Audit of Patient Communications Following Transoral Robotic Surgery

ABSTRACT

Objective

To characterize patient‐initiated communications after transoral robotic surgery (TORS) for human papillomavirus positive (HPV+) oropharyngeal squamous cell carcinoma (OPSCC) and assess associations with 30‐day emergency department (ED) presentation and hospital readmission.

Methods

Retrospective review of adults undergoing TORS at a single tertiary center (July 2014–October 2021). Patient demographics, tumor characteristics, postoperative outcomes, and communications within 30 days were assessed. Logistic regression assessed associations with 30‐day ED presentation or readmission.

Results

Among 249 patients (85.1% male; mean age 60.1 ± 9.9 years), 179 patients (71.8%) initiated communications, for a total of 448 individual communications (2.5 ± 1.8 per patient). Non‐medical concerns accounted for 58.0%, with medical concerns accounting for 42.0%. Fifty‐seven patients (22.9%) presented to the ED within 30 days; patients with ≥ 1 communication had higher ED presentation rates (27.9% vs. 10.0%, p = 0.002). On multivariate logistic regression, ≥ 1 communication (OR 2.98, 95% CI 1.12–7.87, p = 0.03) and longer duration from discharge to first postoperative appointment (OR 1.09, 95% CI 1.01–1.17, p = 0.03) remained significantly associated with 30‐day ED presentation. Notably, triaging providers recommended hospital presentation for only 16.0% of medical complaints.

Conclusions

Patient‐initiated communications following TORS are common and mostly for logistical reasons. Communication frequency and time to first postoperative visit were predictive of 30‐day ED presentation, suggesting that adjusting time to first postoperative appointment, enhanced discharge education, and early postoperative check‐ins may be tangible methods to reduce unplanned healthcare utilization.

Key Points

• •Patient communication after transoral robotic surgery is predominantly logistical in nature, and even medical concerns infrequently require ED presentation. However, number of complaints and time to first follow‐up visit are risk factors for ED presentation in this patient population.
• •This study provides a novel characterization and analysis of patient communications following transoral robotic surgery, and identifies potential areas for improvement in discharge education and triage.

1. Introduction

Transoral robotic surgery (TORS) remains the standard surgical approach for patients with human papillomavirus–positive (HPV+) oropharyngeal squamous cell carcinoma (OPSCC). The typical inpatient stay following TORS is brief, with most patients discharged after 1–2 nights of observation. Nevertheless, postoperative complications such as oropharyngeal hemorrhage, dysphagia, and pain are relatively common and may require intervention, underscoring the importance of the early postoperative period [1, 2].

Patients generally receive both preoperative counseling and postoperative discharge instructions to guide their recovery. However, prior studies have demonstrated that patients often have limited comprehension and recall of these instructions, contributing to confusion, anxiety, and unnecessary emergency department (ED) visits or hospital readmissions [3, 4, 5]. In contemporary practice, patients frequently communicate their postoperative concerns via telephone or electronic messaging. These interactions provide the surgical team with valuable information to triage symptoms and optimize patient care. At the same time, the volume of such communications can place a significant burden on surgeons and clinical staff, highlighting an area for potential improvement.

Patient‐initiated communication represents a unique and underexplored variable that may yield important insights into postoperative concerns. To date, very few studies within head and neck surgery have examined this [6, 7, 8, 9]. A better understanding of the frequency, content, and patterns of these patient‐initiated communications could inform strategies to improve preoperative counseling, refine discharge instructions, and enhance triage protocols. Such interventions may reduce unnecessary hospital utilization while simultaneously alleviating the workload on otolaryngology teams. Accordingly, the objective of this study was to characterize patterns of patient‐initiated communication following TORS for HPV + OPSCC in relation to ED presentation and hospital readmission.

2. Methods

2.1. Population

We conducted a retrospective chart review of patients aged ≥ 18 years who underwent transoral robotic surgery (TORS) for oropharyngeal squamous cell carcinoma (OPSCC) at the University of Pittsburgh Medical Center (UPMC) between July 2014 and October 2021.

2.2. Data Collection and Patient Communications

Patient demographics, tumor staging, operative reports, hospital course, and discharge information were extracted from the electronic medical record (EMR). Postoperative outcomes included ED visits and/or hospital readmissions within 30 days of surgery.

