With limitations on health care costs and recent technical innovations, an increasing number of procedures, including major cancer procedures (e.g., mastectomies, hysterectomies, and prostatectomies), are being performed with an overnight stay. short (one night), or as outpatient surgeries [1-3].
Although short hospital stays have many advantages, that model adds complexities to the delivery of high-quality, patient-centered care, particularly for cancer patients and their caregivers, who are often still struggling with a new cancer diagnosis.
Patients may leave the surgery center while still experiencing symptoms that could have previously been addressed by the hospital’s treating team [4]. Managing symptoms at home can be challenging for patients and their caregivers, who may have difficulty distinguishing normal, expected symptoms from potentially serious adverse events [5]. Without information and awareness of risk, patients may delay seeking care, with severe consequences, or may experience unnecessary anxiety and seek unnecessary care [6].
One approach to that challenge is through the collection of patient - reported outcomes (PROs), which are rapidly becoming a standard of care across medicine to monitor symptom burden. There is extensive experience that PRO data can improve communication with the medical team, quality of life, and patient satisfaction by helping to manage patient symptoms [7-9].
At Memorial Sloan Kettering Cancer Center (MSK), a large randomized trial comparing routine PRO collection with usual care during chemotherapy showed that among patients who received PRO intervention, quality of life improved (34 % vs 18%), and who were less likely to be seen at the emergency care center (CAU) (34% vs 41%; P = 0.02), or to be hospitalized (45% vs 49%; P = 0.08) [10].
PRO-based remote monitoring of patients undergoing outpatient surgery was implemented at MSK in 2016, as a component of routine clinical care [11]. Through this system, called “recovery tracker” ( Recovery Tracker [RT]), patients were asked to report symptoms daily for 10 days after discharge. If symptoms above the threshold were reported, alerts were generated for the care team, and nursing staff called the patient to investigate.
A retrospective analysis of more than 7000 patients demonstrated that the use of RT led to a 42% risk reduction of CAU visits, and did not result in hospital admission, that is, a potentially avoidable visit [12]. However, RT also led to an increase in nursing staff workload, with an average of 2.5 calls to 3.4 calls per patient within 30 days (0.86 call increase; 95% confidence index [CI]: 0.75-0.98; P < 0.001).
To increase the utility of remote monitoring while employing fewer nursing resources, a system was developed that provides patients with real-time automated information about the severity of expected symptoms, and allows them to activate care as needed, here called "enhanced feedback." ".
The objective of this randomized controlled trial was to compare “enhanced feedback” with the described standard of care (nursing extension if symptoms exceed normal limits), called “team monitoring,” and its potential impact on decreasing CAU visits. and readmissions, patient anxiety, and nursing workload, up to 30 days after outpatient cancer surgery.
The authors hypothesized that patients in the enhanced feedback group would be better empowered to make decisions related to their symptoms, improve the patient experience, and prevent potentially avoidable visits to the CAU, using fewer resources than team monitoring.
| Methods |
> Study design
This is a single-center randomized controlled trial with stratified randomization by 1:1 procedure between 2 arms, team monitoring, and enhanced feedback, using PCORI standards as previously described [13]. The protocol was approved by the MSK Institutional Review Board as a low-risk protocol, with a waiver for informed consent. The study was registered at clinicaltrials.gov (NCT03178045).
> Patients and randomization
Patients were eligible to participate if they had undergone oncologic surgery at the Josie Robertson Surgery Center , an outpatient surgery hospital at MSK [11], between September 15, 2017, and September 30, 2019. To be eligible , patients need to be over 18 years old, at risk for and/or diagnosed with breast, gynecologic, urologic, and/or head and neck cancer; and be able to complete electronic surveys from home and in English.
Before surgery, eligible patients received written materials describing the study, via their portal ( MyMSK ), email, or a letter sent to their home. Patients were contacted by the study team to obtain verbal consent to participate. Randomization was performed within 1 week of the patient’s surgical consultation using the MSK Clinical Research Database , a secure, password-protected database that ensures complete allocation concealment.
