accuracy and utility. We hypothesized that the LOS predictions made by the current ACS and DRG calculators do not accurately predict the true LOS for THAs and TKAs.
Methods:
Institutional review board approval was obtained for this retrospective study prior to data collection. The study population was identified using the Current Procedural Terminology (CPT) codes corresponding to primary TKA (CPT 27447) and THA (CPT 27130). We compiled the 25 most recent consecutive TKAs and THAs performed by each of 4 fellowship trained arthroplasty surgeons since January 2020 at our single, tertiary academic institution. This resulted in a collection of 200 primary arthroplasty patients (100 TKAs and 100 THAs). From this initial listing, patients under the age of 65 were removed so that the remaining study group included those who were of Medicare eligibility. This resulted in a final study cohort of 99 total subjects (49 THAs and 50 TKAs). Surgeon A had 15 THAs and 13 TKAs, surgeon B had 10 THAs and 9 TKAs, surgeon C had 13 THAs and 13 TKAs, and surgeon D had 11 THAs and 15 TKAs included. Before surgery, a mandatory education class is required to be completed by the patient, and the surgery will be canceled if the patients do not attend this preoperative class4. Surgical approach for hips (anterior, anterolateral, and posterior) was based solely on surgeon preference. While the only exclusion of TKAs and THAs was patients being under 65, our group’s policy was to avoid surgery on patients with BMI > 40 kg/m2, HbA1C > 8, or those who are chronic high-dose narcotics as outlined in a paper by Edwards et. al.5 All 99 patients were treated with the same rapid recovery pathways utilizing general anesthesia and early mobilization protocols.6 These included the use of general anesthesia without supplemental regional blocks, scheduled anti-emetics, no postoperative urinary retention medication, as well physical therapy on the day of surgery for early joint mobilization. It is customary that patients get out of bed with the physical therapy team within 4 hours of arriving on the orthopedic floor after leaving the post-anesthesia care unit (PACU). Tranexamic acid was used in all 99 cases. Tourniquets and cemented fixation were used in all TKAs while all THAs were cementless.
The actual LOS for each patient was calculated by reviewing the electronic medical record. Patient’s diagnosis related groups were confirmed in the chart, and predicted LOS for THA or TKA had already been calculated through that system’s established methodology. According to the CMS website, a DRG average LOS is calculated in two ways. First, an arithmetic mean LOS (AMLOS) is calculated. This mean was determined by summing all of the patients LOS data available within a given DRG for the previous year and dividing by the total number of patients. Second, a geometric mean LOS (GMLOS) was calculated by taking all of the patients LOS data available within a given DRG for the previous year, multiplying each of these numbers together, and taking the root of the product of that multiplication problem to the nth power (n= number of patients). For example, in 2019, the AMLOS for DRG 470 (major hip and knee replacement) was 2.5 days, while the GMLOS was 2.2 days. In 2020, the AMLOS for DRG 470 was 2.4 days while the GMLOS was 2.1 days. For our study, we chose to utilize the GMLOS methodology because of its propensity to be less affected by outliers.
Patient demographic and comorbidity variables were collected from the medical record to calculate the ACS NSQIP prediction on LOS (Table 1). The publicly-available risk calculator was then used to determine the expected LOS (https://riskcalculator.facs.org/RiskCalculator/index.jsp). The ACS calculator was built by using data from over 5 million surgeries from 2015 to 2019. The data was gathered from 855 hospitals that participate in ACS NSQIP. All surgeries in this study were elective and labeled as “non-emergent”. American Society for Anesthesiologists Physical Status Classification System score (ASA) was recorded from the anesthesiologist’s surgery note on the day of surgery. Diabetes was only reported if patient was on oral medication or insulin, as per the ACS calculator guidelines. Each patient was given an individual LOS prediction and this was recorded as his or her“ACS Prediction”. Outliers were predicted by the ACS and DRG scoring system. Actual LOS outliers were defined as those with a LOS > 1 SD above the mean.
