eFigure. Selection of Patients
eTable 1. Patient and Hospital Characteristics at Index Hospitalization for Patients Undergoing Major Cancer Operations (Total Number of Patients = 126 104)
eTable 2. Thirty-Day and 90-Day VTE Events
eAppendix. International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) Procedure Codes for Procedures Included in the Study
Customize your JAMA Network experience by selecting one or more topics from the list below.
Mallick S, Aiken T, Varley P, et al. Readmissions From Venous Thromboembolism After Complex Cancer Surgery. JAMA Surg. 2022;157(4):312–320. doi:10.1001/jamasurg.2021.7126
What is the incidence and burden of readmissions for postoperative venous thromboembolism after complex cancer surgery?
In this cohort study of 126 104 patients from the Nationwide Readmissions Database who had venous thromboembolism, 30-, 90-, and 180-day venous thromboembolism–associated readmission rates were 0.6%, 1.1%, and 1.7%, respectively.
Venous thromboembolism–related readmissions after complex cancer surgery continue to increase well beyond 30 days after surgery; these readmissions vary by operation type and are associated with significant mortality and cost.
Venous thromboembolism (VTE) is a major cause of preventable morbidity and mortality after cancer surgery. Venous thromboembolism events that are significant enough to require hospital readmission are potentially life threatening, yet data regarding the frequency of these events beyond the 30-day postoperative period remain limited.
To determine the rates, outcomes, and predictive factors of readmissions owing to VTE up to 180 days after complex cancer operations, using a national data set.
Design, Setting, and Participants
A retrospective cohort study of the 2016 Nationwide Readmissions Database was performed to study adult patients readmitted with a primary VTE diagnosis. Data obtained from 197 510 visits for 126 104 patients were analyzed. This was a multicenter, population-based, nationally representative study of patients who underwent a complex cancer operation (defined as cystectomy, colectomy, esophagectomy, gastrectomy, liver/biliary resection, lung/bronchus resection, pancreatectomy, proctectomy, prostatectomy, or hysterectomy) from January 1 through September 30, 2016, for a corresponding cancer diagnosis.
Readmission with a primary diagnosis of VTE.
Main Outcomes and Measures
Proportion of 30-, 90-, and 180-day VTE readmissions after complex cancer surgery, factors associated with readmissions, and outcomes observed during readmission visit, including mortality, length of stay, hospital cost, and readmission to index vs nonindex hospital.
For the 126 104 patients included in the study, 30-, 90-, and 180-day VTE-associated readmission rates were 0.6% (767 patients), 1.1% (1331 patients), and 1.7% (1449 of 83 337 patients), respectively. A majority of patients were men (58.7%), and the mean age was 65 years (SD, 11.5 years). For the 1331 patients readmitted for VTE within 90 days, 456 initial readmissions (34.3%) were to a different hospital than the index surgery hospital, median length of stay was 5 days (IQR, 3-7 days), median cost was $8102 (IQR, $5311-$10 982), and 122 patients died (9.2%). Independent factors associated with readmission included type of operation, scores for severity and risk of mortality, age of 75 to 84 years (odds ratio [OR], 1.30; 95% CI, 1.02-1.78), female sex (OR, 1.23; 95% CI, 1.11-1.37), nonelective index admission (OR, 1.31; 95% CI, 1.03-1.68), higher number of comorbidities (OR, 1.30; 95% CI, 1.06-1.60), and experiencing a major postoperative complication during the index admission (OR, 2.08; 95% CI, 1.85-2.33).
Conclusions and Relevance
In this cohort study, VTE-related readmissions after complex cancer surgery continued to increase well beyond 30 days after surgery. Quality improvement efforts to decrease the burden of VTE in postoperative patients should measure and account for these late VTE-related readmissions.
