Reduced Returns to ED With Post‐Tonsillectomy and Adenoidectomy Chat Program: Outcomes and Costs

ABSTRACT

Objective

Patients who participate in the post‐surgical chat programs will have lower rates of emergency department (ED) visits when compared to those who do not participate.

Methods

Data from a texting program was used to stratify patients into those participating versus those not participating with the program. Individual patients' charts were reviewed for demographic information, if a postoperative ED visit related to surgery occurred, and if the ED visit was caused by a failure of patient education regarding the treatment plan or by expected postoperative symptoms. Odds ratios compared the ED visit rates between participating and non‐participating families. Cost was analyzed by comparing published data on ED visit costs with internal data on chat program cost.

Results

During the study (2/2022–3/2023), 1059 unique patients (children/caregiver dyads) were enrolled in the chat. Five hundred and fifty‐eight patients (52.7%) participated in one or more chats while 501 patients (47.3%) did not participate. Of note, patients with public insurance were less likely to participate. Non‐participants were 9.12 times more likely to visit the ED postoperatively due to a communication lapse (p = 0.047). Estimated savings from 1 year of avoided ED visits did not outweigh the chat program's expense.

Conclusions

Utilization of an automated messaging system for frequently asked questions after pediatric tonsillectomy demonstrated a statistically significant reduction in ED visits and readmissions related to postoperative complications and questions, but an increase in health care costs.

Level of Evidence

3

Data from a post‐surgical texting program was used to stratify patients into those participating versus those not participating in the program. Non‐participants were 9.12 times more likely to visit the ED postoperatively due to a communication lapse (p = 0.0466). Estimated savings from one year of avoided ED visits did not outweigh the chat program's expense.

1. Introduction

Post‐surgical readmission rates serve as a major quality indicator for hospitals. Also, value‐based health insurance care may decline reimbursement for post‐surgical readmissions, resulting in financial losses. Reducing readmissions is of paramount importance for hospitals and healthcare systems.

Pediatric otolaryngology surgery readmission rates have remained low at 2.3% nationally [1]. However, rates of return to the Emergency Department (ED) after tonsillectomy and adenoidectomy are somewhat higher at 4.85% [2]. With ~289,000 tonsillectomies performed annually on children under 15 [3], this would amount to about 14,000 patients returning to the ED annually. Previous studies show up to half of these visits may be attributed to patient education failures in which caregivers have a misunderstanding of the expected postoperative course [4].

The optimal modes for patient education have long been debated. Many practices have focused on postoperative instruction pamphlets to reinforce information discussed in the office. However, surveys have shown parents regard direct conversations with the surgeon, pediatrician, and otolaryngology nurse more highly than informational packets [5]. Additionally, while handouts have been shown to significantly improve recall, recall rates remain low with almost 50% of the information provided forgotten [6]. One study showed that despite expressing satisfaction with the information provided, 38% of parents were unable to correctly state the nature of the surgery their child was having [7]. Similarly, parents have reported appreciation for the addition of figures and diagrams to text‐only handouts. However, data collection did not show a decrease in the volume of parent calls to the nurse's line after switching to visually oriented handouts [8].

The National Institute of Health recommends limiting the amount of information given at any one time to encourage understanding and retention [9]. However, in practice, the pediatric otolaryngologist only has a few opportunities (the clinic visit, the preoperative check‐in, and the postoperative update) to educate the family on what to expect postoperatively. This does not allow ample opportunity to convey the complexities of postoperative care, especially as the family is likely at a heightened state of alert on the day of surgery itself.

Our institution has implemented a post‐tonsillectomy chat program to help fill this communication gap. This program aims to provide pertinent postoperative information for families at a time when they are ready to engage with it and need to use it. While some patients present postoperatively to the ED for unrelated reasons or unavoidable reasons, others present because they did not know what to expect or how to manage the postoperative recovery. This latter group can be considered failures of patient education.

This study compares rates of ED visits due to failures of patient education between participators and non‐participators in the post‐tonsillectomy chat program. The authors hypothesize ED visits due to failures of patient education will be less frequent among families who engage with the post‐tonsillectomy chats.

2. Materials and Methods

As a retrospective chart review study, this research was deemed exempt from the appropriate institutional review board approval based on the applicable federal regulation.