Table 1:
Demographic data for the cohort used in the ACS risk calculator.
| DEMOGRAPHICS | THA | TKA | Total |
|---|---|---|---|
| Number of Patients | 49 | 50 | 99 |
| Age Group | |||
| 65–74 | 30 (61%) | 31 (62%) | 61 (62%) |
| 75–84 | 17 (35%) | 14 (28%) | 31 (31%) |
| 85 years or older | 2 (4%) | 5 (10%) | 7 (7%) |
| Mean Age | 73.23 (SD: 5.98) | 73.72 (SD: 6.77) | 73.48 (SD: 6.37) |
| Sex | |||
| Female | 24 (49%) | 29 (58%) | 53 (54%) |
| Male | 25 (51%) | 21(42%) | 46 (46%) |
| Functional Status | |||
| Independent | 42 (86%) | 44 (88%) | 86(87%) |
| Partially Dependent | 5 (10%) | 5 (10%) | 10 (10%) |
| Totally Dependent | 2 (4%) | 1 (2%) | 3 (3%) |
| ASA Class | |||
| 1 | 0 (0%) | 1 (2%) | 1 (1%) |
| 2 | 18 (37%) | 16 (32%) | 34 (34%) |
| 3 | 28 (57%) | 31 (62%) | 59 (60%) |
| 4 | 3 (6%) | 2 (4%) | 5 (5%) |
| Mean ASA Class | 2.69 (SD: 0.58) | 2.68 (SD: 0.59) | 2.69 (SD: 0.58) |
| Steroid Use for Chronic Condition | 2 (4%) | 7 (14%) | 9 (9%) |
| Ascites within 30 days of surgery | 0 (0%) | 0 (0%) | 0 (0%) |
| Sepsis within 48 hours of surgery | 0 (0%) | 0 (0%) | 0 (0%) |
| Ventilator Dependent | 0 (0%) | 0 (0%) | 0 (0%) |
| Disseminated Cancer | 1 (2%) | 1 (2%) | 2 (2%) |
| Diabetes | |||
| Oral | 9 (18%) | 8 (16%) | 17 (17%) |
| Insulin | 1 (2%) | 1 (2%) | 2 (2%) |
| Hypertension Medication | 34 (69%) | 32 (64%) | 66 (67%) |
| Congestive Heart Failure within 30 days prior to surgery | 0 (0%) | 0 (0%) | 0 (0%) |
| Dyspnea | |||
| With Exertion | 0 (0%) | 4 (8%) | 4 (4%) |
| At Rest | 3 (6%) | 1 (2%) | 4 (4%) |
| Current Smoker within 1 year | 6 (12%) | 2 (4%) | 8 (8%) |
| History of COPD | 3 (6%) | 3 (6%) | 6 (6%) |
| Dialysis | 0 (0%) | 0 (0%) | 0 (0%) |
| Acute Renal Failure | 0 (0%) | 1 (2%) | 1 (1%) |
| BMI (kg/m2) | |||
| Underweight (below 18.5) | 0 (0%) | 0 (0%) | 0 (0%) |
| Normal (18.5–24.9) | 6 (12%) | 6 (12%) | 12 (12%) |
| Overweight (25–29.9) | 16 (33%) | 20 (40%) | 36 (36%) |
| Obese (30.0–40.0) | 27 (55%) | 24 (48%) | 51 (52%) |
| Mean BMI (kg/m2) | 30.77 (SD: 5.32) | 30.85 (SD: 5.65) | 30.81 (SD: 5.46) |
Statistical analysis was performed using Excel (Microsoft, Redmond, WA). LOS data was compared using six paired sample t-tests: actual vs. DRG, actual vs. ACS, and DRG vs. ACS for both THA and TKA. Three two-sample t-tests were performed to compare THA actual, DRG, and ACS values with TKA actual LOS, DRG, and ACS values. Significance was set at p < 0.05. We compared the number of outliers on the ACS calculator to the patients with >1 SD LOS above the mean from our study cohort.
Results:
In both THA and TKA, our results were very similar as far as relationships between actual LOS, ACS-predicted LOS, and DRG-predicted LOS. The actual LOS for our cohort of patients was significantly less than that of both the DRG-predicted and ACS-predicted LOS for both THA and TKA (p <0.001, Table 2). We also determined that the DRG-predicted LOS was significantly lower than the ACS-predicted LOS in both THA and TKA (p <0.001, Table 2). In both THA and TKA, the actual LOS was the shortest, while the ACS-predicted LOS was the longest (Table 2).