Postoperative venous thromboembolism (VTE) is a leading cause of 30-day mortality after complex cancer surgery.1 Approximately 2% of patients undergoing cancer surgery develop deep vein thrombosis or pulmonary embolism, and up to 50% of immediate postoperative mortality is attributable to VTE.1,2 Most prior literature, including randomized trials on VTE prophylaxis and outcomes, used a 30-day postoperative period to assess the rates of VTE.1-4 Hence, there are limited data on the incidence of VTE after specific cancer operations beyond the 30-day postoperative period.5 Although the majority of complications occur within 30 days, this is an arbitrarily chosen cutoff. There are no data to suggest a sudden decline in VTE risk at the 30-day mark. In 1 analysis of postoperative VTE among 239 614 middle-aged British patients undergoing any general surgical operation, the risk of VTE remained elevated for up to 12 weeks after surgery, especially for patients undergoing cancer surgery.6 That study did not consider the risk conferred by specific cancer operations, which can vary significantly in their associated VTE risk.6
Venous thromboembolism events that require readmission are associated with significant cost and have a high risk of mortality.7,8 The 30-day readmission rate after complex cancer surgery varies from 10% to 15%, with approximately 2% to 4% of readmissions attributed to VTE.9-13 Because patients undergoing complex cancer surgery remain at extended risk for VTE, it is likely that readmissions owing to VTE also continue beyond 30 days. These severe VTE events are a potentially preventable delayed complication after complex cancer surgery. Understanding timing and patterns of VTE-related readmission is necessary to develop strategies to reduce the burden. The aim of this study was to assess the incidence of readmission owing to VTE up to 180 days after complex cancer surgery, to describe the associated mortality and cost burden, and to identify factors associated with VTE readmission.
In this cohort study, we performed a secondary analysis of the 2016 Healthcare Cost and Utilization Project Nationwide Readmissions Database,14 a nationally representative, publicly available, deidentified database on approximately 17.2 million discharges for all payers and uninsured patients in the United States. The database was constructed with discharge data from 27 geographically dispersed states and accounts for 57.7% of the total US resident population and 56.6% of all US hospitalizations. Each database entry is assigned a unique patient linkage number, allowing for patients to be tracked across multiple hospitals within the same state. Furthermore, it is designed to be flexible to allow the analyst to determine the relationship between multiple hospital admissions. The database can therefore suitably be used for readmissions analysis. Because this was a secondary analysis of an administrative data set with no identifiable data, the study was exempt from review by the institutional review boards at the University of Wisconsin–Madison, who waived the need for informed consent. The RECORD reporting checklist for observational studies was followed in our study.15
We selected all adult patients (≥18 years) undergoing complex cancer operations from January 1 to September 30, 2016. This time frame allowed for all patients to have a minimum 90-day follow-up period. Separate analyses were performed for patients who had surgery from January 1 to June 30, 2016, to allow for a minimum 180-day follow-up period. Complex cancer operations were defined as patients undergoing an esophagectomy, gastrectomy, colectomy, proctectomy, pancreatectomy, prostatectomy, hysterectomy, cystectomy, liver/biliary tree resection, or lung/bronchus resection for a corresponding cancer diagnosis. The International Statistical Classification of Diseases, Tenth Revision, Procedure Coding System (ICD-10-PCS) was used to identify patients undergoing selected operations (eAppendix in the Supplement). The International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) was used to determine primary diagnoses (eAppendix in the Supplement). Patients who had a prior VTE diagnosis or chronic VTE diagnosis were excluded from our study. The eFigure in the Supplement displays our selection process. Readmission proportions were calculated only for patients who survived to discharge. We also excluded patients who experienced a symptomatic VTE event during the initial hospital stay. The Nationwide Readmissions Database does not collect data on race and ethnicity.
We identified all readmissions to index and nonindex hospitals by using the linkage variable NRD_VisitLink. The Nationwide Readmissions Database provided date of each admission, length of stay during admission, and time between admission dates. These values were combined to calculate the number of days from discharge to readmission in accordance with the Healthcare Cost and Utilization Project database use guidelines. Cause of readmission was sought from ICD-10-CM diagnosis codes describing the primary diagnosis at readmission. We selected all patients with a primary diagnosis of acute VTE (eAppendix in the Supplement). We analyzed the proportion of 30-, 90-, and 180-day readmissions owing to a primary diagnosis of VTE.
Additionally, as a separate secondary analysis, we identified patients with a VTE event who did not require readmission by reviewing all secondary diagnosis codes for VTE. Patients with a diagnosis code or comorbid condition of VTE in the list of additional diagnosis codes were assumed to have had a VTE event sometime after discharge. Because VTE was not a primary diagnosis code for the admission, it was not considered the cause of the readmission. Because all these patients did not have VTE during the index admission, nor did they have a history of VTE, they should not have had an additional diagnosis code of VTE unless they had a VTE event at some point between discharge and readmission. Thus, the only explanation could be that they had a VTE event in the outpatient setting that did not require readmission.