Randomization was stratified by procedure, and implemented through randomly permuted blocks of random length. Although patients were not explicitly informed of their randomization arm, the trial was not blinded as the intervention was obvious to patients and recognizable to the clinical team.
> The recovery tracker
RT is considered the standard of care at the study institution, and was developed with the participation of multiple stakeholders, to monitor patients remotely during the immediate postoperative period [11,12]. In summary, it includes 19 questions adapted from the validated PRO version of the Common Terminology Criteria for Adverse Events (CTCAE) [14], sent to patients via the MyMSK Patient Portal for 10 days after surgery.
Responses above the specific threshold for each symptom, and the duration of time after surgery generate alerts in the clinical team. For symptoms at an intermediate level, a “yellow” alert is sent to the surgeon’s office team, and the patient is evaluated by phone or secure portal message during regular office hours.
For high-risk symptoms, such as severe shortness of breath, a “red” alert is sent to the surgeon’s office team and the patient also instantly receives an electronic pop-up window instructing them to contact the doctor immediately.
> Intervention
All patients in the study completed RT. When symptoms were outside the expected range for patients randomized to the team’s standard care monitoring, the alerts prompted the clinical team to contact the patient in the manner already described. Patients randomized to enhanced feedback also received an immediate formatted report providing automated normative feedback about their expected symptoms for that procedure, but no “yellow alerts” were sent to the clinical team.
Both groups were warned to contact their doctors for “red alert” level symptoms. The development of the feedback reports was informed by real patient data, and feedback from patients, providers, and caregivers [15].
Study patients received daily email reminders to complete RT, from postoperative day 1 to day 10; Additional optional surveys were available to be completed by patients, if they desired, from the 11th to the 30th postoperative day.
> Results
The primary outcome was CAU consultation, without readmission, within 30 days of surgery, an event that could potentially be avoided with proactive symptom management [16,17]. Secondary outcomes included: CAU visits with readmission within 30 days, any readmission within 30 days, patient anxiety, adverse events within 30 days, calls to nursing, referrals for pain management, and consultations. clinics within 30 days of surgery. CAU visits and readmissions were collected from MSK institutional databases.
Patient anxiety was measured with 3 items from the PRO version of the CTCAV survey, which were summed to generate a global score from 0 to 12; patients with the absence of any of the 3 items were considered missing.
Adverse events were captured by the MSK Secondary Surgical Event database , the National Surgical Quality Improvement Program database , and the institutional MSK Clinical Research Database . Other symptom-related events, including the number of nursing calls, pain management referrals, and clinic visits, were extracted from the institutional database and electronic medical records.
> Statistical analysis
Based on data available at the MSK at the time of study design, it was expected that for every 1000 eligible patients treated surgically at the Josie Robertson Surgery Center , 69 patients would have consultations at the CAU, of whom 28 would require readmission and, Therefore, 41 could have consultations in the CAU without readmission.
Using a traditional alpha of 5% and an event rate of 4.1% in the control group, a statistical power of 85% was obtained to detect a relative risk reduction of 50%, based on a total sample size of 2750 patients.
To evaluate the primary outcome, a difference in the risk of the patient having any CAU visit without readmission to MSK within 30 days of surgery, multivariable logistic regression was used to test its association with randomization arm. with procedure type as a covariate.
From that multivariable logistic regression model, a corresponding adjusted risk difference between arms was calculated by calculating a predicted risk in both arms, separately setting the type of procedure to the model value, with a confidence interval ( 95% CI calculated through 2000 samples loaded into computers. That analysis was repeated for the results of CAU visits with readmission within 30 days, and for any readmission within 30 days.
A mixed-effects longitudinal regression was used to test the association between total anxiety score and randomization arm, after adjusting for postoperative day, type of procedure, and an interaction between randomization arm and postoperative day with an intercept. random for the patient.
The interaction of postoperative day by arm was used to test whether the relationship between anxiety and postoperative day differed by arm. The patient’s anxiety score the first time the RT survey was used was excluded, because we wanted to assess patient anxiety as a result of the intervention (enhanced feedback versus team monitoring).
The association between randomization arm and the number of follow-up nursing calls within 30 days was evaluated using negative binomial multivariable regression, with type of procedure as a covariate.