Table 2:
Comparing Actual Length of Stay (LOS) in days, DRG Predicted LOS, and ACS Predicted LOS
| THA | Actual LOS | DRG Predicted LOS | ACS Predicted LOS |
| Average: | 1.29 | 2.15 | 2.90 |
| Geometric Mean: | 1.27 | 2.15 | 2.82 |
| Upper: | 3.22 | 2.20 | 5.00 |
| Lower: | 1.05 | 2.10 | 2.00 |
| Range: | 2.17 | 0.10 | 3.00 |
| ACS Predicted LOS vs. Actual LOS | DRG Predicted LOS vs. Actual LOS | ACS Predicted LOS vs. DRG Predicted LOS | |
| Average difference | 1.61 | 0.87 | 0.74 |
| P-Value | p<0.0001 | p<0.0001 | p<0.0001 |
| TKA | Actual LOS | DRG Predicted LOS | ACS Predicted LOS |
| Average: | 1.46 | 2.15 | 3.14 |
| Geometric Mean: | 1.40 | 2.15 | 3.07 |
| Upper: | 3.25 | 3.40 | 5.50 |
| Lower: | 1.05 | 2.10 | 2.00 |
| Range: | 2.20 | 1.30 | 3.50 |
| ACS Predicted LOS vs. Actual LOS | DRG Predicted LOS vs. Actual LOS | ACS Predicted LOS vs. DRG Predicted LOS | |
| Average difference | 1.68 | 0.69 | 0.99 |
| P-Value | p<0.001 | p<0.001 | p<0.001 |
The actual LOS for TKA patients was significantly longer than that of the THA patients (p = 0.049, Table 3), with no significant differences in demographics. The DRG-predicted LOS was identical in both THA and TKA. While the ACS-predicted LOS did calculate a slightly longer LOS for TKA than THA, this difference did not meet statistical significance (Table 3).
Table 3:
Comparing Actual Length of Stay (LOS, DRG Predicted LOS, and ACS Predicted LOS of both THA and TKA.
| TKA(In Days) | THA(In Days) | THA vs TKA P-Value | |
|---|---|---|---|
| Average: | |||
| Actual LOS | 1.46 (SD 0.54) | 1.29 (SD 0.30) | 0.049 |
| DRG Predicted LOS | 2.15 (SD 0.19) | 2.15 (SD 0.05) | 0.973 |
| ACS Predicted LOS | 3.14 (SD 0.69) | 2.90 (SD 0.71) | 0.090 |
| Geometric Mean: | |||
| Actual LOS | 1.40 | 1.27 | |
| DRG Predicted LOS | 2.15 | 2.15 | |
| ACS Predicted LOS | 3.07 | 2.82 | |
| Maximum: | |||
| Actual LOS | 3.25 | 3.22 | |
| DRG Predicted LOS | 3.40 | 2.20 | |
| ACS Predicted LOS | 5.50 | 5.00 | |
| Minimum: | |||
| Actual LOS | 1.05 | 1.05 | |
| DRG Predicted LOS | 2.10 | 2.10 | |
| ACS Predicted LOS | 2.00 | 2.00 | |
| Range: | |||
| Actual LOS | 2.20 | 2.17 | |
| DRG Predicted LOS | 1.30 | 0.10 | |
| ACS Predicted LOS | 3.50 | 3.00 | |
| Average Difference: | |||
| ACS Predicted LOS vs. Actual LOS | 1.68 | 1.61 | |
| DRG Predicted LOS vs. Actual LOS | 0.69 | 0.87 | |
| ACS Predicted LOS vs. DRG Predicted LOS | 0.99 | 0.74 |
There were 9 (9%) outliers in actual LOS, 13 (13%) outliers in the ACS prediction, and 0 outliers in DRG predictions from the 99 total patients reviewed (Table 4). Of the 13 predicted outliers by ACS, 7 (54%) of the subjects correlated with being an actual LOS outlier. Seven of the 9 (78%) actual LOS outliers were correctly identified as ACS prediction outliers.