Other variables included in the analysis were patient demographic characteristics, disease-specific diagnostic codes for each admission, severity measures, outcomes, and hospital-level characteristics. We included age as a categorical variable in deciles. Severity of illness and risk of mortality were assessed with the “all-patient-refined diagnosis related groups severity of illness” variable and “risk of mortality” variable, respectively. These 2 data elements were calculated by 3M with regression analysis and provided with the database for each patient at each admission to be used for risk adjustment. Comorbid conditions were identified with diagnosis codes from the Agency for Healthcare Research and Quality–generated Elixhauser Comorbidity Index. These variables were used to generate a modified comorbidity score for each patient, in which each patient was grouped into 1 of 4 categories (0, 1, 2, or >2 comorbid conditions). We omitted malignant tumor as part of the comorbidity score because all patients had such a diagnosis. We used ICD-10-CM codes to generate variables for major complications, which were defined as cardiac arrest, acute cerebrovascular accident, sepsis, septic shock, hemorrhage, pneumonia, adult respiratory distress syndrome, myocardial infarction, and kidney failure (eAppendix in the Supplement). We also used ICD-10-CM codes to track patients who received adjuvant chemotherapy at any point after their index surgery until readmission (eAppendix in the Supplement). Length of stay was included as a continuous variable. Hospital costs were generated by merging hospital charges with the cost-to-charge ratio files provided with the data set. We also analyzed outcomes of discharge disposition, length of stay, cost, major morbidity, and in-hospital mortality. Outcomes during readmission, including length of stay, mortality, hospital cost, multiple readmissions (readmitted more than once after primary admission for VTE), disposition at discharge, and index vs nonindex readmissions, were calculated for patients who were readmitted within 90 days with VTE as the cause of the readmission.
Descriptive analyses were performed, using proportions for categorial variables and means with SDs or medians with IQRs for continuous variables where appropriate. All 30-, 90-, and 180-day readmissions in which VTE was the cause of readmission were reported as proportions. The incidence of VTE readmission was plotted by days to readmission to analyze trends. Multivariable logistic regression was used to identify factors present on the primary admission that were associated with 90-day VTE readmission. Variables considered for the model included demographic characteristics, patient clinical characteristics, type of surgery, facility-level variables, and postoperative in-hospital major complications and length of stay. Mann-Whitney U test or χ2 test was used to test for significance, and all analyses used 2-sided P. All variables that were significant at P < .10 on univariate analysis and those that were a priori decided by the authors to be clinically relevant were selected to be included in the initial model. The final multivariable model was derived with a manual backward elimination process, in which variables were removed 1 at a time, checking for 10% difference in the pseudo-R2 values. The linktest was also used to check for any additional variables that were statistically significant. Finally, the F test was used to test for model fit. This method was used to arrive at the most parsimonious best-fit multivariate model. All analyses were performed with Stata, version 16 (StataCorp LLC).
We analyzed a total of 197 510 visits from 126 104 patients undergoing a complex cancer operation from January 1 to September 30, 2016. Patient and hospital characteristics are presented in eTable 1 in the Supplement. A total of 74 047 patients were men (58.7%), 52 057 were women (41.3%), mean age was 65 years (SD, 11.5 years), 58 048 were younger than 65 years (46.0%), and 12 603 (10.0%) had no other comorbid conditions at admission. For the primary admission, 22 580 patients (17.9%) experienced at least 1 major complication, median length of stay was 4 days (IQR, 2-8 days), and the overall in-hospital mortality was 1.15% (1452 patients).
Of the 124 644 patients who survived until discharge and did not have a discharge diagnosis of VTE, 23 140 (18.6%) were readmitted within 90 days for any cause. Venous thromboembolism was the primary cause for 5.8% of 90-day readmissions, which corresponded to 767 patients (0.6%) within 30 days of discharge and 1331 patients (1.1%) within 90 days. The proportions of VTE readmissions by surgery type are presented in Table 1. Venous thromboembolism requiring readmission at 90 days was most common after cystectomy, pancreatectomy, and esophagectomy. Patients who had any VTE event (necessitating readmission or not) are described in eTable 2 in the Supplement. Of 23 140 patients who were readmitted within 90 days for any cause, 1517 (6.6%) died during their readmission visit. Of the 1517 patients readmitted for any cause who subsequently died, 239 (15.8%) had VTE as a primary diagnosis. For our secondary analysis, 1536 (1.2%) 30-day VTE events and 2610 (2.1%) 90-day VTE events were diagnosed that did not lead to readmission. Of the 90-day VTE events, 1206 (1.0%) were due to pulmonary embolism and 1404 (1.1%) to deep vein thrombosis.
