+ Indicates censored data.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Deutschmann MW, Sykes KJ, Harbison J, Cabrera-Muffly C, Shnayder Y. The Impact of Compliance in Posttreatment Surveillance in Head and Neck Squamous Cell Carcinoma. JAMA Otolaryngol Head Neck Surg. 2015;141(6):519–525. doi:10.1001/jamaoto.2015.0643
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Posttreatment surveillance (PTS) is a key component in the treatment of patients with head and neck cancer. It is unclear how beneficial this is in improving patients’ survival.
To determine how compliance with follow-up affects clinical outcomes in patients with head and neck squamous cell carcinoma.
Design, Setting, and Participants
This was a retrospective cohort study at a tertiary academic center of a total of 332 patients with head and neck squamous cell carcinoma who had completed both treatment and follow-up at the University of Kansas Medical Center. Patient and tumor characteristics, socioeconomic status, and geographic data were collected.
Compliance with PTS.
Main Outcomes and Measures
The effect of compliance with PTS on overall survival.
Compliance with PTS, US Census tract income level, and the distance patients travel for follow-up had significant effects on survival (P = .001, P = .001, and P = .01, respectively). Cox proportional hazard models revealed that more advanced disease (hazard ratio [HR], 1.76 [95% CI, 1.21-2.58]; P = .003), middle (HR, 1.64 [95% CI, 1.13-2.39]; P = .009) and moderate (HR, 1.90 [95% CI, 1.18-3.06]; P = .008) census tract income level, and age (HR, 1.03 [95% CI, 1.01-1.04]; P < .001), were significantly associated with an increased risk of death. There was an association between compliance and tobacco cessation (P = .003), as well as the distance a patient lived from the medical center (P = .008).
Conclusions and Relevance
Patients with head and neck squamous cell carcinoma were significantly more likely to survive with completion of follow-up and tobacco cessation. Compliance with PTS was associated with smoking cessation and traveling less than 200 miles for follow-up.
Despite advances made in its diagnosis and treatment in recent years, head and neck squamous cell carcinoma (HNSCC) continues to have an unacceptably high morbidity and mortality rate. Although a combined modality approach (consisting of a combination of surgery, radiation, and/or chemotherapy) is often used for treatment of advanced cancers, the 5-year survival rate for HNSCC is still only slightly above 50%.1 This is often due to the high rate of persistent disease or recurrence. Quiz Ref IDMerkx et al2 found that 50% of recurrences and over 50% of second primary tumors were identified on routine follow-up visits. This emphasizes the importance of posttreatment surveillance (PTS) in patients with HNSCC.
Despite the importance of this aspect of cancer care, the frequency and total number of posttreatment surveillance visits are somewhat controversial. Some authors3,4 cite limited effectiveness of follow-up on improving prognostic outcomes either immediately following treatment or after 3 years in certain subsets of patients. However, reduced life expectancy was found when follow-up was withheld.5 The National Comprehensive Cancer Network guidelines recommend that follow-up should consist of visits at least every 1 to 3 months during the first year after treatment, every 2 to 4 months in the second year, every 4 to 6 months in the third to fifth years, and yearly after that.
The ability of patients to complete posttreatment surveillance may potentially be influenced by a host of factors, including travel distance to the treatment site, social support system, and socioeconomic status. Quiz Ref IDThe correlation between the distance to the treatment center and patient outcomes paradoxically showed higher survival rates in patients who had a greater the distance to travel.6 When compared with income level quintiles, survival was improved in the groups with higher income.7 In addition, tobacco cessation is suspected to affect survival in the posttreatment surveillance time period.8
The purpose of this study was to determine whether compliance with PTS contributes to survival. We hypothesized that patient adherence to a PTS regime would improve overall survival. We also wished to determine whether certain patient characteristics may play a role in compliance with PTS and, subsequently, with outcomes measures. This would then allow us to predict which patients would benefit from more aggressive posttreatment surveillance schedules.
Medical records from the University of Kansas Medical Center (KUMC) from March 2003 to December 2008 were reviewed for cases of HNSCC. A total of 398 patients who were diagnosed as having HNSCC were identified. Our inclusion criteria included any patient older than 18 years with a diagnosis of mucosal HNSCC who received treatment at KUMC. A diagnosis prior to January 1, 2007, was required to allow for sufficient length of follow-up and documentation of at least 1 posttreatment surveillance visit after completion of the treatment regimen. The study received approval from the Human Subjects Committee at the University of Kansas Medical Center.
Multiple data variables were collected, including patient characteristics, tumor characteristics, tobacco use and cessation, alcohol use, compliance with PTS, distance to KUMC from home, socioeconomic factors, and clinical outcomes. Tumor characteristics included tumor site and American Joint Committee on Cancer (AJCC) staging. Clinical outcome data included follow-up, recurrence status, and survival status. If survival data were not available in the medical record, it was obtained from the Kansas State Cancer Registry.
Patients’ tumors were staged according to the AJCC guidelines. Every patient who was currently smoking at the time of diagnosis and subsequent visits was counselled in smoking cessation. Distance to KUMC and US Census tract code income classification were based on the residence of the patient at diagnosis. Census tract code income classifications are made by the US Census based on the median family income in the census tract relative to the median family income within that geographical area.