Multivariable logistic regression adjusted for type of procedure as a covariate was used to test for an association between randomization arm and the risk of referral for pain management or risk of an adverse event within 30 days. All calculations were carried out with the R program, version 3.6.1, using the lme4 version 1.1-23 package to fit the linear mixed effects model.
| Results |
A total of 8659 patients were evaluated for eligibility, and 2823 of them met the criteria. Of these, 2793 were randomized and, of those randomized, 93 underwent procedures deemed postoperatively ineligible, 64 had their surgery suspended or relocated, and 12 declined consent, leaving 2624 patients for analysis. The analysis was performed following intention-to-treat principles and results were compared by randomization group: 1314 in the enhanced feedback arm vs. 1310 in the team monitoring arm.
The median age in both arms was 54 years and >70% of the cohort were women (a result of the predominance of breast and gynecological surgical procedures). Almost all patients (99%) received general anesthesia.
The majority of patients (65%) underwent outpatient extended recovery procedures, defined as outpatient procedures with a planned overnight stay [18] (a small fraction of patients with outpatient extended recovery procedures were discharged on the same day of surgery). .
Participation through RT was high. There was no difference in completion rate between the two arms. A total of 90% of patients completed RT at least once, 84% ≥ 4 times, and 62% on all 10 postoperative days. On average, patients submitted 7.2 RT responses.
Overall, 173 patients presented to the MSK CAU after surgery, and 71 were readmitted within the first 30 days. Multivariable analysis found no evidence of a difference between groups in the risk of CAU visits without readmission ( P = 0.12), nor in the risk of CAU visits with readmission, or any readmission ( P = 0.12). 4 for both).
The adjusted absolute risk difference between the risk of a CAU visit without readmission was 1.0% (95% CI: -0.23–3.1%) higher among those randomized to receive improved feedback.
A small proportion of subjects randomized to receive enhanced feedback (130) did not receive some features of that feedback due to a technical error, while all patients randomized to equipment monitoring received correct feedback.
Therefore, a per-protocol sensitivity analysis was performed comparing those randomized to equipment monitoring with the subgroup of patients who correctly received enhanced feedback. The sensitivity analysis estimates moved slightly toward the null, confirming the primary conclusion.
Patients randomized to enhanced feedback had a more rapid reduction in their anxiety score, where the maximum score is 12, by 0.04 (95% CI: 0.02-0.06) per postoperative day, compared to those randomized for equipment monitoring. That translates to a greater reduction in anxiety in the enhanced feedback arm of 0.36 points of the survey over the course of 9 days.
The number of calls to nursing was 10% (95% CI: 5-16%) lower in the enhanced feedback group at 30 days. When the analysis was repeated limited to the 10 days in which the enhanced feedback was provided, the number of calls to nursing was 14% lower (95% CI: 0.08-0.19; P < 0.001).
Overall, 106 patients (4.0%) experienced one adverse event, with 3 (0.1%) having two adverse events. There was no significant difference in the incidence of adverse events between the groups (0.4% higher among patients receiving enhanced feedback: 95% CI: -1.1–2.0%).
There were also no differences in pain management with 12 (0.9%) and 11 (0.8%) in the enhanced feedback and team monitoring arms, respectively, or in clinical consultations (incidence rate ratio 0 .99; 95% CI: 0.92-1.06; P = 0.7).
| Discussion |
Enabled by technical innovations in surgery and anesthesiology and driven by pressures to reduce costs, an increasing number of operations are performed as outpatient procedures [1-3]. Previous work by the authors’ group suggests that remote monitoring of PROs can significantly reduce unplanned care [12].
In the large randomized controlled trial reported here, we found that adding automated feedback to routine PRO assessment reduced patient anxiety and nursing workload, while maintaining the previously observed reduction in CAU visits. .
The finding that rates of CAU consultations and adverse events were similar between the two groups provides important reassurance that immediate, normative, and contextual feedback to a post-surgical patient about self-reported symptoms can safely supplant a discussion with knowledgeable nursing staff, to support the patient’s decision making, or manage expected moderate symptoms.