In the subgroup analysis examining 180-day VTE readmissions, 84 176 patients underwent a complex cancer operation between January 1, 2016, and June 30, 2016. Of these patients, 83 337 (99.0%) survived until discharge and 1449 (1.7%) were readmitted owing to VTE within 180 days of discharge. Readmissions at 180 days because of VTE are described in Table 2.
The variation in VTE readmission rates over time for each operation is depicted in the Figure. The highest rate of 180-day readmissions because of VTE was observed in patients undergoing cystectomy (104 of 3170; 3.3%), followed by those undergoing pancreatectomy (93 of 2958; 3.1%). The lowest rates of 180-day readmissions because of VTE were observed in patients undergoing prostatectomy (216 of 19 343; 1.1%), followed by those undergoing lung/bronchus resections (176 of 13 235; 1.3%). The readmission rates varied significantly over time among operations. There was a high rate of early readmissions and plateauing of the curve for patients undergoing esophagectomies, cystectomies, colectomies, and prostatectomies. However, for patients undergoing pancreatectomies, liver/biliary resections, gastrectomies, and proctectomies, there were several other inflection points beyond the 30-day period.
Of the 1331 patients readmitted owing to VTE within 90 days, 456 (34.3%) were readmitted to a hospital other than where the surgery was performed (nonindex hospital). The median length of stay was 5 days (IQR, 3-7 days), median cost was $8102 (IQR, $5311-$10 982), and the total cost of all VTE readmissions combined was $23.8 million. Of 1331 patients, a total of 122 (9.2%) died during their readmission visit, whereas 83 (6.2%) were readmitted multiple times. The outcomes of 90-day readmissions are presented in Table 3.
On multivariate analysis, independent factors associated with increased 90-day readmission for VTE included type of operation (cystectomy: OR, 2.57 [95% CI, 2.20-2.98]; esophagectomy: OR, 1.45 [95% CI, 1.10-1.87]; liver/biliary resection: OR, 1.38 [95% CI, 1.12-1.67]; lung/bronchus resection: OR, 0.75 [95% CI, 0.65-0.86]; pancreatectomy: OR, 1.88 [95% CI, 1.57-2.23]; prostatectomy: OR, 0.78 [95%, 0.69-0.88]; proctectomy: OR, 1.03 [95% CI, 0.87-1.21]; gastrectomy: OR, 1.20 [95% CI, 0.97-1.46]; hysterectomy: OR, 0.14 [95% CI, 0.10-0.19] compared with colon resection), scores for severity of illness (moderate: odds ratio [OR], 1.60 [95% CI, 1.42-1.82]; major: OR, 1.60 [95% CI, 1.42-1.82]; extreme: OR, 3.54 [95% CI, 2.98-4.19]), scores for risk of mortality (moderate: OR, 2.12 [95% CI, 1.91-2.34]; major: OR, 3.48 [95% CI, 3.12-3.88]; extreme: OR, 3.10 [95% CI, 2.64-3.62]), age of 75 to 84 years (OR, 1.30; 95% CI, 1.02-1.78) compared with age of 18 to 44 years, female sex (OR, 1.23; 95% CI, 1.11-1.37), nonelective index admission (OR, 1.31; 95% CI, 1.03-1.68), higher number of comorbidities (OR, 1.30; 95% CI, 1.06-1.60), and experiencing a major postoperative complication during the index admission (OR, 2.08; 95% CI, 1.85-2.33) (Table 4). A laparoscopic approach was associated with decreased 90-day readmission for VTE (OR, 0.66; 95% CI, 0.59-0.73). Receiving adjuvant chemotherapy was not associated with increased VTE readmissions. The model satisfied postestimation testing criteria, with a significant F test result at P < .001 and the linktest at P = .72 (R2 = 0.77).
