Association Between End-of-Rotation Resident Transition in Care and Mortality Among Hospitalized Patients

Key Points

Question  Are patients who are exposed to end-of-rotation resident transition in care at risk for greater mortality, and is this association related to the 2011 duty-hour regulations?

Findings  In this multicenter cohort study of 230 701 patients admitted to internal medicine services in 10 Veterans Affairs hospitals, end-of-rotation house staff transition in care was associated with significantly higher in-hospital mortality (3.5% for intern only transition vs 2.0% for control and 4.0% for intern + resident transition vs 2.1% for control). The association was significantly stronger following institution of ACGME duty hour regulations.

Meaning  End-of-rotation transitions may introduce risk in internal medicine inpatient care.

Importance  Shift-to-shift transitions in care among house staff are associated with adverse events. However, the association between end-of-rotation transition (in which care of the patient is transferred) and adverse events is uncertain.

Objective  To examine the association of end-of-rotation house staff transitions with mortality among hospitalized patients.

Design, Setting, and Participants  Retrospective multicenter cohort study of patients admitted to internal medicine services (N = 230 701) at 10 university-affiliated US Veterans Health Administration hospitals (2008-2014).

Exposures  Transition patients (defined as those admitted prior to an end-of-rotation transition who died or were discharged within 7 days following transition) were stratified by type of transition (intern only, resident only, or intern + resident) and compared with all other discharges (control). An alternative analysis comparing admissions within 2 days before transition with admissions on the same 2 days 2 weeks later was also conducted.

Main Outcomes and Measures  The primary outcome was in-hospital mortality. Secondary outcomes included 30-day and 90-day mortality and readmission rates. A difference-in-difference analysis assessed whether outcomes changed after the 2011 Accreditation Council for Graduate Medical Education (ACGME) duty hour regulations. Adjustments included age, sex, race/ethnicity, month, year, length of stay, comorbidities, and hospital.

Results  Among 230 701 patient discharges (mean age, 65.6 years; men, 95.8%; median length of stay, 3.0 days), 25 938 intern-only, 26 456 resident-only, and 11 517 intern + resident end-of-rotation transitions occurred. Overall mortality was 2.18% in-hospital, 9.45% at 30 days, and 14.43% at 90 days. Adjusted hospital mortality was significantly greater in transition vs control patients for the intern-only group (3.5% vs 2.0%; odds ratio [OR], 1.12 [95% CI, 1.03-1.21]) and the intern + resident group (4.0% vs 2.1%; OR, 1.18 [95% CI, 1.06-1.33]), but not for the resident-only group (3.3% vs 2.0%; OR, 1.07 [95% CI, 0.99-1.16]). Adjusted 30-day and 90-day mortality rates were greater in all transition vs control comparisons (30-day mortality: intern-only group, 14.5% vs 8.8%, OR, 1.17 [95% CI, 1.13-1.22]; resident-only group, 13.8% vs 8.9%, OR, 1.11 [95% CI, 1.04-1.18]; intern + resident group, 15.5% vs 9.1%, OR, 1.21 [95% CI, 1.12-1.31]; 90-day mortality: intern-only group, 21.5% vs 13.5%, OR, 1.14 [95% CI, 1.10-1.19]; resident-only group, 20.9% vs 13.6%, OR, 1.10 [95% CI, 1.05-1.16]; intern + resident group, 22.8% vs 14.0%, OR, 1.17 [95% CI, 1.11-1.23]). Duty hour changes were associated with greater adjusted hospital mortality for transition patients in the intern-only group and intern + resident group than for controls (intern-only: OR, 1.11 [95% CI, 1.02-1.21]; intern + resident: OR, 1.17 [95% CI, 1.02-1.34]). The alternative analyses did not demonstrate any significant differences in mortality between transition and control groups.

Conclusions and Relevance  Among patients admitted to internal medicine services in 10 Veterans Affairs hospitals, end-of-rotation transition in care was associated with significantly higher in-hospital mortality in an unrestricted analysis that included most patients, but not in an alternative restricted analysis. The association was stronger following institution of ACGME duty hour regulations.