Patient‐initiated communications (phone calls or EMR messages) that were documented in the EMR and occurred within 30 days postoperatively were reviewed. Per institutional protocol, patient‐initiated communications are triaged by clinic nurses or otolaryngology residents. Each communication was assigned a primary reason for communication (i.e., dysphagia). For communications in which the patient had multiple complaints, the most pressing complaint was selected as the primary reason. Each communication was assessed for temporal association with a subsequent ED visit or readmission.

Communications were broadly categorized into non‐medical and medical concerns. Non‐medical communications included logistical requests (e.g., work forms, referrals, scheduling), medication questions or refills, pathology or laboratory requests, and or clarifications. Medical communications included oropharyngeal bleeding, pain, neck swelling, dysphagia, Jackson‐Pratt (JP) drain/tube/line issues, wound dehiscence, muscle stiffness, or systemic symptoms (i.e. fever, tachycardia, hypotension, or constipation).

2.3. Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 29.0.2 (IBM Corp., Armonk, NY, USA). Descriptive statistics summarized the study cohort. Univariate and multivariate logistic regression was used to evaluate associations between communications and 30‐day ED presentation or hospital readmission. Statistical significance was defined as α < 0.05. Figures were generated using GraphPad Prism version 10 (Boston, Massachusetts USA).

3. Results

3.1. Demographics

A total of 249 patients who underwent TORS for OPSCC from July 2014 to October 2021 were included (85.1% male, 95.1% Caucasian, 60.1 ± 9.9 years old). The majority had a history of smoking (62.2%). Average T category was 1.4 ± 0.6 and N category was 1.1 ± 0.8. Average hospital stay following surgery was 3.2 ± 2.3 days. Average time from discharge to first clinic visit was 7.5 ± 4.2 days. When comparing patients who had ≥ 1 communication versus those who had none, there were no statistically significant differences in clinical or demographical variables (Table 1).

Table 1.

Demographics.

Variable	Total (n = 249)	Communication (n = 179)	No communication (n = 70)	p‐value
Sex (male)	212 (85.1%)	149 (83.2%)	63 (90.0%)	0.18
Age (years)	60.1 ± 9.9	59.5 ± 9.7	61.4 ± 10.4	0.16
Race (Caucasian)	237 (95.1%)	168 (93.8%)	69 (98.6%)	0.12
Smoking history	155 (62.2%)	107 (59.8%)	48 (68.5%)	0.20
Alcohol history	170 (68.3%)	125 (69.8%)	45 (64.2%)	0.40
Length of stay (days)	3.2 ± 2.3	3.2 ± 2.2	3.3 ± 2.8	0.70
Duration from discharge to first follow‐up (days)	7.5 ± 4.2	7.6 ± 4.2	7.5 ± 4.3	0.83
T category	1.4 ± 0.6	1.5 ± 0.7	1.2 ± 0.6	0.81
N category	1.1 ± 0.8	1.0 ± 0.7	1.1 ± 0.2	0.75
Emergency department presentation	57 (22.9%)	50 (27.9%)	7 (10.0%)	0.002
Hospital readmission	40 (16.1%)	29 (16.2%)	11 (15.7%)	0.10

3.2. Reasons for Communication

Of the 249 patients included, 179 patients (71.8%) reached out to the department within 30 days postoperatively for a total of 448 communications (2.5 ± 1.8 per patient, range 1–10). The average time from discharge to first communication was 5.4 ± 6.2 days. Of the 448 communications, non‐medical complaints made up the majority (n = 260, 58.0%). Of the 260 non‐medical communications, the most common reasons for communication were logistical (i.e., appointment timing, n = 124, 47.7%), medication questions or refills (n = 97, 37.3%), or chart results (i.e., questions about Epic notes or lab results, n = 39, 15.0%). Communications related to medical complaints were less common (n = 188, 42.0%). Of these 188, the most common reasons for communication included pain (n = 54, 28.7%), dysphagia (n = 44, 23.4%), systemic symptoms (i.e., fever, n = 34, 18.1%), neck swelling (n = 17, 9.0%), JP drain concerns (n = 16, 8.5%), oropharyngeal bleeding (n = 15, 8.0%), and neck wound dehiscence (n = 2, 1.1%). Reasons for communication are shown in Figure 1.

Figure 1.