The pediatric otolaryngology service in a tertiary care children's hospital launched a postoperative tonsillectomy chat program in partnership with Conversa. Beginning February 2022, legal guardians of all patients undergoing tonsillectomy and adenoidectomy who gave permission to receive texts were enrolled in the chat program to access postoperative instructions and report symptoms to the clinical team. Texts were sent directly to the legal guardian, and did not require the Conversa app to be downloaded on the guardian's phone. Message prompts were created by the hospital's otolaryngology team without the use of any artificial intelligence. Conversa tracks data on chat participation with respect to engagement and responses selected. Institutional review board exemption was obtained before data collection began. Data was extracted from the Conversa system on patients enrolled in the system between February 2022 and February 2023.

The chat program begins sending automated messages one day before surgery and continues on postoperative days 1, 3, 6, 10, 14, and 30. Participants also can start an unscheduled session by reentering the chat without receiving an automated message. The automated messages offer a variety of options for families, including reviewing discharge instructions, what to expect postoperatively, a medication dosing guide, recovery tips, and reporting symptoms. The prompts on what to expect provide additional information on common temporary postoperative symptoms including bad breath, congestion, sore throat, nausea/vomiting, voice changes, fever, and snoring. Recovery tips provide additional information on topics including managing pain, resuming activity, adequate fluid intake, and what to eat. The symptom reporting tool includes poor appetite, nausea, vomiting, pain, fever, and bleeding from the mouth. The chat engages participants to provide further detail on symptom severity. Depending on symptom severity, automated messages provide reassurance or encourage contacting clinical staff (Figures 1 and 2).

FIGURE 1.

Infographic shows symptoms about which parents may select additional information. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

FIGURE 2.

Infographic shows an example of recovery tips related to hydration. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

Individual patient charts were reviewed to ascertain whether the patient had presented to the ED in the 30 days following surgery, and whether the reason for this ED visit was related to the surgery. ED visit notes were reviewed to identify whether the ED return was prompted by a failure of patient education regarding the treatment plan or expectations of postoperative symptoms. Failures of patient education were defined as lack of understanding of expected postoperative symptoms and lack of understanding of how to administer pain medications. Patients who contacted the clinical staff or were directed by the chat to present to the ED were not considered patient education failures. If patients presented to the ED for persistent issues despite following postoperative recommendations, this was not considered an education failure.

Patients were then divided into two groups: those who participated in the chat program (interacted with more than one set of the chat prompts) and those who did not participate. These groups were defined by participation rather than enrollment to best represent engagement with the chat program. Demographics were compared between groups to assess for any pre‐existing selection bias. The number of ED return visits for communication failures between these two groups was then compared using a Fisher's exact test to generate an odds ratio and p value.

3. Results

Between February 2022 and February 2023, 1059 unique patients were enrolled in the post‐tonsillectomy chat. Five hundred and fifty‐eight patients (52.7%) participated in one or more chats, while 501 patients (47.3%) did not participate. There were no significant differences between participators and non‐participators with respect to patient age, race, or primary language. Families of patients with public insurance were less likely to participate compared with families of patients with private insurance (Table 1).

TABLE 1.

Insurance, language, and race demographics between participators and non‐participators.

		Total	Non‐Participator	Participator	p
Insurance	Public	452 (43.3%)	254 (46.5%)	198 (39.8%)	0.03
	Private	592 (56.7%)	292 (53.5%)	300 (60.2%)
Primary language	Not english	48 (4.6%)	30 (5.5%)	18 (3.6%)	0.19
	English	1000 (95.4%)	518 (94.5%)	482 (96.4%)
Race	Black	224 (21.4%)	138 (25.2%)	106 (21.2%)	0.66
	Latino	125 (11.9%)	63 (11.5%)	62 (12.4%)
	Asian	14 (1.3%)	10 (1.8%)	4 (0.8%)
	White	618 (59.0%)	318 (58.0%)	300 (60.0%)
	Other	47 (4.5%)	19 (3.5%)	28 (5.6%)

Abbreviation: POC, person of color.

Fifty‐four patients (5.1%) returned to the ED postoperatively for reasons related to their surgery. Primary ED diagnoses for revisit included pain/dehydration (44%), bleeding (42.6%), fever (1.9%) and nausea/vomiting (1.9%). No presentations due to bleeding were determined to be caused by patient education failures. All patients presenting for pain/dehydration that were determined to be patient education failures were non‐participators in the chat program (Table 2).

TABLE 2.

Reasons for ED visits divided by participators versus non‐participators and stratified by the presence of patient education failure.