Table 4:
Actual LOS outliers and the comorbidities/demographics associated with these outliers.
| OUTLIERS | Actual LOS | ACS Predicted | ||||
|---|---|---|---|---|---|---|
| TKA(7 Total) | THA(2 Total) | Total (9 total) | TKA(7 total) | THA(6 total) | Total(13 total) | |
| ASA Class >/= 3 | 7 (100%) | 2 (100%) | 9 (100%) | 7 (100%) | 6 (100%) | 13 (100% |
| On Hypertension Medication | 4 (57%%) | 2 (100%) | 6 (67%) | 6 (86%) | 5 (83%) | 11 (85%) |
| Female Sex | 5 (71%) | 1 (50%) | 6 (67%) | 4 (57%) | 4 (67%) | 8 (62%) |
| Partially Supported At Home | 4 (57%) | 1 (50%) | 5 (56%) | 5 (71%) | 4 (67%) | 9 (69%) |
| History of Falls | 1 (14%) | 1 (50%) | 2 (22%) | 1 (14%) | 2 (33%) | 3 (23%) |
| Overweight BMI | 5 (71%) | 1 (50%) | 6 (67%) | 4 (57%) | 5 (83%) | 9 (69%) |
| Dyspnea On Exertion | 3 (43%) | 0 (0%) | 3 (33%) | 3 (43%) | 1 (17%) | 4 (31%) |
| Oral Diabetes Medication | 2 (29%) | 0 (0%) | 2 (22%) | 3 (43%) | 1 (17%) | 4 (31%) |
Discussion:
Improvements in rapid recovery protocols after joint arthroplasty over the last decade have led to reductions in hospital LOS across the country as well as at our institution6,7. The ability of commonly used LOS prediction tools to adapt to these changes is not known. In this retrospective review, we found that both the DRG and ACS risk calculators predict longer LOS after both THA and TKA than what occurs at our academic medical center that employs a modern rapid recovery protocol. We determined that the actual LOS for both THA and TKA patients is significantly less than the predictions made by both DRG and the ACS risk calculator by almost a full day. We also noted that subjects undergoing a TKA (1.46 days) spent a significantly longer time in the hospital than did those who underwent THAs (1.29 days, p=0.049).
The ACS-predicted LOS has been used with varying success when applied to in other surgical fields. Walker et al.8 showed that the actual observed LOS of adrenalectomies at a single institution was 1.6 days, which was significantly shorter than the ACS-predicted LOS of 2.1 days. This mirrors our findings both in that the ACS-predicted longer LOS than the actual LOS and also in the magnitude of overestimation of LOS by the ACS prediction. Jiang et al.9 found that the ACS calculator accurately predicted their observed LOS in Whipple procedures performed at a tertiary health center. On the other hand, Cusworth et al.10 found that patients undergoing a Whipple procedure at a single institution had a significantly longer observed LOS than what the ACS-predicted. Reviews of three single institutional studies, all using the NSQIP ACS LOS predictor found that one showed a significantly shorter observed LOS8, one showed no significant difference between ACS predictions and observed LOS9, and one showed that they observed a longer observed LOS than what the ACS-predicted.10 It is possible that the ACS calculator predicts LOS correctly when compared to all procedures throughout all institutions, but when compared to single institutions, it performs worse at predicting LOS correctly. This could be due to individual differences in surgeon treatment methods, as well as differences inherent to hospital rules, regulations, and recovery strategies.
Although the actual LOS was significantly shorter than both LOS prediction tools, we did discover that certain patient factors and comorbidities were possibly associated with increased actual LOS. These attributes included ASA score, hypertension requiring medication, female sex, use of a mobility aid before surgery, increased BMI, and not being fully independent preoperatively. There is also evidence in multiple previous publications that these factors have been found to independently increase LOS times, but it was more common that an increase LOS was associated with combinations of these factors. A study performed by Piuzzi et al11 found that increased BMI and female gender were both correlated with longer LOS in a sample of 4,509 TKA patients. Another study, looking at TKA patients, found that preoperative reliance on a caregiver in the home was significantly correlated to an increased LOS and that BMI and female sex were not significantly tied to a longer LOS12. Both Piuzzi and Halawi found that an increase in comorbidities led to an increase actual LOS. Lastly, Richards et al13 found multiple factors associated with an increased LOS when examining patients following a hip fracture. They noted that poor mobility status before surgery predicted a longer hospital stay, which was consistent withour findings regarding longer LOS and the use of mobility aids and being dependent on care givers. Secondly, they found that patients with ASA scores of 3 or 4 had significantly longer LOS. While this list of factors is not exhaustive, it can help counsel patients and guide providers and payers on specific functional and baseline health considerations that may adverselyaffect LOS. From our observations, and the above publications findings, it may be important to appropriately counsel patients of an increased LOS risk if they have a higher ASA score, have hypertension requiring medication, are a female, use a gait aid before surgery, have an increased BMI, or are not fully independent preoperatively.
