Outcomes of a Telephone-Based Questionnaire for Follow-up of Patients Who Have Completed Curative-Intent Treatment for Oral Cancers

This diagnostic study uses data from patients treated for oral cancer at a tertiary care cancer center in India to investigate the clinical value of a telephone-based questionnaire in detecting cancer recurrence during follow-up evaluation.

Key Points

Question

What is the role of a symptom-based telephone questionnaire in detecting recurrences in patients with radically treated oral cancer?

Findings

In this diagnostic cohort study of 400 patients with oral cancer, a symptom-based questionnaire was found to have a sensitivity of 90% and a negative predictive value of 99% for detecting cancer recurrences.

Meaning

This questionnaire may help in identifying patients at higher risk of cancer recurrence; furthermore, the questionnaire may help avoid overcrowding in clinics, enhance proper utilization of health care resources, and deter unnecessary clinic visits for patients.

Abstract

Importance

Follow-up assessment of patients who had treatment for head and neck cancer is critical and an important part of the overall treatment program. During each visit to the hospital, patients are evaluated by a physician and may undergo additional tests. Because it has been observed that the symptoms mentioned by patients often guide the treating clinicians in identifying cancer recurrence, an appropriately constructed questionnaire can help clinicians determine which patients need further testing as a result of a recurrence and which patients can continue to be monitored remotely.

Objective

To evaluate the role of a symptom-based telephone questionnaire in detecting recurrences in patients with radically treated oral cancer.

Design, Setting, and Participants

This prospective diagnostic cohort study was conducted from October 1, 2018, to February 28, 2019. This study took place at Tata Memorial Centre, an apex referral cancer center in India. A total of 615 consecutive patients with oral cancer were screened, of whom 400 patients consented to be a part of the study. After completion of curative treatment, these patients were followed up for 2 months to 2 years. We excluded patients younger than 18 years or older than 80 years, those with Eastern Cooperative Oncology Group status greater than 2, and those who had already been diagnosed with recurrent disease.

Interventions/Exposures

The patients were contacted 2 weeks before their clinic appointment, and a telephone interview was conducted using a predefined questionnaire. Based on patients’ responses to these questions, their disease status was estimated. Subsequently, during the actual follow-up visit, a trained head and neck surgeon, who was blinded to the questionnaire result, examined them.

Main Outcomes and Measures

The sensitivity and specificity of the telephone questionnaire were calculated by comparing its results with those of the trained head and neck surgeon after the clinical evaluation.

Results

Of the 615 patients screened, 400 consented to be part of the study. Participants had a median (interquartile range) age of 49 (41-56) years, and 334 (83.5%) were men. Recurrence was noted in 20 patients (5.0%). The telephone-based questionnaire was found to have sensitivity of 90.0%, specificity of 74.2%, positive predictive value of 15.5%, and negative predictive value of 99.3%. Clinical examination values were 100.0%, 92.9%, 42.5%, and 100.0%, respectively. A total of 124 patients (31.0%) said that they would prefer such a telephone follow-up compared with an actual physical follow-up visit.

Conclusions and Relevance

This diagnostic cohort study found that a symptom-based telephone questionnaire had good sensitivity and negative predictive value for detecting recurrences in patients with oral cancer on follow-up evaluation after completion of definitive treatment.

Introduction

Oral cancers are No. 11 on the list of most common cancers in the world.¹ Surgery is the preferred modality for treating patients with oral cancer; this is often followed by adjuvant therapy based on the histopathology report. Because of the risk of recurrences as well as the higher incidence of second primary cancers in this group, close follow-up after treatment is recommended.^2,3 Because there are many patients completing treatment over a period of time, follow-up assessment of such patients forms the major workload in the outpatient department. At Tata Memorial Centre in Mumbai, India, patients travel from great distances to complete the follow-up examinations as per the schedule, which increases economic burden. Many patients are daily wage workers and do not have ample resources to make frequent long-distance trips. We have noticed that all patients visiting our center have mobile devices for communication, irrespective of their financial status, and are easily accessible by telephone. We have also observed that many patients themselves identify new symptoms and notify the examining physician of these symptoms during follow-up visits or in telephone conversation. Some of these symptoms enable the physician to detect a new primary or a recurrent disease. Several existing studies have found that symptoms mentioned by patients help in diagnosing the majority of cancer recurrences (66%-80%).^4,5