The authors’ previous retrospective work showed that, by itself, monitoring reduces potentially avoidable visits to the CAU, that is, those that do not result in readmission, by 44%, but at the expense of 0.9 additional calls to nursing. per patient [12].
By offering improved automated feedback, the data presented indicate that the same decrease in unplanned care can be achieved, with fewer nursing resources. This reduction in calls to nursing is clinically significant when translated into the cumulative number of calls per patient.
At MSK, which performs nearly 10,000 outpatient cancer surgical procedures annually, a reduction of 0.4 calls per patient represents 4,000 phone calls. Symptoms reported with enhanced feedback can also be used to help support patients in settings with fewer nursing resources for postoperative management, or with higher rates of "unanticipated care."
In addition to the positive impact on resource utilization, automated feedback also appears to help decrease patient postoperative anxiety. After outpatient cancer surgery, symptoms are very common [6,19] and a considerable source of distress for patients [20], which may be compounded by the fact that they are not able to distinguish postoperative symptoms." expected,” from those of concern, without education on the part of health care providers.
In this study, patients who received improved feedback had a more rapid reduction in anxiety scores, with a cumulative decrease of 0.36 points over 9 days. This reduction in anxiety is clinically significant, given that the intervention is risk-free and its marginal cost is close to zero [21]. That finding suggests that real-time interpretation and guidance of postoperative symptoms provides important and timely reassurance to patients during a stressful time.
It should be noted that on day 1 before the intervention, patients in the enhanced feedback arm had a slightly higher level of anxiety. Given the randomized design of this study, this difference was due to chance; Likewise, it was controlled by using ANCOVA, which takes into account the relationship between the baseline score and the follow-up score.
This study has several limitations.
1. First, it was carried out in a single healthcare center, which is a well-equipped urban hospital, with highly protocolized care and advanced computing capabilities. The patients in this study were likely more computer literate and of a higher socioeconomic status than may be seen in other hospital systems. For example, the portal’s adoption rate among its patients is close to 90%, which is higher than national averages [22-24]. Access to digital platforms can also be challenging for patients in the face of stress due to illness and treatment, or pre-existing health conditions [25-27]. However, given the ubiquity of smartphones, and the increasing availability and utilization of patient portals, minimal training is required to use enhanced feedback.
2. Second, the study was conducted in an outpatient center where patients are generally healthier and less likely to require urgent care visits or readmissions. Overall, symptom burden was modest for this population, and most patients recovered well at home with minimal involvement of the clinical team or caregivers. Future work on symptom reporting and improved feedback will be applied to sicker patients undergoing more complex surgeries, where higher frequencies of complications have potential for greater beneficial effect.
3. Third, only consultations to the CAU of the MSK and not to the community emergency departments could be captured. Although the MSK CAU readmission and visit rates therefore represent an underestimate of the overall need for acute care for these patients, there is no reason to expect the rate of outpatient visits to differ between randomization arms. Additionally, this study was empowered to detect a 50% relative risk reduction in CAU consultations. In retrospect, equivalence between the two arms was perhaps the most likely outcome, but at the time of study design, the impact of RT alone was not yet known.
Finally, we were unable to randomize patients to a “no monitoring” arm, which may reflect the current status of most surgery programs, and may limit the generalizability of these results. However, PROs are rapidly becoming available and integrated into electronic health record systems [28-30], so these findings may encourage adoption, and inform the design of maximally effective systems in the near future.
In conclusion , it was found that providing patients with automated feedback on the severity of symptoms during recovery, from an ambulatory surgery center, reduces anxiety and nursing workload, without affecting CAU consultation rates or of readmissions, compared to care provider monitoring of electronically reported symptoms.
The symptoms captured by RT are not unique to outpatient surgery, and therefore the concept of providing real-time normative feedback to patients may also be useful after more complex inpatient surgeries, as well as treatments such as radiotherapy or chemotherapy, where patients need to manage their symptoms at home.
As such, consideration should be given to incorporating automated feedback more broadly into systems for routine monitoring of PROs. This innovative, patient-centered approach cost-effectively increases the positive impact of PRO measurement in surgical care.