We describe the 30-, 90-, and 180-day rates of readmission for postoperative VTE after complex cancer surgery in the United States. The risk of VTE-related readmission varied by operation and was not constant during the postoperative period. Some operation types demonstrated a progressive increase in VTE-related readmission during the 180-day postoperative period (such as pancreatectomy), whereas others plateaued by 60 days (such as prostatectomy). Venous thromboembolism was listed as the primary diagnosis in 1331 of 23 140 readmissions (5.8%) and had a high mortality and cost burden. We found that 239 of 1517 patients (15.8%) who died during their initial readmission had VTE as their primary diagnosis. Overall, these data demonstrate the significant burden of VTE-related readmission beyond the standard 30-day outcome reporting period.
We chose to focus on VTE-related readmission rather than the overall incidence of postoperative VTE because these events were thought to be associated with the highest resource burden and risk of mortality. Venous thromboembolism–related readmissions, on average, were associated with a 4-day median length of stay, $8123 median hospital cost, and 5.1% mortality rate. Although the high morbidity and mortality associated with postoperative VTE has been previously documented,1,16 our results demonstrate that these trends extend beyond the 30-day readmission period. The development of delayed VTE-related readmission is likely multifactorial. Cancer-related factors, such as advanced disease, likely contribute to the risk of VTE beyond 30 days.1 Surgical factors may also contribute, because return to baseline function for patients undergoing complex abdominal surgery may take 6 months or more.17,18 Quality improvement efforts to decrease the burden of postoperative VTE should measure and account for these late VTE–related readmissions and associated risk factors.
Postoperative VTE is a potentially preventable complication. Venous thromboembolism prophylaxis is recommended and routinely used in the immediate postoperative period.19-21 Additional approaches to reduce the risk of postoperative VTE include pneumatic compression stockings and early ambulation.22,23 Randomized data exist to support the routine use of postdischarge VTE chemoprophylaxis among patients undergoing cancer surgery.3,4,24 In accordance with these findings, the National Comprehensive Cancer Network recommends 4 weeks of outpatient VTE chemoprophylaxis for high-risk abdominal or pelvic operations.20 Risk factors identified by Agnelli et al1 include anesthesia time greater than 2 hours, advanced-stage disease, bed rest longer than 3 days, older than 60 years, and previous VTE. Similarly, the American Society of Clinical Oncology recommends 7 to 10 days of postoperative chemoprophylaxis for all patients, with select high-risk patients receiving longer courses.19 However, these recommendations are often not followed because of perceived bleeding risk and concerns regarding cost-effectiveness.25,26 Adherence to recommended VTE prophylaxis is not a tracked quality measure.
Although longer-term chemoprophylaxis is often not used owing to limited data supporting routine use and the high cost of extended VTE prophylaxis, the frequency and morbidity of late VTE–related readmissions in our study suggest that there may be value in identifying a subset of high-risk patients most likely to benefit from extended VTE chemoprophylaxis or screening ultrasonography.27-30 Validated VTE risk assessment tools such as the Caprini or Rogers score might assist in identifying such patients.31-33 Future cost-effectiveness studies would help clarify the benefit of these interventions in select patient populations. Variables such as age (>65 years), metastatic disease, operation type, body mass index (>25, calculated as weight in kilograms divided by height in meters squared), and albumin level (<3.0 g/dL) have been shown in previous studies to be associated with postoperative VTE after cancer operations.16 The cost and tolerability of long-term prophylaxis might be improved by direct-acting oral anticoagulants, which have been shown in a recent randomized clinical trial of 400 women undergoing surgery for gynecologic malignancies to provide efficacy similar to that of low-molecular-weight heparin for postoperative VTE prophylaxis.34 This trial also demonstrated superior patient satisfaction with direct-acting oral anticoagulants, although adherence to both therapies statistically remained the same.34
The 90-day VTE-related readmission rate was highest among patients undergoing cystectomy (107 of 4688; 2.3%) and pancreatectomy (76 of 4467; 1.7%). In accordance with 30-day data from the National Surgical Quality Improvement Program, these 2 operations have previously been identified as being among the highest risk for overall VTE, although few studies have assessed the rate of VTE-related readmission specifically.35,36 A study by Doiron et al37 identified an overall 90-day VTE rate of 5.4%, and another study by Schmid et al38 observed that 8.5% of 30-day readmissions after radical cystectomy were due to VTE. Despite the high rate of VTE among patients undergoing cystectomy and pancreatectomy, postdischarge VTE prophylaxis is often underused in this cohort. A recent survey of US hepatobiliary surgeons demonstrated that only 45% used postdischarge VTE chemoprophylaxis after pancreaticoduodenectomy.25 A similar survey of UK urologists demonstrated that only 67% used postdischarge VTE chemoprophylaxis after radical cystectomy.39 In addition to emphasizing the importance of postdischarge VTE prophylaxis in this population, our data suggest that extended monitoring for VTE, prolonged VTE prophylaxis beyond 30 days, or both may be indicated for patients at highest risk.