The purpose of close follow-up is to identify a new disease or recurrence in its early stage so that appropriate curative treatment can be initiated as soon as possible. However, it has been observed that even when patients are followed up intensively, salvage therapy remains feasible in only 20% to 65% of patients.^{5,6,7,8,9,10,11,12} Moreover, data are scarce regarding the exact proportion of asymptomatic patients who are found to have recurrence on routine clinical examination. Therefore, we conducted this study to gather data related to asymptomatic recurrences and evaluate the role of a symptom-based telephone questionnaire in detecting cancer recurrence in patients who have been radically treated for oral cancer. We also assessed the patient’s acceptance of a less intensive telephone-based follow-up. This study holds special relevance during situations such as the coronavirus disease 2019 (COVID-19) pandemic, when social distancing is important and a telephone consultation with a validated questionnaire would help patients achieve close follow-up without unnecessary contact and exposure.

Methods

This prospective diagnostic cohort study was conducted among 400 eligible patients at Tata Memorial Centre, a busy tertiary care cancer center, from October 1, 2018, to February 28, 2019. We included patients with oral cancer who, after completion of curative treatment, had been monitored for more than 2 months and less than 2 years. We excluded patients younger than 18 years or older than 80 years, those with Eastern Cooperative Oncology Group status greater than 2, and those who had already been diagnosed with recurrent disease. Patients assessed once for the study were not considered during their subsequent visits. The study was approved by the center’s institutional ethics committee and was registered with the Clinical Trials Registry of India (No. CTRI/2017/08/009358). This study followed the Standards for Reporting of Diagnostic Accuracy (STARD) reporting guideline.

A questionnaire was developed in consultation with senior oncologists from the department of head and neck oncology. It included common questions routinely asked during the outpatient follow-up visit. These questions were related to symptoms suggestive of recurrence and also functional incapacitation during the posttreatment phase. The questionnaire was developed in 3 languages: English, Hindi, and Marathi (eAppendix in the Supplement).

The primary objective of the study was to detect the sensitivity of a nurse-led telephone questionnaire in identifying cancer recurrence compared with an office-based clinical evaluation by a trained head and neck surgeon. The secondary objectives were to estimate the proportion of asymptomatic patients with recurrent disease on routine follow-up, to detect the rate of salvage surgery for patients found to have cancer recurrence, and to evaluate patient acceptance of a telephone-based questionnaire administered every 3 months and a physical examination by the physician every 6 months.

We used hospital records to obtain the file numbers, appointment dates, and contact information of the patients coming for scheduled follow-up 15 days before their actual visit. They were contacted via telephone by a nurse 10 to 14 days before their appointment. The nurse conducting the interview was initially trained with a few mock telephone conversations and was conversant in English, Hindi, and Marathi (regional language). The initial 15 calls made by the nurse were supervised by 1 of the investigators (A.M.). During each call, the nurse explained the purpose of the study, and verbal informed consent was obtained. The participant’s language preference was requested, and the subsequent interview was conducted in that language. They were then asked questions as per the questionnaire. At the end of the conversation, the participants were instructed to come for their routine follow-up visit as per the schedule provided to them or at least within 3 weeks from the date of conversation. A minimum of 5 minutes was given to each participant to answer the questionnaire. Based on the responses to the questionnaire, an a priori decision was made to suspect recurrence if any of the questions numbered 4, 5, 9, or 10 were answered positively or if the answers to more than one of questions 1, 2, 3, 8, or 11 were suggestive of any problem. Although any of the responses in the latter group could also suggest recurrence, recurrence was suspected only if a positive answer was given for more than 1 question. This is because these findings can also result from functional problems after radical oral cancer treatment. This was decided in consultation with the senior head and neck oncologists in the department at the time of protocol development.