Many authors have attempted to come up with better calculators to predict LOS after joint replacement14. Poitras et al performed a study to determine the accuracy of using ASA, Charlson comorbidity index (CCI), and the Time-Up-And- Go (TUG) performance measure to predict LOS in TJA patients. They found that both TUG (odds ratio (OR) 2.18) and ASA (OR 3.57) could be used to help predict LOS in these patients15. Secondly, a study by Kim et al attempted to find if the same CCI and the Elixhauser Comorbidity Measure (ECM) could be used to predict LOS on patients having shoulder arthroplasty. What was found is that both the CCI (C-Statistic= 0.698) and the ECM (C-Statistic= 0.752) could predict extended LOS, but the ECM predicted it slightly better. While both of these showed correlations with certain comorbidities with increased or extended LOS, neither was able to give an actual number of days16. Future work is needed to see if other factors are needed to develop more accurate calculators.
The decrease in LOS over time may have affected the ability of the ACS and DRG calculators to accurately predict LOS. LOS has decreased from an average of 3.7 days from 2006 to 2009, to an average of 3.3 days from 2010 to 2013, and finally down to an average of 3.0 days from 2014 to 201617. Other authors found that in 2013, their hospital’s average LOS for TJA patients was 2.66 and in 2017, their average LOS was 1.6318. In 2019, outpatient knee replacement became accepted by CMS and the same is occurring with THA, which will further decrease average LOS across institutions 19. Further work is required to develop predictive models to determine the ability of patients to safely undergo same day total joint replacement20.
Although certain patient-related factors have been associated with increased LOS in THA and TKA, it is plausible that procedure-related factors could explain this increase as well. A recent study21 examined 5,281 THA patients and multiple procedure-related factors, including as hospital site, surgical approach, day of surgery, and surgery start time. They discovered that hospital site was significantly associated (p<0.001) with a >1-day LOS. They also found that undergoing TJA on a Friday was almost twice as likely (OR=1.81) as Monday to have a >1-day LOS and that a PM surgery was almost twice as likely (OR=1.8) as AM surgery to have a >1-day LOS. Perhaps the most glaring prediction for a >1-day LOS was the use of non-direct anterior hip arthroplasty. Anterolateral, direct lateral and posterior based THA approaches were much more likely to have a >1-day LOS (OR=9.13, 7.59, and 5, respectively). According to these findings, it’s possible that newer LOS predictors should take into account procedure-related factors as well as patient-related factors to improve model precision.
While our study approach is unique, there are potential weaknesses to note. First, we were reliant on demographic and functional status information entered correctly into our electronic medical record, and there were instances that required extrapolation. There are some values that were put into the ACS Risk Calculator that involved potential observer effects, as certain elements were not directly stated in the medical record and instead inferred from reviewing the patient’s encounters around the time of surgery. These elements were most commonly patient dependence status and recent history of falls. Secondly, with only 49 TKA patients and 50 THA patients, it is possible that a larger sample size could have led to a truer representation of our patient population as a whole. We do not necessarily think this is true since all but one THA and 4 TKA patients experienced a shorter actual LOS than what either of the two prediction tools calculated. Further prospective study designs would better account for appropriately powered subjects. It is also important to note that all patients considered in this study were at least 65 years of age, which may limit the broad generalizations when applying such LOS predictive tools to younger patients undergoing joint arthroplasty.
The accurate prediction of LOS is a keymetric surrounding total joint arthroplasty due to its importance in resource allocation, patient cost, and hospital cost. LOS is a very important data point for hospital administrators and is even more important when predicted LOS are accurate. This allows hospitals near capacity to allocate appropriate resources, including beds and nursing staff. High through put services with short LOS require higher nursing ratios because of the increased work burden of admitting and discharging patients. We have previously shown that a predictable LOS can also help appropriate manage LOS fluctuations by day of the week when looking at TJA22. As many primary total joints move to the outpatient setting, whether HOPD (hospital outpatient department) or ASC (ambulator surgery center), with or without overnight or longer observation capabilities, the ability to predict this LOS becomes vital to many – including the surgeons who may own the ASC. Importantly, better predictors of who should stay overnight in a hospital setting are very much needed moving forward. The recent pandemic has further shown the need to predict resource allocation. Many total joint surgeons around the country were impacted greatly and restricted from operating due to hospital resource problems with limited available beds, nursing staff, and supplies. Accurate predictors for LOS would have been certainly been very helpful in this setting. The results of our study demonstrate that the current LOS achievements at our institution after THA and TKA consistently beat the most widely used LOS prediction models by one full day. Our ability to reproduce these shorter LOS proves that modern rapid recovery protocols are achieving their goal, as well as underscores the need to refine modern prediction tools to more accurately anticipate the costs and resource allocation surrounding arthroplasty surgery.