Pie charts demonstrating (A) overall reasons for communication, (B) communication reasons for non‐medical concerns, and (C) communication reasons for medical concerns.

3.3. 30‐Day ED Presentations

Of the 249 patients, 57 presented to the ED within 30 days of surgery, with 13 patients presenting multiple times for a total of 74 ED visits. The average time from discharge to first ED presentation was 5.6 ± 3.8 days. Of the 74 total ED visits, the most common reason for ED presentation was oropharyngeal bleed (n = 27, 36.5%), followed by systemic symptoms (i.e., fever/chills) (n = 18, 24.3%), dysphagia (n = 11, 14.9%), JP drain issues (n = 8, 10.8%), pain (n = 6, 8.1%), neck swelling (n = 3, 4.1%), and wound dehiscence (n = 1, 1.4%) (Figure 2).

Figure 2.

Pie charts demonstrating reasons for (A) 30‐day ED presentation and (B) 30‐day hospital readmission.

Patients who had ≥ 1 postoperative communication demonstrated a higher rate of 30‐day ED presentation (27.9% vs. 10.0%, p = 0.002) (Table 1). Those who presented to the ED had a higher average number of communications (2.5 ± 1.9 vs. 1.6 ± 1.8, p = 0.001). On univariate logistic regression, having a postoperative communication (odds ratio [OR] 3.49, 95% confidence interval [CI] 1.49–8.13, p = 0.004) as well as longer time from discharge to first follow‐up (OR 1.08, 95% CI 1.01–1.15, p = 0.03) increased the chance of 30‐day ED presentation. On multivariate logistic regression, ≥ 1 communication (OR 2.98, 95% CI 1.12–7.87, p = 0.03) as well as a longer duration from discharge to first follow‐up (OR 1.09, 95% CI 1.01–1.17, p = 0.03) remained significantly associated with 30‐day ED presentation when controlling for gender, race, age at surgery, and tumor stage (Table 2).

Table 2.

Logistic regression to predict 30‐day ED presentation.

Variable	Univariate logistic regression		Multivariate logistic regression
	Odds ratio (95% CI)	p‐value	Odds ratio (95% CI)	p‐value
≥ 1 communication	3.49 (1.49–8.13)	0.004	2.98 (1.12–7.87)	0.03
Gender (male)	0.77 (0.34–1.70)	0.52	1.01 (0.42–2.40)	0.97
Race (Caucasian vs. other)	1.72 (0.94–4.30)	0.62	1.70 (0.48–6.30)	0.39
Age at surgery (y)	0.58 (0.16–1.98)	0.38	0.98 (0.94–1.01)	0.17
Time from discharge to first follow‐up	1.08 (1.01–1.15)	0.03	1.09 (1.01–1.17)	0.03
T category	0.87 (0.51–1.49)	0.61	1.15 (0.66–2.03)	0.62

Abbreviations: CI, confidence interval; ED, emergency department.

Of the 188 patient communications regarding medical concerns, only 30 patients (16.0%) were recommended by the healthcare provider to present to the ED for evaluation. When examining communications that led to ED presentation, of the 74 presentations, only 28 (37.8%) were predated by a relevant communication. Specifically (regardless of whether the triaging provider recommended ED presentation), 11/15 communications (73.3%) for oropharyngeal bleeding, 5/44 communications for dysphagia (11.4%), 4/54 communications for pain (7.4%), 2/34 communications for systemic symptoms (5.9%), 3/17 communications for neck swelling (17.6%), 2/16 communications for JP drain concerns (12.5%), and 1/2 communications for neck wound dehiscence (50.0%) resulted in ED presentation.

We elected to further examine dysphagia, given its relevance to TORS. Overall, there were 44 communications for difficulty swallowing and/or throat pain causing difficulty swallowing (9.8% of all communications, 23.4% of communications for medical concerns). However, of these 44, only 11 were recommended by the provider to present to the ED for further evaluation, with only five of these patients ultimately presenting to the ED. Overall, there were 11 patients who presented to the ED for dysphagia, of which three resulted in hospital admission.

Similarly, we further investigated ED presentations for pain, as this was the second most common reason for 30‐day ED presentation (n = 18, 24.3%) behind oropharyngeal bleed, and the most common medical reason for patient communication (n = 54, 12.1% of all communications, 28.7% of communications for medical concerns). Of the 54 communications for pain, only two patients (3.7%) were recommended by the assessing provider to present to the ED. Neither of these patients required hospital readmission. Overall, there were 18 presentations to the ED for pain, of which four were predated by a relevant communication. Only 1 (5.6%) of these presentations resulted in hospital readmission for pain management.