During the actual physical visit at the outpatient department, written informed consent was obtained. The patients were asked about their acceptance of and preference for the new model of follow-up vs the traditional 3-month face-to-face examination, and their responses were recorded in the case reporting form. They were also asked how far they had to travel for the follow-up visit and how much such a visit affected them financially. Thereafter, the patients were examined by a head and neck surgeon (who was unaware of the outcome of the telephone conversation) as they would have been at their normal visit. Depending on complaints and clinical findings, if recurrence was suspected, a biopsy, fine-needle aspiration cytology, or imaging (computed tomography, magnetic resonance imaging, or positron emission tomography and computed tomography) was performed. Thereafter, the case was discussed in a joint clinic where surgical, medical, and radiation oncologists were present and a decision was made by consensus. In the case of a recurrence, the site of recurrence, salvageability, and treatment modality used were recorded. Patients who had a telephone interview but did not have the follow-up outpatient department visit within 3 weeks of conducting the telephone interview were counted as missing for the follow-up, and it was decided to exclude these patients from the study.

Sample Size

The primary analysis was performed to calculate the CI around the estimates of sensitivity and specificity. The CI was based on the normal approximation

s ^± z_1−∝/2 √s ^(1 − s ^)/n,

where s is the estimate of sensitivity and specificity, and n is either the number of recurrences (when calculating sensitivity) or the number without recurrence when calculating specificity. The sample size calculation is based on the margin of error with an estimated proportion of 0.5 since this will give the maximum margin of error, m. Therefore, the sample size for a specified margin of error is

n = (z_1−∝/2)²s ^(1 − s ^)/m²,

where z_1−∝/2 is the 1 – α /2 quantile from the cumulative normal distribution. A study by Kothari et al¹³ estimates a 10% suspicion of recurrence in routine follow-up of patients with head and neck cancer. As we included all stages of oral cancer in different follow-up periods ranging from 2 months to 2 years after surgery, we adopted a 10% asymptomatic recurrence rate for calculating the sample size. The sensitivity and specificity margins expected were 0.094 and 0.031, respectively. If the recurrence rate turned out to be 5%, then the expected sensitivity and specificity margins were 0.132 and 0.03, respectively. Sensitivity of the telephone-based questionnaire was assessed by comparing its outcome with the result of the actual physical visit. Confidence intervals were calculated by normal approximation.

Statistical Analysis

Statistical analyses were conducted from April 15, 2019, to May 30, 2019, using SPSS version 21 software (IBM Corp). Secondary analysis was performed to examine covariates that may be predictive of high sensitivity of the questionnaire among the patients. Covariates that were examined were stage, subsite, age, sex, socioeconomic status, education level, and completion of adjuvant therapy. Simple descriptive statistics were used to estimate proportions of asymptomatic recurrences and salvage rates.

Results

Based on the inclusion and exclusion criteria as per the approved protocol, we identified 615 patients for screening in the initial round. Of these, 400 patients consented to participate in the study. Participants had a median (interquartile range) age of 49 (41-56) years, and 334 (83.5%) were men. All patients who consented to the telephone questionnaire arrived for their hospital appointment within a 3-week time frame. Patient demographic characteristics, cancer site, and treatment-related details are given in Table 1. All patients had been on a regular 3-month follow-up schedule.

Table 1. Distribution of Various Clinical, Pathological, and Social Factors in the Cohort .

Parameter	No. (%)a
Total No.	400
Age, median (IQR), y	49 (41-56)
≤60	336 (84.0)
>60	64 (16.0)
Sex
Male	334 (83.5)
Female	66 (16.5)
Education
<Secondary school	150 (37.5)
>Secondary school	250 (62.5)
Subsite
Tongue	175 (43.8)
Buccal mucosa, alveolus, or RMT	212 (53.0)
Otherb	13 (3.3)
Stage
Early	123 (30.8)
Advanced	277 (69.3)
pT stage
T1	74 (18.5)
T2	146 (36.5)
T3	65 (16.3)
T4a	115 (28.7)
pN stage
N0	232 (58.0)
N+	168 (42.0)
Adjuvant therapy
Adjuvant RT	205 (51.2)
Adjuvant CTRT	143 (35.8)
Follow-up preference
Every 3 mo	276 (69.0)
Telephone questionnaire and every 6 mo	124 (31.0)
Acceptability
Yes	385 (96.3)
No	11 (2.8)
Maybe	4 (1.0)
Financial implications of present visit
Not at all	9 (2.3)
Little bit	96 (24.0)
Somewhat	196 (49.0)
Very much	99 (24.8)
Expenditure, median (IQR), ₹	5000 (2000-7000)
Duration of travel, median (IQR), h	28.5 (20-36)

Abbreviations: CTRT, chemoradiotherapy; IQR, interquartile range; pN, pathological nodal stage; pT, pathological tumor; RMT, retromolar trigone; RT, radiotherapy.