	Education failure	Total	Non‐participator	Participator
Pain/Dehydration	No	24	7 (29.2%)	17 (70.8%)
	Yes	5	5 (100%)	0
Bleeding	No	23	7 (30.4%)	16 (69.6%)
	Yes	0	0	0
Fever	No	0	0	0
	Yes	1	1 (100%)	0
Nausea/Vomiting	No	0	0	0
	Yes	1	0	1 (100%)

Of the 54 patients (5.1%) who visited the ED postoperatively for reasons related to their surgery, seven (13.0%) were determined to be the result of a patient education failure. Patient education failures were closely divided between a lack of understanding expected postoperative symptoms and a lack of understanding how to administer pain medications (Table 3).

TABLE 3.

Types of patient education failures and frequency of occurrence.

Type of patient education failure	Number of patients
Did not understand expected postoperative symptoms	3 (42.9%)
Did not understand how to give pain meds	4 (57.1%)

When ED visits were stratified based on chat participation, 27 of the patients who presented to the ED (50%) had participated in the chat program, and 27 of the patients (50%) were non‐participators. However, when ED visits due to patient education failures were stratified by chat participation, seven of the patients who presented to the ED as a result of a patient education failure (87.5%) were non‐participators, while only one patient (12.5%) was a participator (Figures 3 and 4).

FIGURE 3.

Stacked bar chart compares the proportions of patients with ED visits related to patient education failures, patients with ED visits unrelated to patient education failures, and patients without ED visits between chat participants and non‐participants. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

FIGURE 4.

Flow diagram showing the distributions of enrolled patients and the breakdown of ED revisits and their causes between participators and non‐participators. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

Fisher's exact test was performed, which generated an odds ratio of 9.10 with a p‐value of 0.0466. This denotes that non‐participants in the chat program were 9.12 times more likely to visit the ED for a patient education failure compared to participants. This finding was statistically significant (p‐value = 0.047). There were no statistically significant differences in the frequency of communication lapses based on race, primary language, or insurance type.

4. Discussion

Today's personal communication technology facilitates faster and easier bidirectional information transfer. One such technology, Conversa, triages postoperative questions and reinforces patient education after pediatric tonsillectomy with or without adenoidectomy. In this study, we demonstrate the positive impact an automated virtual care platform, such as Conversa, has in reducing ED visits post‐tonsillectomy among participants in this technology compared to non‐participants. We observed a notably higher rate of communication lapses (defined as a failure of patient education regarding the treatment plan or postoperative recovery symptoms leading to postoperative ED presentation) among families who declined to enroll in the chat program when compared to families who elected to participate. This suggests that the use of the Conversa chat system was successful in providing education to families regarding routine postoperative instructions and reducing the frequency of avoidable returns to the hospital for postoperative pain management. While patients who participated in the chat program may also have contacted the nurse line at the direction of the chat program, this was not considered a failure of the chat program but rather an opportunity to perform additional education and avoid a more costly and burdensome ED visit.

Previous studies examining parental recall of surgical risks and postoperative instructions, even after a formal discussion like informed consent, have shown poor recall rates [10, 11]. Parents of tonsillectomy patients who were quizzed on their postoperative knowledge answered only three‐quarters (76.3%) of questions with correct responses [12]. Printed handouts only marginally improved recall knowledge, and overall recall rates were still low [13]. Additionally, Camilleri et al. showed that, while 90% of patients think leaflets are a good idea, only 75% of patients read and used the leaflets provided [14]. In addition to possibly being lost or misplaced, these printed handouts require a certain baseline level of health literacy from the reader, and 28% of adult patients undergoing otolaryngologic surgery have low health literacy [15]. Multiple studies have demonstrated paper resources do not improve recall rates to an acceptable level [16, 17, 18, 19, 20, 21, 22].

Newton and Sulman conducted a text messaging pilot study among families with a child undergoing tonsillectomy [23]. The investigators postulated that providing small amounts of information at pertinent time points during the patient's postoperative recovery would be more useful than a handout provided at the time of surgery [23]. The authors' follow‐up study of 85 families showed that no families who received texts visited the ED postoperatively, and only 25% of these families called the nurses' line postoperatively [24].

In 2020, a prospective study pilot‐tested the text messaging system, Tonsil‐Text‐To‐Me [25]. Among seven parents of children undergoing tonsillectomy with or without adenoidectomy, the parents reported reduced anxiety and increased enthusiasm and engagement in their children's postoperative care [25]. Conversa represents a similar tool and resource. These messaging systems are capable of not only providing real‐time feedback but also delivering smaller amounts of relevant information at a time, which may be easier for patients and families to digest.

