Delirium and neuropsychological recovery among emergency general surgery survivors (DANE): study protocol for a randomized controlled trial and collaborative care intervention

This is a multicenter, prospective randomized controlled trial which will evaluate the efficacy of a collaborative care model, delivered via telehealth, in improving the cognitive, functional, and psychological recovery of older emergency general surgery delirium survivors. Our study sites span three states in the Midwest and include a safety net hospital serving many underprivileged and minority patients (Eskenazi Hospital), four community hospitals (Indiana University Health (IUH) North, IUH West, Meriter Hospital, and Swedish American Hospital), and three tertiary academic health centers (IUH Methodist, IUH University Hospital, and the University of Wisconsin Health University Hospital). Together, our team has assessed close to 8000 intensive care unit patients as well as older injured patients for eligibility and enrolled nearly 1000 patients into two NIH-funded trials. For this study, we plan to randomize 528 patients.

Patients will be eligible for inclusion in this study if they are English-speaking, aged 65 years and older, admitted to one of our participating hospitals, and have undergone one of the following procedures in an emergent, urgent, or unplanned fashion: colectomy, small bowel resection, repair of peptic ulcer disease, lysis of adhesions, laparotomy, esophagectomy, liver resection, hepatopancreatobiliary surgery, gastrectomy, hernia repair, cystectomy, gynecological surgery, major vascular procedures, and major orthopedic (e.g., fractures) procedures. They will also need to be able to provide consent or have a legally authorized representative to provide consent, have access to a telephone or Internet-connected computer or smart device, be discharged to home or sub-acute rehabilitation, and have had at least one episode of delirium in the period prior to discharge from acute care.

Broadly speaking, the exclusion criteria will target patients in overall declining health, for whom it may be unethical to intervene or for whom there is no expectation of full recovery. These include a self-reported diagnosis of cancer with short life expectancy (< 6 months), current chemotherapy or radiation therapy confirmed in the electronic medical record (EMR), a history of Alzheimer’s disease and related dementias (ADRD) and other neurodegenerative disorders such as Parkinson’s disease or vascular dementia, incarcerated or unhoused status at the time of enrollment, acquired neurologic injury (stroke, traumatic brain injury, cerebral edema/swelling, anoxic brain injury, or any other acute/subacute neurologic deficit) as the admitting diagnosis or during the course of the hospital stay, a history of bipolar disorder or schizophrenia confirmed by EMR, current alcohol consumption > 5 drinks per day by self-report and/or confirmed by EMR, and history of drug abuse within the last 3 months.

Both the University of Wisconsin and Indiana University Institutional Review Boards (IRB) approved this randomized controlled trial. This trial is registered on ClinicalTrials.gov (NCT05373017). Our team will approach all eligible participants prior to discharge for enrollment. We will obtain informed consent from the patient or a legally authorized representative by a trained research assistant (RA) or research coordinator. After the explanation of the study, all potential trial participants will be given a printed and/or electronic consent document and HIPAA release form. All personnel will be IRB-approved and trained on confidentiality, ethical research practices, and enrollment procedures. All participant data will be collected, shared, and maintained within IRB and HIPAA guidelines, and only approved study team members will have access to the collected data. The data will be stored in a secure Research Electronic Data Capture (REDCap) database on a password-protected University of Wisconsin server.

Adverse events will reported to the co-principal investigators, statisticians, and the Data Safety Monitoring Board (DSMB) as well as the IRB, with appropriate action as deemed per policy. We will also formally document such occurrences. Re-hospitalizations, emergency department visits, and deaths will be reported via EMR notifications or from family members/caregivers at the time of outcome assessments, which are provided to the DSMB at regular meetings. Any other adverse events will be provided by reports from patients or their families/caregivers.