^aValues are expressed as No. (%) unless otherwise specified.

^bOther subsites consist of the floor of the mouth, lip, and palate.

Based on the response to the questionnaire, recurrence or new disease was suspected in 111 patients (27.8%). When patients arrived for their physical examination, the examining physician suspected recurrence in 47 patients (11.8%) and ordered further evaluation. On further investigation, recurrence was confirmed in 20 patients (5.0%). Of these 20 patients, the telephone questionnaire identified recurrences in 18 of them (90%) correctly.

The telephone questionnaire was found to have a sensitivity of 90.0%, specificity of 75.5%, positive predictive value (PPV) of 15.5%, and negative predictive value (NPV) of 99.3%. In comparison, the direct clinical examination had a sensitivity of 100.0%, specificity of 92.9%, PPV of 42.5%, and NPV of 100% (Table 2). Of all the recurrences diagnosed, 7 (35.0%) were local, 6 (30.0%) were regional, 6 (30.0%) were distant, and 1 (5.0%) had a second primary cancer. Only 3 (15.0%) of these recurrences could be salvaged. The telephone questionnaire identified 2 of the 3 (66.7%) salvageable recurrences. All 3 of these recurrences required surgical treatment. Fourteen recurrences (70.0%) required palliative chemotherapy, 1 (5.0%) required palliative radiotherapy, and 2 (10.0%) were deemed fit for best supportive care.

Table 2. Results of the Telephone Questionnaire With Respect to Clinical Examination .

Examination type	%
	Sensitivity	Specificity	Positive predictive value	Negative predictive value
Telephone questionnaire	90.0	74.2	15.5	99.3
Clinical examination	100.0	92.9	42.5	100.0

In this study, we noticed that patients traveled a median (interquartile range) of 28.5 (20-36) hours to reach the hospital for the scheduled follow-up visit. While the Indian government allows free travel by train for patients with cancer in order to visit hospitals, a patient typically spends approximately ₹5000 ($72.60 US dollars) during each visit for travel-related expenditures. On being asked to grade the financial implications of their current visit, approximately half (196 [49.0%]) of the patients responded by saying that it affected them somewhat, whereas for one-quarter of the patients (99 [24.8%]), the visit caused significant financial hardship. Although the questionnaire-based follow-up was acceptable to 385 (96.3%) patients, only one-third (124 [31.0%]) preferred this modality of follow-up (consisting of a telephone questionnaire conducted every 3 months and a physical visit every 6 months). However, these are the patients who were willing to visit the hospital for regular follow-up and physical examination by the treating physician. Approximately 30% of the patients in our hospital do not follow the strict 3-month schedule, and these patients probably opted out of the present study.

We further analyzed the sensitivity and specificity of the questionnaire after stratifying the cohort by various covariates. These values are shown in Table 3. Sensitivity and specificity of the questionnaire were better among patients aged 60 years and younger (sensitivity, 94.4% vs 50.0% for patients >60 years) and those with educational attainment greater than secondary school (91.6% vs 87.5% for patients with less than a secondary school education). The questionnaire was more sensitive among women compared with men (100.0% vs 88.2%), but less specific (63.4% vs 77.9%). Sensitivity and specificity were better for those with tongue (100.0% and 73.4%, respectively) and early-stage tumors (100.0% and 83.4%, respectively). Sensitivity and specificity for those receiving adjuvant therapy were 90.9% and 74.3%, respectively, for radiotherapy and 87.5% and 77.0%, respectively, for chemoradiotherapy. When stratified by the site of recurrence, sensitivity was highest for those with regional recurrence and for those with a secondary primary cancer site (both 100.0%). In the current study, local recurrences had a sensitivity of 85.7%, and distant metastasis had a sensitivity of 83.3%.