An analysis of 90- and 180-day readmission with VTE after cancer surgery was recently conducted by Jarvis et al.40 The authors found a rate of admission with VTE of 1.7% at 90 days and 2.3% within 180 days. There are important differences in methodology between our study and theirs. The primary outcomes of our study were postoperative VTE as a primary cause of readmission, whereas Jarvis et al40 reported VTE when listed as either a primary or secondary diagnosis. The latter approximates the overall rate of postoperative VTE, although it does not capture individuals treated entirely as outpatients, which is a significant proportion of patients. Hence, the rates described by Jarvis et al40 have ascertainment bias. In contrast, our analysis accurately captures instances of VTE that were severe enough to warrant readmission. We also reported the rate of deep vein thrombosis and pulmonary embolism separately and grouped our results by operation type rather than cancer type. Overall, this granularity is intended to capture the most severe and life-threatening complications of postoperative VTE and aids in the development of further studies and interventions.
There are important limitations to our analysis. This was a retrospective analysis of administrative data, and therefore errors in coding are possible. It is possible that some of these readmissions were primarily due to causes other than VTE. Conversely, we may have missed patients with VTE readmissions if the admission diagnosis was not coded appropriately. By including only patients with a primary admission diagnosis of VTE, we selected a subset of patients with the highest likelihood of VTE-related readmission. Although this method gave us the most precise subset of patients, it may have underestimated the true burden. The analysis does not account for patients who were readmitted in another state, those who experienced fatal pulmonary embolism outside of the hospital, or those who had a VTE event that was not serious enough to warrant readmission. However, these numbers are likely small. One strength of the Nationwide Readmissions Database is that it captures nonindex hospital readmissions, which are often missed by institutional data sets. Excluding patients with previous VTE and those who developed VTE during the index admission may have slightly underestimated the observed rate of VTE, which would also then underestimate the burden of VTE-related readmissions. These patients are at higher risk of postoperative VTE.31 Additionally, they may also have been more likely to be receiving therapeutic anticoagulation for a prolonged period after surgery. Including these patients in the analysis may have introduced a higher potential for labeling errors for patients readmitted because of VTE as opposed to other causes in patients with known VTE. It is impossible in the Nationwide Readmissions Database data set to segregate patients with new outpatient postoperative VTE events from those with a prior VTE diagnosis. A final limitation of our study is that the data used were from 2016; however, we believe that our findings remain relevant because postoperative standards of care have remained relatively stable during the past 5 years.
In this cohort study, VTE-related readmissions after complex cancer surgery continued to occur well beyond 30 days after surgery. Venous thromboembolism was a primary diagnosis in 15.8% of deaths during readmission after complex cancer surgery. Patients were frequently readmitted to nonindex hospitals and bore a significant mortality and cost burden. These results suggest that the burden of VTE after complex cancer surgery is underappreciated by registries focused on 30-day outcomes. Quality improvement efforts on the burden of VTE should ensure a longer follow-up period.
Accepted for Publication: November 7, 2021.
Published Online: January 26, 2022. doi:10.1001/jamasurg.2021.7126
Corresponding Author: Syed Nabeel Zafar, MD, MPH, Division of Surgical Oncology, Department of Surgery, University of Wisconsin School of Medicine and Public Health, 600 Highland Ave, Madison, WI 53792 (firstname.lastname@example.org).
Author Contributions: Dr Zafar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Mallick, Aiken, Tzeng, Weber, Wasif, Zafar.
Acquisition, analysis, or interpretation of data: Mallick, Varley, Abbott, Weber, Zafar.
Drafting of the manuscript: Mallick, Aiken, Tzeng, Weber, Zafar.
Critical revision of the manuscript for important intellectual content: Varley, Abbott, Tzeng, Weber, Wasif, Zafar.
Statistical analysis: Mallick, Aiken, Weber, Zafar.
Administrative, technical, or material support: Weber, Zafar.
Supervision: Abbott, Tzeng, Weber, Wasif, Zafar.
Conflict of Interest Disclosures: None reported.