The perioperative period can be a time of heightened stress and anxiety for children and families of children undergoing pediatric surgery. Although families routinely undergo preoperative counseling and may receive additional printed resources and instructions, information recall after surgery remains low. Misunderstanding postoperative instructions and expectations can lead to avoidable ED visits and readmissions, especially for those with limited knowledge on how to access their health care teams. This burden can be reduced with the implementation of automated messaging systems, such as Conversa.

Avoiding unnecessary ED visits is important for hospital quality metrics and for patient satisfaction. However, programs like Conversa are not without cost. The platform costs our institution about $50,000 annually. With an approximate volume of 5000 patients annually, this comes to an average cost of $10 per patient. Considering approximately half of the families in our data sample participated in the program, costs could be estimated at $20 per participant. In our study sample of more than 1000 patients, we concluded six ED visits were avoided with the use of the chat program. Extrapolating that to a pool of 5000 patients, 30 ED visits would be avoided over the course of an entire year. According to a 2022 study by Tran et al., the average ED visit cost is $1420 [26]. The costs for 30 ED visits would be $42,600, which is $7400 less than the annual cost for the messaging program. While these stated costs are not specific to our institution, they are meant to place this technology into a cost–benefit framework. Costs and benefits at each individual institution will vary based on the size of the patient population impacted, the negotiated cost of the technology platform, and the cost of ED visits. There is benefit to a family avoiding the hassle and inconvenience of another ED visit is significant; the cost to the hospital system may not show a financial return on investment, especially in smaller centers where the number of tonsillectomies performed is even lower. With increased hospital costs and tight margins, individual centers will have to decide if this is an investment they can afford to make.

It also is important to consider the public policy implications of chat programs when all patients may not have equal access to advanced communication technologies. This study showed a lower rate of participation in the chat program for patients with public insurance. It may be that these patients do not have access to cell phones with unlimited data plans, thus resulting in a decreased participation rate. If chat programs do decrease ED utilization, as our data suggest, how they can be funded to allow equitable access may need to be addressed above the level of the individual hospital or health system.

This is a retrospective review of a single institutional experience from a metropolitan‐area pediatric medical center. As such, the data presented herein may not be representative of other pediatric health care facilities. Furthermore, the inherent subjectivity of chart review to identify communication lapses may further limit the validity of our conclusions. However, while this study is limited to a single tertiary care hospital, we are part of a 5‐state two million patient health system with a long electronic health record utilization experience. Our next steps could involve leveraging our 18 pediatric otolaryngologists across 6 operative locations to better understand the opportunities and limitations of this important educational and patient care technology moving forward.

The impact of demographics and socioeconomic status (SES) on our outcomes remains unclear. No significant differences in ED returns were observed when comparing factors such as age, race, and preferred language between chat participators and non‐participators. This may be attributed in part to the small sample size. However, a notable finding was that those enrolled in public health insurance were found to have significantly lower rates of chat participation using Conversa. Further evaluation is necessary to provide insight on the implications of SES on avoidable ED visits and readmissions after pediatric otolaryngologic surgery.

Overall, only about half of families enrolled in the chat chose to participate. Reasons for this low participation rate may be due to SES, demographic, and/or technological variances, and perhaps social determinants of health including race, English‐language comprehension, or technological access. Future study could explore these as possible causes. Comparisons also need to be made between Conversa, other automated messaging systems or similarly developed tools, and other communication modalities such as phone calls and written discharge instructions. Multi‐institutional assessments of the Conversa chat system will also help to assess the viability of this program in other areas with different language and cultural milieus.

In the current state, messaging through electronic medical record (EMR), phone calls, and even secure testing is very labor‐intensive for the pediatric otolaryngology clinical staff. As artificial intelligence (AI) becomes more prominent, there may be utility in using this technology to help triage phone calls, screen incoming messages, and draft responses. This may make other communication and education modalities more advantageous than they have previously been.

5. Conclusion

Patient comprehension and communication are crucial in the perioperative period. An automated messaging system for frequently asked questions after pediatric tonsillectomy accomplished a reduction in ED visits and readmissions related to failures of patient education. However, program cost outweighed the savings generated by avoided ED visits. Improved patient safety and reduced rates of avoidable ED visits are expected with the use of this technology, but the impact on health care costs associated with a new technology must be weighed against the magnitude of benefit delivered.

Conflicts of Interest

The authors declare no conflicts of interest.

Xu V. and Aaronson N. L., “Reduced Returns to ED With Post‐Tonsillectomy and Adenoidectomy Chat Program: Outcomes and Costs,” The Laryngoscope 135, no. 9 (2025): 3389–3396, 10.1002/lary.32195.