Supplementary Material
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Weiss A, &Elixhauser A (2014). Overview of Hospital Stays in the United States, 2012. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs, 425. [Google Scholar]
- 2.Krueger CA, Kerr JM, Bolognesi MP, Courtney PM, & Huddleston JI (2020). The Removal of Total Hip and Total Knee Arthroplasty from the Inpatient Only List Increases the Administrative Burden of Surgeons and Continues to Cause Confusion. The Journal of Arthroplasty, 0(0). 10.1016/j.arth.2020.04.079 [DOI] [PubMed] [Google Scholar]
- 3.Brown TS, Bedard NA, Rojas EO, Anthony CA, Schwarzkopf R, Barnes CL, Stambough JB, Mears SC, Edwards PK, Nandi S, Prieto HA, Parvizi J; AAHKS Research Committee. The Effect of the COVID-19 Pandemic on Electively Scheduled Hip and Knee Arthroplasty Patients in the United States. J Arthroplasty. 2020July;35(7S):S49–S55. doi: 10.1016/j.arth.2020.04.052. Epub 2020 Apr 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Edwards PK, Mears SC, Lowry Barnes C. Preoperative Education for Hip and Knee Replacement: Never Stop Learning. Curr Rev Musculoskelet Med. 2017September;10(3):356–364. doi: 10.1007/s12178-017-9417-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Edwards PK, Mears SC, Stambough JB, Foster SE, Barnes CL. Choices, Compromises, and Controversies in Total Knee and Total Hip Arthroplasty Modifiable Risk Factors: What You Need to Know. J Arthroplasty. 2018October;33(10):3101–3106. doi: 10.1016/j.arth.2018.02.066. Epub 2018 Feb 23. [DOI] [PubMed] [Google Scholar]
- 6.Stambough JB, Bloom GB, Edwards PK, Mehaffey GR, Barnes CL, Mears SC. Rapid Recovery After Total Joint Arthroplasty Using General Anesthesia. J Arthroplasty. 2019September;34(9):1889–1896. doi: 10.1016/j.arth.2019.04.066. Epub 2019 May 9. [DOI] [PubMed] [Google Scholar]
- 7.Stambough JB, Nunley RM, Curry MC, Steger-May K, Clohisy JC. Rapid recovery protocols for primary total hip arthroplasty can safely reduce length of stay without increasing readmissions. J Arthroplasty. 2015;30(4):521–526. doi: 10.1016/j.arth.2015.01.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Walker JB, Gogoj A, Saunders BD, Canter DJ, Lehman K, Raman JD. Accuracy of the NSQIP risk calculator for predicting complications following adrenalectomy. Int Urol Nephrol. 2019August;51(8):1291–1295. doi: 10.1007/s11255-019-02187-1. Epub 2019 Jun 10. [DOI] [PubMed] [Google Scholar]
- 9.Jiang HY, Kohtakangas EL, Asai K, Shum JB. Predictive Power of the NSQIP Risk Calculator for Early Post-Operative Outcomes After Whipple: Experience from a Regional Center in Northern Ontario. J Gastrointest Cancer. 2018September;49(3):288–294. doi: 10.1007/s12029-017-9949-2. [DOI] [PubMed] [Google Scholar]
- 10.Cusworth BM, Krasnick BA, Nywening TM, Woolsey CA, Fields RC, Doyle MM, Liu J, Hawkins WG. Whipple-specific complications result in prolonged length of stay not accounted for in ACS-NSQIP Surgical Risk Calculator. HPB (Oxford) 2017February;19(2):147–153. doi: 10.1016/j.hpb.2016.10.015. Epub 2016 Dec 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Piuzzi NS, Strnad GJ, SakrEsa WA, et al. The Main Predictors of Length of Stay After Total Knee Arthroplasty: Patient-Related or Procedure-Related Risk Factors. J Bone Joint Surg Am. 2019;101(12):1093–1101. doi: 10.2106/JBJS.18.00758 [DOI] [PubMed] [Google Scholar]