Table 3. Sensitivity and Specificity of the Telephone Questionnaire Stratified by Various Covariates .

Parameter	%
	Sensitivity	Specificity
Age, y
≤60	94.4	75.7
>60	50.0	74.1
Sex
Male	88.2	77.9
Female	100.0	63.4
Educational attainment
<Secondary school	87.5	69.0
>Secondary school	91.6	79.4
Subsite
Tongue	100.0	73.4
Buccal mucosa, alveolus, or RMT	71.4	77.0
Othera	NA	76.9
Stage
Early	100.0	83.4
Advanced	88.8	71.8
Adjuvant therapy
None	100.0	74.0
RT	90.9	74.3
CTRT	87.5	77.0
Site of recurrence
Local	85.7	NA
Regional	100.0	NA
Distant	83.3	NA
Second primary	100.0	NA

Abbreviations: CTRT, chemoradiotherapy; NA, not available; RMT, retromolar trigone; RT, radiotherapy.

^aOther subsites consist of the floor of the mouth, lip, and palate.

Discussion

In any oncology practice, follow-up assessment of patients who have completed their treatment is important. In fact, this posttreatment monitoring constitutes the main workload in the outpatient department. Studies have shown that when patients with head and neck cancer are followed up for 5 years, local and regional recurrence may be seen in 25% to 50% of the patients,^6,14 and second primary cancer sites can develop in 3% to 5% of patients.¹⁵ To detect these cases, patients should be kept on regular surveillance.^2,3 The patients need to be assessed not only for oncologic control but also for their functional status.

Symptoms mentioned by patients help physicians in diagnosing the majority of recurrences (66%-80%).^4,5 Studies have noted no change in disease-free survival or overall survival between the groups in which recurrence was detected by the patient vs the physician. In one study evaluating 3645 patient visits, it was found that cancer recurrence or a second primary cancer was suspected in 5% of the patients. A total of 79% of the patients suspected to have a recurrence had themselves identified new symptoms. In the absence of any symptoms, recurrence was rare and seen in only 1.2% of patients.⁵ Another study found that cancer recurrence was suspected in only 10% of patients seen routinely and in 68% of those who requested a consultation. Results showed that 56% of those who requested a consultation actually had a recurrence, whereas recurrence was found in only 0.3% of asymptomatic patients.¹³ According to another study, the number needed to treat to detect 1 asymptomatic recurrence was 99.⁶ All these studies demonstrate that patients with recurrences often present with symptoms, which lead to the identification of such recurrences.

In countries like India, people have to travel long distances to reach specialized cancer hospitals. In the current study, the median duration of travel roughly translates to 3 working days. Thus, the indirect costs for these trips are substantial. Three-fourths of the patients felt that the trip affected them financially. Such factors prompted us to consider whether an alternative to an in-person evaluation could be found in order to maintain follow-up. We decided to explore the option of having a symptom-based telephone questionnaire, which would help in identifying patients with recurrences. We compared the sensitivity of this questionnaire against actual physical follow-up visits. These in-person examinations can be considered the criterion standard, as it is the accepted practice currently.

Because we do not want to miss any potential cancer recurrences, it is important that a telephone questionnaire has a high sensitivity. Our study results suggest that the questionnaire has a sensitivity comparable to that of a clinical examination. In a large percentage of patients (72.2%), the questionnaire ruled out cancer recurrence, which was validated in the subsequent clinical evaluation. However, the long-term performance of the questionnaire needs to be further assessed. The specificity was also reasonable, which clearly demonstrates the utility of the questionnaire in detecting recurrences and also in avoiding unnecessary clinic visits, thereby reducing the pressure on the limited health care resources and also reducing the financial burden of the patients. However, the questionnaire missed 10% of patients with recurrence who were asymptomatic for recurrent disease, highlighting a limitation of this approach.