Quiz Ref IDA handoff is defined as the transition of patient care to another clinician through the transfer of information, responsibility, and authority.¹ Increasing evidence indicates shift-to-shift handoffs are a source of adverse events and errors,² and conversely, that interventions to improve such handoffs may have meaningful benefits to patient safety.^3-5 However, the risk surrounding handoffs may be contingent on the type of transition. An understudied area is end-of-rotation handoff,^6-8 in which resident physicians turn over their service—often to entirely new teams, rather than just a brief overnight period. Although longer length of stay has been associated with service change and hospitalist transitions,^6,8-10 few investigators have demonstrated an association with mortality.^7,11

Quiz Ref IDOn July 1, 2011, Accreditation Council for Graduate Medical Education (ACGME) duty-hour regulations limited first-year residents (interns) to 16 continuous hours of work,¹² which led to an increase in shift-to-shift handoffs among residency programs.¹³ Despite these changes, safety outcomes and mortality rates have remained unchanged,¹⁴ but transition-related outcomes have not been examined. Although the end-of-rotation transition would not have been directly affected by duty hour regulations, residency programs complied with these rules by adopting a day-float and night-float shift schedule in contrast to the former 24-hour call schedule.¹⁵ This may have increased the proportion of residents receiving handoffs of new patients each rotation, with possible detrimental effects for patients already at risk of handoff-related miscommunication, such as those undergoing end-of-rotation transition in care. Therefore, the study objectives were to determine if end-of-rotation house staff transition was associated with greater mortality, and, if so, to assess the association of 2011 duty hour regulations with this relationship.

This study was reviewed and approved by the Veterans Affairs (VA) New York Harbor Healthcare System institutional review board, which provided a waiver of consent for both the patients and physicians in the study. A contact from the internal medicine residency program at each university affiliate was informed about the study, and the contact provided information on their rotation handoff dates. Verbal consent was obtained with assurance of program and hospital anonymity regarding any publication of data.

This was a retrospective cohort study of adult patients (aged ≥18 years) discharged from 10 university-affiliated US Veterans Health Administration hospitals from July 1, 2008, through June 30, 2014. The 10 hospitals were chosen because of the affiliated residency’s ability to provide resident schedules and specific service characteristics. Three residency programs were unable to provide switch dates back to 2008. For these programs only the available switch dates were included; the earlier years were excluded from the data set for each program before analysis.

Data from the VA Corporate Data Warehouse was accessed through the VA Informatics and Computing Infrastructure (VINCI). VINCI compiles and stores data from all VA medical centers nationwide, and provides investigators the analytical tools needed to complete research studies in the VA. The Corporate Data Warehouse is a national repository of data within VINCI from several administrative and clinical sources.

We included medical patients discharged from a general medicine team, intensive care unit, coronary care unit, stepdown unit, or subspecialty team (such as cardiology, pulmonology, hematology, or oncology). The majority of interns and residents for each service were internal medicine–specific house staff, but various nonmedical house staff rotated as well; these data were not available for this study. Nonmedical services or services that rotated on a different schedule were excluded, as were patients with length of stay less than 1 day to ensure patients were potentially exposed to an end-of-rotation care transition during their stay. Exact intern and resident switch dates were collected for each program, but we were unable to further quantify transitions of other clinicians potentially involved in the care of included patients. In anticipation of this limitation, internal medicine residency programs were included only if they provided all inpatient coverage to their respective VA hospitals, rotated all teams with the same block schedule, and did not share rotation switch dates with attending physicians, fellows, or other mid-level clinicians (ie, house staff switched on distinct weeks from other clinicians). Some programs switched both interns and residents within the same week (eg, Monday and Tuesday), whereas some switched up to a week apart (eg, the first and seventh of each month).

Data included age, sex, race/ethnicity, discharge service, length of stay, all comorbidities indexed by the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) codes, and discharge disposition. Race/ethnicity data were included to estimate the characteristics of the population served. They were based on standard, fixed categories used within the US Veterans Health Administration and reflect the most recent patient-identified category recorded at entry to VA health services or point-of-care available in the VA’s electronic medical record system. Comorbidities were used to construct the Elixhauser comorbidity score,^16,17 which has been associated with the risk of hospital death.¹⁸ We also obtained records of any mortality after discharge and readmission. These data were incorporated from the Beneficiary Identification Records Locator Subsystem death records administrative file (an additional administrative data source available within VINCI) that compiles mortality data from multiple VA sources, including benefits and US Veterans Health Administration services. Whether mortality at 30 or 90 days was directly related to the index hospitalization could not be determined from the available data. Only data on discharge and admission to VA facilities were available.