- 12.Halawi MJ, Vovos TJ, Green CL, Wellman SS, Attarian DE, Bolognesi MP. Preoperative predictors of extended hospital length of stay following total knee arthroplasty. J Arthroplasty. 2015;30(3):361–364. doi: 10.1016/j.arth.2014.10.025 [DOI] [PubMed] [Google Scholar]
- 13.Richards T, Glendenning A, Benson D, Alexander S, Thati S. The independent patient factors that affect length of stay following hip fractures. Ann R Coll Surg Engl. 2018;100(7):556–562. doi: 10.1308/rcsann.2018.0068 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Howie CM, Mears SC, Barnes CL, Stambough JB. Readmission, Complication, and Disposition Calculators in Total Joint Arthroplasty: A Systemic Review. J Arthroplasty. 2020. Nov 3:S0883-5403(20)31151-7. doi: 10.1016/j.arth.2020.10.052. Epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Poitras S, Au K, Wood K, Dervin G, Beaulé PE. Predicting hospital length of stay and short-term function after hip or knee arthroplasty: are both performance and comorbidity measures useful?. Int Orthop. 2018;42(10):2295–2300. doi: 10.1007/s00264-018-3833-y [DOI] [PubMed] [Google Scholar]
- 16.Kim CY, Sivasundaram L, LaBelle MW, Trivedi NN, Liu RW, Gillespie RJ. Predicting adverse events, length of stay, and discharge disposition following shoulder arthroplasty: a comparison of the Elixhauser Comorbidity Measure and Charlson Comorbidity Index. J Shoulder Elbow Surg. 2018;27(10):1748–1755. doi: 10.1016/j.jse.2018.03.001 [DOI] [PubMed] [Google Scholar]
- 17.Sarpong NO, Boddapati V, Herndon CL, Shah RP, Cooper HJ, Geller JA. Trends in Length of Stay and 30-Day Complications After Total Knee Arthroplasty: An Analysis From 2006 to 2016. J Arthroplasty. 2019;34(8):1575–1580. doi: 10.1016/j.arth.2019.04.027 [DOI] [PubMed] [Google Scholar]
- 18.Stone AH, Dunn L, MacDonald JH, King PJ. Reducing Length of Stay Does Not Increase Emergency Room Visits or Readmissions in Patients Undergoing Primary Hip and Knee Arthroplasties. J Arthroplasty. 2018;33(8):2381–2386. doi: 10.1016/j.arth.2018.03.043 [DOI] [PubMed] [Google Scholar]
- 19.Scully RD, Kappa JE, Melvin JS. "Outpatient"-Same-calendar-day Discharge Hip and Knee Arthroplasty. J Am AcadOrthop Surg. 2020October15;28(20):e900–e909. doi: 10.5435/JAAOS-D-19-00778. [DOI] [PubMed] [Google Scholar]
- 20.Ziemba-Davis M, Caccavallo P, Meneghini RM. Outpatient Joint Arthroplasty-Patient Selection: Update on the Outpatient Arthroplasty Risk Assessment Score. J Arthroplasty. 2019July;34(7S):S40–S43. doi: 10.1016/j.arth.2019.01.007. Epub 2019 Jan 15. [DOI] [PubMed] [Google Scholar]
- 21.The Cleveland Clinic Arthroplasty Group, Understanding the Main Predictors of Lengthof Stay After Total Hip Arthroplasty: Patient-Related or Procedure-Related RiskFactors?,The Journal of Arthroplasty (2021), doi: 10.1016/j.arth.2020.11.029. [DOI] [PubMed] [Google Scholar]
- 22.Edwards PK, Hadden KB, Connelly JO, Barnes CL. Effect of Total Joint Arthroplasty Surgical Day of the Week on Length of Stay and Readmissions: A Clinical Pathway Approach. J Arthroplasty. 2016December;31(12):2726–2729. doi: 10.1016/j.arth.2016.05.057. Epub 2016 Jun 3. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.