Any health care intervention can be successful only if it is accepted by the patients in the community. In this study, a majority of patients found the questionnaire to be acceptable. In spite of the traditional mindset of seeing the doctor face to face and also being aware that this was an experimental surveillance option, one-third of the patients mentioned that they would prefer this form of surveillance over a physical, in-person follow-up examination. These numbers may actually improve with further refining of the questionnaire and feedback from the patients.

It is generally observed that the salvage rate after radical treatment is poor owing to unexpected extent of the disease at the time of its detection. In our study, results suggest that the questionnaire could identify 90% of recurrences, and 2 of the 3 salvaged cases were identified in the telephone questionnaire.

As sensitivity of the questionnaire can be affected by various factors, we looked at the effect of various covariates on the questionnaire’s outcome. We found that the sensitivity was higher for a younger population of patients, which could be the result of their ease of conversing and communicating over the telephone. Higher sensitivity was seen among women, but the specificity was relatively lower compared with men. The questionnaire also performed better among those with higher educational attainment. Over one-third of the patients had been educated to the secondary school level. Among them, the sensitivity of the questionnaire was found to be high. Among various cancer subsites, any changes in the tongue are quite easily observed. Probably for this reason, patients with tongue lesions had a higher sensitivity but lower specificity as compared with other sites. Sensitivity and specificity were somewhat lower for those with advanced disease. This may be because these patients had major resections, reconstruction, and adjuvant therapy, which left them with persistent, nonspecific symptoms; this could cause difficulty for these patients in identifying or feeling any new symptoms in their mouth or neck.

Approximately 87% of patients received some form of adjuvant therapy. Although sensitivity was highest among those who did not receive any adjuvant therapy, among those receiving adjuvant chemoradiation, a sensitivity of 87.5% was seen. It is also important to evaluate whether the questionnaire was sensitive enough for different types of recurrences so that it could correctly identify regional recurrence as well as a second primary cancer site in all patients.

The results of this study suggest that a telephone-based questionnaire has potential utility in identifying recurrence in patients treated for oral cancer. It has shown a reasonable sensitivity for older and less educated patients and even among those with advanced disease and those having received adjuvant therapy.

Especially during times when an infectious disease affects a wide geographic area, such as during the COVID-19 pandemic, lockdown measures including social distancing and quarantine make a telephone-based questionnaire of particular value in triaging patients who otherwise would need to come to the clinic for an in-person consultation. It is also important to note that our telephone questionnaire was conducted by a trained nurse. This model can significantly reduce the pressure on specialist oncologists, who are limited in numbers, especially in places where they are grossly outnumbered by patients.

Strengths and Limitations

This prospective diagnostic cohort study has the strengths of a robust sample size and good representation of the various covariates in the population. It also assesses the acceptability of the questionnaire among the cohort. However, the study has a few limitations as well. Sometimes the patients were not available at the telephone number provided, and repeated attempts at contacting them were required. The functional assessment was difficult to carry out with a telephone-based questionnaire. The questionnaire may be further refined by adding more specific questions to evaluate the functional speech and swallowing status of the patient. Although it was not possible for the present study, in the future, with wider availability of smartphones and the internet, a video consultation may be of great help. In this study, we followed our institutional protocol and considered imaging only in cases of any clinical suspicion. However, routine imaging may be added for the surveillance of such patients, and this may help to detect additional patients with recurrences in early stages. For some patients, their speech was not adequately intelligible over the phone. We overcame this difficulty by speaking with the patient’s carer or partner with their consent. Although none of the patients in this study lived alone, this possibility could make implementing the telephone consultation difficult. The telephone consultation may also miss asymptomatic disease recurrences. In addition, this study was performed at a single point in time. Hence, a longitudinal evaluation of the questionnaire may be required for assessing its long-term performance.

Conclusions

The results of this study suggest that the proposed telephone-based questionnaire has a high sensitivity and NPV in detecting recurrences in patients treated for oral cancer. The questionnaire was useful and well received among patients irrespective of education, age, stage of the disease, and adjuvant therapy received. Such a questionnaire may help in decongesting the clinics, thereby allowing for the proper use of health care resources and avoiding unnecessary clinic visits for patients.

Supplement.

eAppendix. Telephone-Based Questionnaire