This study is a multi-site, randomized controlled clinical trial. Due to the nature of the intervention, we will perform a two-arm (1:1), single-blinded study to evaluate the superiority of the proposed DANE collaborative care model over usual care. At or near the time of discharge, enrolled patients will be randomized to receive the DANE intervention using a computer-generated randomization scheme, stratified by recruitment site and whether the procedure was primarily vascular, due to the known delirium risk with that operation type. The comparator will be standard care, which will consist of postoperative care visits and calls as would be typical at the respective facility and under the guidance of the care team. Patients will then be followed for 18 months (Fig. 1).

The Delirium and Neuropsychological Recovery among Emergency General Surgery Survivors (DANE) Recovery Model, our intervention, will occur across five phases corresponding to the guidelines established for conducting follow-up studies in injured patients [10]. Additionally, this study adheres to a conceptual model and classification scheme that organizes health outcomes into five levels [11]. During the acute stage (0–1 month after surgery), patients will undergo the initial case review and two initial virtual visits, and we will develop the recovery care plan. During the recovery stage (2–6 months after surgery) and maintenance stage (7–12 months after surgery), the patients will undergo the interaction phase of the DANE Recovery Model. Based on our prior work in a critical care recovery program targeting ICU survivors, we know that improvements in the recovery begin to manifest at 6 months, and we expect survivors of postoperative delirium to follow a similar trajectory [12].

At the initial case review, randomized patients, the care coordinator (CC) will introduce the DANE Recovery Model and determine the best mode of contact for future meetings. The time for the first visit will also be arranged, which marks the beginning of the recovery model.

After completing a pre-visit review, the CC will conduct the first virtual visit using a mobile visit technology to perform an initial assessment. The CC will also perform a social and community needs assessment, reconcile all prescribed and over-the-counter medications, and note all scheduled and recommended appointments with specialists, physical and occupational therapists, and other providers. The Healthy Aging Brain Care (HABC) Monitor will be used to evaluate and track cognitive, functional, and psychological symptoms of patient and caregiver stress and to trigger the use of specific treatment protocols [13].

The second virtual visit will include a detailed discussion of the individualized recovery care plan with the patient and caregivers, which will include the process of monitoring recovery progress, implementation of recovery protocols, distribution and explanation of educational materials, and connection to in-home and community resources as needed.

The 12-month interaction period with patients and/or informal caregivers will include virtual visits, phone calls, email, and mail. The minimum amount of contact during this period will be every 2 weeks. At the end of the interaction period, the CC will transition the care to the patient’s primary care physician. Activities during these interactions will include addressing any questions, collecting feedback, reviewing/reconciling medications and adherence, scheduled appointment reviews, HABC Monitor assessments, sleep monitoring, and facilitating access to any other resources in the community if needed.

Participants are free to withdraw from participation in the study at any time upon request. Those who do so will be informed that no new procedures will be performed after notice of study withdrawal. Information obtained up to that point will remain part of the study. The reason for withdrawal will also be recorded.

Patients randomized to usual care will also receive an initial review by the care coordinator prior at or near hospital discharge to confirm eligibility criteria, contact information, review discharge and rehabilitation plans, identify the primary care physician, and compose and send care transition information. Patients in the usual care group will receive no further interventions after this initial review. Concomitant care including prescription medications, over-the-counter medications, and supplements is permitted and will be tracked by the study team.

On or within 48 h of anticipated hospital discharge and after obtaining informed consent, the RA will obtain baseline measurements of physical function, depression, anxiety, quality of life, and cognitive ability. Subjects will then be randomized in a 1:1 manner via a computer-generated randomization scheme, stratified by recruitment site and surgery type (vascular or non-vascular surgery). All subjects will receive the interventions described under usual care. We will make every effort to blind the research staff performing the assessments to the intervention group. Intervention allocation will be coordinated centrally through the University of Wisconsin to minimize staff exposure to randomization assignments. Furthermore, the study staff will adhere to a strict study script during survey administration with patients and caregivers, and a standardized chart abstraction protocol with a data dictionary will be used. Due to the type of intervention, blinding of the patients is not possible. Physicians and other providers caring for the patient will be made aware of the study but not its objectives. Though collaborative care adverse event rates are low, blinded adverse event rates will be presented to the statistician and principal investigators throughout the trial. Decisions regarding unblinding will be made if there are elevated adverse event rates. The research team will be unblinded should any serious adverse events be deemed to be related to the intervention and the DSMB notified immediately.