Intern and resident transition were considered independently because they were sometimes separated in time and because level of training may be associated with the quality of the transition. Patients were divided into 2 groups: transition and control. For the main analysis, the transition group included patients admitted at any time prior to an end-of-rotation transition who either were discharged or died within 7 days following the transition (length-of-stay outliers >99th percentile [>33 days] excluded). This definition was chosen to ensure that the exposure group was limited to patients who underwent end-of-rotation house staff transition, and to try to ensure that longer stay patients—who may be at greater risk for both transition and transition-related complications—were included in the exposure group. The exposure group was limited to those discharged or deceased within 7 days of transition⁷ because effects of transition are likely attenuated after prolonged exposure to the new team. All other discharges were considered controls.

Transition patients whose admission spanned more than 1 transition were still included if discharged within 7 days following the final transition. However, this occurrence was uncommon because patients with length of stay greater than 33 days were excluded. Additionally, some transition patients may have been represented in the control group if admitted prior to transition and not discharged within 7 days of transition. To evaluate if level of training was associated with outcomes, we separately defined 3 sets of transition and control cohorts stratified by exposure to intern transition, resident transition, or both intern + resident transition.

An alternate analysis was then conducted using a different exposure definition, in which the transition group included all patients admitted on either of the 2 days before transition, and the control group included all patients admitted on the same 2 days of the week, 2 weeks later.¹⁹ Patients in the exposure group did not have to be discharged within a week of transition. This definition ensures more comparable patient populations but is limited to a subset of the patients, excludes patients with long stays prior to transition, and has more crossover between groups. We assessed the outcomes of all admissions based on the assigned groups, regardless of length of stay.

The primary outcome for all analyses was adjusted in-hospital mortality rate. Secondary outcomes included 30-day and 90-day mortality and readmission rates. In all analyses, 30-day and 90-day mortality were measured from date of admission.

The main objective was to determine whether end-of-rotation transition in care was associated with greater mortality. First, χ² tests were used to determine if there were significant differences in unadjusted mortality in-hospital, at 30 days, and at 90 days between each transition group and control group. Separate multivariable logistic regression models were constructed for intern-only, resident-only, and intern + resident transitions to examine the association of each type of transition with hospital mortality, controlling for confounders. All models included age, sex, race/ethnicity, length of hospital stay (outliers >99th percentile deleted in main analyses), calendar month, calendar year, individual hospital site, and the Elixhauser comorbidity score as covariates. Because the number of individual hospital sites (N = 10) was small compared with the sample size, their effects were considered fixed in regression models. To further evaluate for hospital outliers, we repeated analyses using separate models for each hospital in the main analysis.

To evaluate whether the 2011 ACGME duty hour restrictions had differential associations with patients undergoing end-of-rotation transitions, multivariable logistic regression models were constructed that included a variable for ACGME period (before duty hour regulations [July 1, 2008, to June 30, 2011] or after duty hour regulations [July 1, 2011, to June 30, 2014]) and an interaction between ACGME period and transition (difference-in-differences models). Similar to other outcomes, these analyses were stratified by type of transition.

After completing the primary analyses, we repeated the analyses to assess secondary outcomes including 30-day and 90-day mortality and 30-day and 90-day readmission rates. For 30-day and 90-day mortality, we used individual patient data and selected the last available admission per patient, rather than including all admissions. Statistical analyses were performed using SAS Enterprise Guide, version 6.1, and SAS, version 9.4 (both from SAS Institute). All the P values were 2-sided. P values less than 0.05 were considered statistically significant.

Individual residency program data are displayed in eTable 1 in the Supplement. Only 1 out of the 10 residency programs required in-person sign out at the time of rotation-to-rotation transition, whereas the other 9 used written and verbal sign out. Program-specific end-of-rotation sign out processes were reported to be similar at the VA compared with other hospitals where house staff rotated. Individual hospital census and house staff switch schedules varied, but 90% of programs reported the presence of a formal handoff education program (eTable 1 in the Supplement).