3.4. 30‐Day Hospital Readmission

Of the 249 patients, 40 patients required 30‐day hospital readmission, with some patients being admitted multiple times for a total of 52 readmissions. Patients who had at least one postoperative communication demonstrated a statistically similar chance of 30‐day hospital readmission (16.2%) compared to those who had no communications (15.7%, p = 0.10). Moreover, those who had at least 1 hospital readmission demonstrated a statistically similar number of communications (2.18 ± 1.93) than those who did not present to the ED (1.73 ± 1.84, p = 0.17). On univariate logistic regression, only time from discharge to first follow‐up was significantly associated with 30‐day hospital readmission (OR 1.12, 95% CI 1.04–1.21, p = 0.003). On multivariate logistic regression, this remained a significant predictor (OR 1.17, 95% CI 0.83–1.29, p = 0.01) when controlling for communications, gender, race, age at surgery, and tumor stage (Table 3).

Table 3.

Logistic regression to predict 30‐day hospital readmission.

Variable	Univariate logistic regression		Multivariate logistic regression
	Odds ratio (95% CI)	p‐value	Odds ratio (95% CI)	p‐value
≥ 1 Communication	2.03 (0.85–4.84)	0.11	1.71 (0.51–3.30)	0.33
Gender (male)	0.64 (0.27–1.53)	0.32	1.12 (0.46–3.10)	0.70
Race (Caucasian vs. other)	0.96 (0.21–4.32)	0.95	1.16 (0.23–5.80)	0.85
Age at surgery (y)	1.01 (0.97–1.05)	0.52	0.98 (0.95–1.03)	0.55
Time from discharge to first follow‐up	1.12 (1.04–1.21)	0.003	1.17 (0.83–1.29)	0.02
T category	0.73 (0.39–1.39)	0.35	1.23 (0.66–2.50)	0.45

Abbreviation: CI, confidence interval.

Of the 52 readmissions, the most common reason was oropharyngeal bleed (n = 26, 50.0%), followed by systemic symptoms (n = 9, 17.3%), dysphagia/poor PO intake (n = 8, 15.4%), pain management (n = 4, 7.7%), neck swelling (n = 3, 5.8%), and JP drain issues (n = 2, 3.8%) (Figure 2). Of the 52 readmissions, 17 (32.7%) were predated by a relevant communication.

4. Discussion

This study provides a novel characterization of postoperative patient‐initiated communication following TORS for OPSCC and its relationship to 30‐day ED presentation and hospital readmission. Nearly three‐quarters of patients reached out to the department within 30 days after surgery, with most communications pertaining to non‐medical issues such as logistical questions and medication refills. In fact, the majority (84%) of communications for medical concerns were effectively triaged without the need for ED presentation. However, having ≥ 1 communication (OR 2.98) as well as longer time to first postoperative appointment (OR 1.09) was associated with a higher chance of 30‐day ED presentation on multivariable analysis.

Patient‐initiated communications for medical concerns accounted for approximately 40% of all communications. Interestingly, only a minority of these patients (16.0%) were considered by the triaging provider to require ED presentation. Similarly, in a study by Prince et al., only 25% of patient communications after parotidectomy required action by the provider [6]. Dysphagia, a known complication after TORS, exemplifies this trend. In our cohort, 11 patients presented to the ED with dysphagia, but only 3 required hospital admission. Moreover, of the patient‐initiated communications for dysphagia (n = 44, 23.4% of medical‐related communications), only 11 (25.0%) were recommended by the triaging provider to seek evaluation in the ED. This suggests that providers were largely able to effectively triage these patients, minimizing unnecessary hospital visits. These data identify a potential opportunity for intervention. For example, implementing a protocol for early postoperative check‐ins could help address patient concerns promptly and identify those who would truly benefit from in‐person evaluation. Notably, the average time from discharge to first communication was 5.4 days, and the average time to first ED presentation was 5.6 days, indicating that early check‐in prior to these timepoints may be essential. In a study by Shah et al., implementation of postoperative check‐ins within 72 h after discharge decreased ED presentations for total laryngectomy patients [10].