The primary aims of this study are to evaluate the efficacy of the DANE Recovery Model to improve the cognitive recovery of older emergency surgery delirium survivors, as determined by the Repeatable Assessment of Neuropsychological Status (RBANS). Secondary aims include evaluation of physical and psychological recovery, as measured by the Short Physical Performance Battery, Medical Outcomes Study Short Form (SF-36), Patient Health Questionnare-9 (PHQ-9), and Generalized Anxiety Disorder Scale (GAD-7).

Chart review will be used to abstract relevant covariables. Baseline assessments will occur at or as close (within 4 weeks) to hospital discharge as possible. After baseline assessments, outcomes will be collected at 6, 12, and 18 months. We will make every effort to retain patients during the 18-month study period and complete all follow-up instruments. We will attempt to do this in person. We will abstract data in a standardized fashion into a REDCap database. We will ensure data quality by independent abstraction and double data entry of 10% by the site PI or suitable designee. Any changes to the study protocol, including those related to the inclusion and exclusion criteria, will be communicated to key personnel, trial participants, program officers, and IRBs.

We will compare randomization results to the planned randomization schedule. Baseline characteristics between the intervention and standard care arm will be compared using analysis of covariance (ANCOVA) for continuous variables and the Cochran-Mantel-Hansel statistic for categorical variables while adjusting for stratifying variables. Patients will be stratified by hospital and surgery (vascular vs. non-vascular procedures). We will also examine the distributions of continuous variables in order to verify the normal distribution assumption, and use alternative procedures (transformation or non-parametric methods) if this assumption is violated. Frequency distributions of all categorical variables will also be examined, and exact inference procedures will be used in cases of zero or small cell size. Intention-to-treat analysis will be conducted for the primary and secondary aims.

For our primary outcome of cognitive recovery, we will use mixed effects models with repeated RBANS scores collected at baseline and at 6, 12, and 18 months as outcome measures. Group, time, a group-by-time interaction, and other baseline covariates found to be significantly different in univariate comparisons will be used as independent variables. We will adjust for stratification variables. We will also conduct post hoc comparisons at each follow-up time following a significant interaction between group and time to determine the time when a group difference can be detected. Parameter estimation and inference for the mixed-effects models are conducted using the maximum likelihood approach which are robust under many missing data mechanisms.

For our secondary aims of physical and psychological recovery, we will develop separate mixed effects models for physical function scores (SPPB and physical component score on the SF-36) collected at baseline and at 6, 12, and 18 months as the outcome measures for the former and PHQ-9, GAD-7, or mental component score on the SF-36 for the latter. In both sets of models, we will use the same independent variable selection and stratification variables as in our primary outcome.

In an exploratory analysis, we will plan to investigate the heterogeneity of treatment effects whenever a treatment effect is detected in order to determine if the treatment has different effects in various subgroups defined by sex, race/ethnicity, socioeconomic status, and age.

We completed a sample size calculation using a mixed effects model and adjusted for correlations among repeated measures. Effect sizes were derived from prior work [9, 12, 19, 20]. Assuming a base correlation of 0.3 and a decay rate of 0.1 in a linear exponent autoregressive correlation structure for repeated measures and effect sizes of 0.2 SD at 6 months, 0.4 SD at 12 months, and a maintained effect size of 0.4 SD at 18 months, 185 participants per group are required to yield a power of 80.6% to detect differences in the change in RBANS between the intervention and usual care groups with type I error rate at α = 0.05. Assuming a 30% attrition rate, we need to enroll 264 patients per group into the study with a total enrollment target size of 528. Similar power will be achieved for the secondary outcome measures. Power estimates were calculated using the GLMPower procedure in SAS 9.4.