The main study cohorts included 230 701 patient discharges, of which 25 938 (11.2%) were exposed to intern-only transition, 26 456 (11.5%) were exposed to resident-only transition, and 11 517 (5.0%) were exposed to intern + resident transition (Table 1). Among them, 95.8% were men with a mean age of 65.6 years. Mean length of stay was 4.2 days (SD, 4.7; with outliers >99th percentile [>33days] excluded) with a median of 3.0 days (interquartile-range [IQR], 1.0-5.0). Median length of stay was greater for transition patients than controls by 3 days for both the intern-only group (5.0 [IQR, 3.0-9.0] for transition vs 2.0 [IQR, 1.0-5.0] for control) and the resident-only group (5.0 [IQR, 3.0-9.0] for transition vs 2.0 [IQR, 1.0-5.0] for control), and by 4 days in the intern + resident group (6.0 [IQR, 4.0-10.0] for transition vs 2.0 [IQR, 1.0-5.0] for control). Baseline characteristics including comorbidities are outlined in Table 1.

The alternative analyses included 14 669 admissions in the intern-only transition group, 15 068 admissions in the resident-only transition group, and 4729 admissions in the intern + resident transition group. Baseline characteristics were similar between groups, with no significant differences in length of stay (eTable 2 in the Supplement). Because the alternative analysis groups were defined by admission date rather than transition in care, there was crossover between groups. A total of 5652 patients (38.5%) in the intern-only transition group, 5883 patients (39.0%) in the resident-only transition group, and 2231 patients (47.2%) in the intern + resident transition group were discharged or died prior to transition and did not experience a transition in care. Conversely, among the controls, 703 patients (4.8%) in the intern only group, 772 patients (5.1%) in the resident only group, and 216 patients (4.5%) in the intern + resident group remained in the hospital long enough to experience a transition in care (eTable 3 in the Supplement). There was also incomplete overlap between the 2 analytic groups, with only 59% to 66% of alternative analysis transition patients being represented in the main analysis transition cohorts (eTable 3 in the Supplement).

Unadjusted and adjusted mortality rates from the main analyses are presented in Table 2. Unadjusted differences were larger than adjusted differences. Adjusted hospital mortality was significantly greater in transition vs control patients for the intern-only group (3.5% vs 2.0%; odds ratio [OR], 1.12 [95% CI, 1.03-1.21], P < .01) and intern + resident group (4.0% vs 2.1%; OR, 1.18 [95% CI, 1.06-1.33], P < .01), but not for resident-only group(3.3% vs 2.0%; OR, 1.07 [95% CI, 0.99-1.16]).

Quiz Ref IDAdjusted 30-day and 90-day mortality rates were greater in all transition vs control comparisons (30-day mortality: intern-only group, 14.5% vs 8.8%, OR, 1.17 [95% CI, 1.13-1.22]; resident-only group, 13.8% vs 8.9%, OR, 1.11 [95% CI, 1.04-1.18]; intern + resident group, 15.5% vs 9.1%, OR, 1.21 [95% CI, 1.12-1.31]; 90-day mortality: intern-only group, 21.5% vs 13.5%, OR, 1.14 [95% CI, 1.10-1.19]; resident-only group, 20.9% vs 13.6%, OR, 1.10 [95% CI, 1.05-1.16]; intern + resident group, 22.8% vs 14.0%, OR, 1.17 [95% CI, 1.11-1.23]).

The Figure shows unadjusted and adjusted ORs of hospital mortality for each hospital. Unadjusted mortality rates were consistently greater for patients undergoing transition without any apparent outliers and with only 1 program not reaching statistical significance at any level of transition. Single-center–adjusted in-hospital mortality was similarly greater for many hospitals, but statistical significance varied.

Mortality rates from the alternative analyses, including in-hospital, 30-day, and 90-day mortality rates are shown in Table 3. There were no statistically significant differences between transition and control groups for in-hospital, 30-day, or 90-day mortality rates.

Adjusted 30-day and 90-day readmission rates from both the main analyses (Table 2) and alternative analyses (Table 3) were not significantly different for any comparison.

As shown in Table 4, adjusted hospital mortality for all transition groups (intern only, resident only, and intern + resident) before duty hour regulations was not significantly different than controls. However, adjusted hospital mortality was significantly greater in both the intern-only and intern + resident transition groups compared with controls in the period after duty hour regulations (intern only: 3.3% vs 1.8%; OR, 1.16 [95% CI, 1.08-1.25], P < .001; intern + resident: 3.7% vs 1.8%; OR, 1.24 [95% CI, 1.11-1.39], P < .001).