The predominance of non‐medical inquiries underscores ongoing challenges in postoperative education and care coordination. Nearly 60% of communications were logistical or administrative in nature, consistent with prior literature showing that patients often experience uncertainty about follow‐up logistics, medication schedules, and expected recovery symptoms after head and neck surgery [11, 12]. Patients have been shown to often retain less than half of information discussed in preoperative counseling [13, 14]. Moreover, poor medical literacy can be a significant barrier to accurately interpreting or following postoperative instructions [15]. These data highlight the importance of enhancing the clarity and accessibility of discharge instructions.

Recent publications have demonstrated success with alternative methods of perioperative instructions, such as the use of animated videos [16] or ChatGPT to design more readable instructions [14]. Our findings show that over 50% of patient‐provider communications following TORS are for logistical reasons or asymptomatic concerns. Therefore, to effectively triage patient communications, we propose the development of a secure AI large language model that integrates post operative education materials and patient‐specific information accessible in the EMR or scheduler. This AI model could serve as an efficient first‐pass assessment of patient communications regarding upcoming appointments, work forms, referrals and potentially medication refill requests. Rapid and personalized responses could be generated for what we have demonstrated are the majority of patient communications, while those of greater concern are escalated to medical personnel. In fact, recent studies have shown that AI responses to patient questions are often higher quality and more empathetic when compared to physician responses [17, 18]. Additionally, patients may benefit from structured libraries of short videos created by medical professionals [19]. This allows for repeated viewing of discharge instructions, and addresses issues of incomplete information retention and variable reading level. The proposed AI model and video library could be hosted on a mobile application for ease of use. Overall, the goal of these materials is to empower patients to answer their own questions when possible and to otherwise reduce avoidable utilization of hospital resources.

Our findings also identify postoperative communications as a potential novel risk marker for postoperative outcomes. In this study, patients who initiated ≥ 1 postoperative communication had a higher rate of 30‐day ED presentation (OR 2.98) while controlling for time to follow‐up appointment, tumor stage, and demographic variables. However, communication reasons likely vary in their risk for ED presentation. Therefore, this observed relationship may be disproportionately influenced by high‐risk communication reasons, such as oropharyngeal bleeding. Importantly, the association between ≥ 1 postoperative communication and 30‐day ED presentation does not imply causality, as variables such as greater symptom burden could underscore both measures. Identifying predictors of postoperative complications is important to identify at‐risk patients. To date, no studies have identified number of postoperative communications as an independent predictor of 30‐day ED presentation. By recognizing postoperative communications as a potential risk factor for unplanned ED visits, clinicians may identify patients and intervene before escalation to the ED.

4.1. Limitations

The present study has several limitations. Its retrospective design is subject to documentation bias, as some patient interactions may not have been recorded in the EMR. Additionally, communication tone, urgency, and content granularity could not be fully captured through message logs alone. Variability in discharge paperwork could also not be evaluated, as the discharge paperwork is not typically saved in the EMR. Lastly, despite statistical adjustments for TNM category, unaccounted for differences in disease severity across patients may introduce outcome bias.

5. Conclusion

Patient‐initiated communications after TORS are common and mostly for logistical reasons. Here, we identify number of communications as a novel independent risk factor for 30‐day ED presentation. Moreover, the majority of medical‐related communications did not require ED presentation, identifying a potential area for improved triage to reduce unnecessary hospital utilization. Prospective studies assessing whether early postoperative check‐ins or enhanced discharge education can reduce unplanned hospital utilization are warranted.

Author Contributions

Study conception and design: Jonathan Goulazian, Micah Harris, and Shaum S. Sridharan. Literature review: Jonathan Goulazian and Micah Harris. Data collection: Jonathan Goulazian, Micah Harris, and Kelly Daniels. Statistical analysis and interpretation of results: Jonathan Goulazian and Micah Harris. Manuscript drafting: Jonathan Goulazian and Micah Harris. Study design and clinical oversight: Matthew E. Spector, Kevin J. Contrera, Joshua D. Smith, Seungwon Kim, Jessica Maxwell, and Shaum S. Sridharan. All authors reviewed the results and approved the final version of the manuscript.

Funding

The authors have no funding to report.

Ethics Statement

The research presented in this paper was approved by the Institutional Review Board of the University of Pittsburgh, School of Medicine.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.