This trial will be monitored every 6 months by a Data Safety Monitoring Board (DSMB) and reviewed annually by the University of Wisconsin IRB. The board will consist of three members: a surgeon, a biostatistician, and a safety officer. This team will be separate from the study team without competing interests. Review reports will include any necessary additional actions as necessary (e.g., corrective intervention, ad hoc review, stopping rule, communication requirements with study participants or investigators). The DSMB will meet twice annually either in-person or by teleconference call to review the study progress, data quality, and participants’ safety. We will submit quarterly reports to the DSMB in addition to these in-person meetings. Our investigator group will continue to meet on a weekly basis to discuss trial progress, weekly enrollments, specific concerns, and amendments to the trial protocol. All clinical outcomes will be systematically tracked throughout the study period. Clinical outcomes including death, hospital or ICU readmission, depressive symptoms, anxiety symptoms, agitation and behavioral disturbances, and fall/mobility problems will be tracked as clinical outcomes and are not required to be reported as adverse events. Adverse events will be classified in terms of severity, expectedness, and relatedness. The study coordinator and biostatistician will generate reports for the PI, safety officer, and DSMB which will contain summaries of adverse events, complaints, retention, intervention compliance, and protocol violations, including how they were handled. Adverse events will be reported within 24 h to the study team, to the IRB within 14 business days, and to the NIA within 48 h. A summary of all reportable adverse events and clinical outcomes will be reported at least twice a year (initially quarterly and then every 6 months beginning at the time determined by the DSMB).

All electronic data is password and firewall security-protected, and any data set containing patient identifiers will be managed by the investigators, sub-investigators, or a designated programmer. Data sets with patient identifiers will be stored indefinitely and for a minimum of 7 years on limited-access folders on UW and IU servers in which only the key study personnel listed on the IRB application will have access.

Patients who meet the screening criteria will be approached while in the hospital following their operation. They will be screened for the development of delirium with EMR documentation or with the CAM-ICU. Surrogate consent will be obtained for patients deemed unable to provide their own consent. Based on our experience with previous studies by our team, we assume a very conservative consent rate of 40% over the 36-month recruitment period. Thus, we expect to enroll at least 24 patients per month which is more than enough to meet our target enrollment of 15 patients per month in order to reach our target sample size of 528 subjects. To enhance consent and recruitment rates, we have established a strong relationship with the emergency general surgery clinical teams at all the enrolling hospitals and have enrolled subjects within these hospitals for previous studies. The primary investigator on this trial and the co-investigators who all have direct involvement in the care of emergency surgery patients.

With respect to retention, the usual care group presents the greatest challenge. As part of our quality control procedures during enrollment, we will monitor the subjects’ perceptions about the risks and benefits of participation. We have and will continue to recruit research personnel who are representative of our target population and seek to identify individuals who have life experiences in the community from which they are recruiting. If retention drops below 80%, we will have staff follow up with the participants to troubleshoot issues and provide coaching if necessary. We will also institute gift card incentives, and we will use fair subject payments contingent on the completion of the baseline and 6-, 12-, and 18-month follow-up assessments. Finally, we have developed alternatives to face-to-face visits for measuring the SPPB and RBANS in the event that these visits are not possible.

Trial results will be prepared and submitted in manuscript form for peer review. Trial results will also be communicated to participants and participating institutions and presented at local and national surgical and aging conferences. Access to collected data will depend on the outcome to be measured. Interested individuals will make a data request and prepare a project proposal that includes project goals, outcomes data to be collected, and the planned analysis. This project, if approved, will be carried out under the direction of the principal investigators after it has been deemed to fall under the intended use of this application. Publication authorship will be based on the relative scientific contributions of the PIs and key personnel.