When these associations were tested using difference-in-differences analyses (Table 4), mortality rates for intern-only and intern + resident transition patients worsened compared with controls following duty hour regulations (intern only: OR, 1.11 [95% CI, 1.02-1.21], P < .05; intern + resident: OR, 1.17 [95% CI, 1.02-1.34], P < .05).

Alternative analysis results regarding duty hour period comparisons are presented in Table 4. These results were not statistically significant for any comparison before or after the implementation of duty hour regulations.

In the main analysis of this large multicenter study, hospitalized medical patients exposed to end-of-rotation house staff transition in care experienced an absolute increase of 1.5% to 1.9% in unadjusted in-hospital mortality risk, and an adjusted OR of 1.12 to 1.18 for in-hospital mortality compared with control patients. Furthermore, the adjusted ORs for 30-day and 90-day mortality were 1.10 to 1.21 for transition vs control patients for all types of transitions. The associations were stronger following the institution of ACGME duty hour regulations. There were no significant associations with readmission rates. In contrast, the findings of the alternative analysis showed no significant difference in mortality between hospitalized medical patients admitted just before end-of-rotation transition and control patients admitted on the same days in the middle of the rotation. Similarly, these groups were not significantly different in 30-day and 90-day mortality or readmission rates. Together, these results showed that end-of-rotation transitions in care were associated with increased mortality; however, this increased risk may be limited to longer-stay, complex patients with greater comorbidities or those discharged soon after the transition.

Physician transitions in care contribute to adverse patient events and outcomes,^2,4,20 but studies have focused predominantly on shift-to-shift handoff or hospital discharge.^1,21-23 The results of this study suggest that end-of-rotation handoff is a critical transition that deserves further attention. In 1994, a study by Rich et al⁸ evaluated intern rotations, showing longer length of stay for intern transition patients but no association with mortality or readmissions. In 2002, a study by Smith et al⁶ evaluated end-of-the-month admissions, finding a nonsignificant increase in length of stay without differences in mortality or in-hospital adverse events when compared with middle-of-the-month admissions. More recently, in 2015 a similar study compared mortality the week following end-of-rotation transition in care with all other weeks at a large urban medical center, observing an association between end-of-rotation handoff and adjusted hospital mortality.⁷

There are a number of potential explanations for the discordant findings of mortality risk with end-of-rotation transitions between the 2 analyses. First, they included different populations of patients. The main analysis included the majority of patients who experienced a transition, particularly complex and long-stay patients. By contrast, the alternative analysis included only patients admitted immediately prior to transition, by definition excluding long-stay and potentially more complex patients. Quiz Ref IDIt is therefore possible that transitions in care increase the risk of mortality only for patients who have already had a complex hospital course or prolonged length of stay. For these patients, incomplete information transfer or nonfamiliarity with patients, which have been shown to accompany transitions in care,^9,10,24-26 may be particularly harmful, increasing mortality risk.

Second, limitations of the alternative analysis may have impaired its ability to demonstrate a mortality effect. The smaller sample size of the alternative analysis reduced its power and excluded many patients undergoing transition. Additionally, it produced substantial crossover, such that over a third of patients in the “transition” groups did not actually experience a transition. Third, it is possible that the definition of the analytic groups in the main analysis may have created unmeasured confounding, as illustrated by the substantial differences in the main analysis between the exposure and control groups for length of stay and comorbid conditions. Moreover, the increased length of stay in the transition group may have produced greater opportunity for in-hospital mortality. To mitigate these potential biases, the analyses were adjusted for comorbidities and length of stay. However, it is possible that the differences observed in patient populations might represent a direct consequence of clinical decisions made because of an upcoming transition rather than confounding. That is, if clinicians try to discharge as many patients as possible prior to transitioning off service, but have more difficulty discharging complex and long-stay patients, the average severity or complexity of patients exposed to transitions in care could be increased. In addition, disruptions from transitions in care may have delayed patients’ discharge and prolonged their length of stay. In that case, the differences between groups in the main analysis would not be bias, but a consequence of the transition.

The increased 30-day and 90-day mortality risks observed in the main analysis suggest that the delayed discharge of these complicated patients following transition could be detrimental. A reasonable explanation for these findings could be communication failures that led to errors at discharge or possibly another complication changing the trajectory of the patient’s prognosis after discharge. The alternative analysis, however, did not demonstrate these findings, which could be related to the noted differences between analyses. For example, the main analysis specifically evaluated patients discharged soon after transition whereas the alternative analysis measured patients admitted just prior to transition, excluding those patients with longer stays before transition who were likely the most at risk for misinformation transfer. The difference in patient population between analyses is further demonstrated by the finding that only 59% to 66% of alternative analysis transition patients were also considered a transition patient in the main analysis.

The transition groups that were consistently associated with increased mortality in-hospital and after discharge involved the transition of interns, whether alone or in combination with residents. The mortality risk increased after the 2011 ACGME duty hour restrictions, which primarily affected interns, even though studies have found that duty hour regulations may not be harmful^14,27-29 and may be associated with benefit.^30,31 The importance of intern transitions may be attributable to a number of factors. First, interns perform a majority of the day-to-day work of a medical team. They gather, maintain, and transfer the majority of day-to-day information for hospitalized patients, but they are also the most inexperienced residents, increasing the risk for misinformation transfer and potential errors. Because 2011 duty hour regulations primarily limited intern hours,¹² shift-to-shift handoffs have increased noticeably,¹³ prompting a prospective trial of medical residencies.³²

Quiz Ref IDSecond, for residency programs to comply with the duty hour regulations, interns now rotate on shorter “night-float” rotation schedules, in contrast with longer 24-hour call schedules.¹⁵ As a result, the proportion of residents taking handoffs for new patients each block has likely increased correspondingly. This change in day-to-night continuity exposes patients each month to a new team of physicians during the day and at night, when previously they would have only 1 team of physicians taking 24-hour call. Third, interns are now expected to complete the same tasks in less time than in the past, leaving them with limited time to prepare for service change. Although research has shown that medical errors were reduced when interns worked shorter shifts,³³ the present findings suggest that current tools and schedules to standardize end-of-rotation transition may need to evolve, particularly because only 1 of 10 surveyed programs in this study required in-person sign out at the time of rotation-to-rotation transition.

This study has additional limitations. First, VA medical centers include a patient population made up predominantly of older men with multiple comorbidities. Nonetheless, the included programs indicated that end-of-rotation sign out processes were similar at the VA compared with the other hospitals through which house staff rotated. The analysis did not account for other clinicians, such as attending physicians, physician assistants, or nurse practitioners, but only programs in which these clinicians had a different rotation schedule from the house staff were included to minimize potential bias. Although the number of non–internal medicine house staff that rotated on medical services could not be quantified, end-of-rotation transitions would not be expected to differ. Certain handoff dates for 3 of the residency programs were unavailable; therefore, these missing years were omitted from each hospital’s respective data set prior to analysis. Non–teaching service patients could not be excluded at some of the included hospitals. These patients were covered by nonrotating clinicians without monthly rotating residents. However, this would likely lead to a conservative bias that would be expected to dilute any true effect. Other limitations include the retrospective, observational design; that these compared cohorts may have differed in important ways not accounted for, such as socioeconomic status; and the unavailability of data on specific handoff processes of care, such as the type and amount of education provided at each program.

Among patients admitted to internal medicine services in 10 Veterans Affairs hospitals, end-of-rotation transition in care was associated with significantly higher in-hospital mortality in an unrestricted analysis that included most patients, but not in an alternative restricted analysis. The association was stronger following institution of ACGME duty hour regulations.

Corresponding Author: Joshua L. Denson, MD, Division of Pulmonary Sciences and Critical Care Medicine, University of Colorado, 12700 East 19th Ave, Room 9023, Mail Stop C272, Aurora, CO 80045 (joshua.denson@ucdenver.edu).

Correction: This article was corrected for omission of a funding/support acknowledgment and a disclaimer on January 3, 2017.

Author Contributions: Drs Denson and Sherman had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Denson, Jensen, Saag, Horwitz, Evans, Sherman.

Acquisition, analysis, or interpretation of data: Denson, Jensen, Wang, Fang, Sherman.

Drafting of the manuscript: Denson, Jensen, Saag, Horwitz, Evans, Sherman.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Wang, Fang.

Administrative, technical, or material support: Denson, Jensen.

Supervision: Denson, Horwitz, Evans, Sherman.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This material is the result of work supported with resources and the use of facilities at the Veterans Affairs New York Harbor Healthcare System.

Role of the Sponsor: The Veterans Affairs New York Harbor Healthcare System had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.

Additional Contributions: We thank Jack Naggar, MD (Tufts Medical Center), for his uncompensated assistance in the initiation of this project.