Compliance with PTS was measured according to attendance at follow-up visits as recommended by the treating head and neck surgeon. At our institution, we define posttreatment surveillance as every 6 to 8 weeks for the first posttreatment year, every 2 to 3 months for the second posttreatment year, every 4 months for the third posttreatment year, every 6 months for the fourth posttreatment year, and yearly for all subsequent years.
Patients were given a 4-week grace period to allow for scheduling difficulties or illness. Patients were categorized as compliant if they attended all scheduled appointments within 4 weeks of the suggested timeline, partially compliant if they missed 1 or 2 appointments, or noncompliant if they missed 3 or more appointments or were completely lost to follow-up after the initial appointment.
To be included in the study, patients needed to be seen for at least 1 follow-up visit at the study institution (KUMC). Those patients subsequently seeking care at other institutions would be classified as noncompliant.
Relationships of categorical variables were summarized by frequencies and percentage and tested using χ2 or Fisher exact test, as appropriate. Quantitative relationships were summarized by means (SDs) and tested using 2-sample t tests.
Using SPSS statistical software (version 22; IBM Corp), Kaplan-Meier survival models and Cox proportional hazard models were created to model time to event status with contributing variables and a study period of 60 months. Both models allow for data contribution from all participating patients regardless of the amount of time that they spent in the study. Patients were censored when the designated follow-up time had not been fully met, but their data contributed to the model. Patients not yet having 60 months of follow-up but surviving at the time of completion were censored, allowing for the inclusion of their data. Overall survival was used for these models. Cox proportional hazard models compare survival using multiple variables to create proportional hazards of the event (in our case mortality).
A total of 332 patients met inclusion criteria. Mean (SD) overall follow-up time was 45 (30) months. The patient characteristics were typical for this patient population (Table 1). The oral cavity was the most common primary site (124 patients [37%]), followed by the larynx (95 [29%]) and oropharynx (94 [28%]). A total of 246 patients (74%) presented with advanced disease. Treatment modalities included surgery alone (65 patients [20%]), radiation therapy alone (11 [3%]), surgery plus adjuvant radiation therapy (65 [20%]), chemoradiation (105 [32%]), and surgery plus chemoradiation (86 [26%]). Most patients (213 [64%]) did not develop a recurrence.
Over half the patients (198) lived within 50 miles of the treatment center, while 22 (7%) lived more than 200 miles away (Table 2). Over half the patients (180 [54%]) lived in areas falling within the middle census tract income level. We categorized patients according to their compliance with the posttreatment surveillance appointments. Nearly half (163 [49%]) did not miss any appointments. Patients missing 1 or 2 appointments (67 [20%]) were considered partially compliant. The remaining 101 patients (30%) were considered to be noncompliant, having missed 3 or more appointments during the surveillance period.
Quiz Ref IDCompliance was not significantly related to treatment modality (P = .43), sex (P = .37), race (P = .68), tumor stage (P = .15), census tract income level (P = .10), or age group (P = .57). There was, however, a relationship between compliance and tobacco cessation (P = .003), as well as the distance a patient lived from the medical center (P = .008) (Table 3). Patients who quit smoking, lived in higher-income census tracts, and lived closer to medical center were more likely than expected to have kept all of their appointments.
Kaplan-Meier survival analyses were also performed. There was a significant impact on overall survival when patients were stratified by tumor stage, compliance with PTS, census tract income level, smoking status, and distance from home to KUMC (Figure) (P = .02, P = .001, P = .001, P = .01, and P = .02, respectively).
Cox proportional hazard models were constructed to explore overall survival differences in populations that may be caused by multiple factors (Table 4). For categorical variables, the resulting hazard ratios (HRs) or odds ratios (ORs) provide a measure of increased or decreased odds of death relative to reference categories within the defined variables. For continuous variables, the ORs refer to increases in hazard based on a single unit change in the variable (eg, age increased by 1 year).
The final multivariable model includes statistically significant changes in survival with advanced stage at presentation (HR, 1.76 [95% CI, 1.21-2.58]; P = .003), middle (HR, 1.64 [95% CI, 1.13-2.39]; P = .009) or moderate (HR, 1.90 [95% CI, 1.18-3.06]; P = .008) census tract income levels, age at diagnosis (HR, 1.03 [95% CI, 1.01-1.04]; P < .001), and compliance with PTS (Table 4). All other variables fell out of the model based on a lack of statistical significance (established a priori P ≤ .05). As expected, when controlling for all other variables in the model, patients with advanced-stage disease were 76% (95% CI, 1.21-2.58) more likely to die before 60 months than those with early-stage disease. For every year of increasing age at diagnosis, when controlling for the other variables there was a 3% (95% CI, 2%-5%) increase in the odds of death. Again controlling for other factors, individuals residing in census tracts designated as middle or moderate income were more likely to die than those living in upper income tracts (ORs, 1.64 [95% CI, 1.13-2.39]; P = .009, and 1.90 [95% CI, 1.18-3.06]; P = .008, respectively).
Posttreatment surveillance compliance as a whole significantly influences survival (P = .001) (Figure 1B), but Quiz Ref IDdifferences between the HRs of the levels of compliance were not statistically significant (P = .12) (Table 4). The ORs indicate that those who missed 1 or 2 appointments had the lowest odds of death by 60 months relative to those who missed no appointments, and those who missed 3 or more appointments were at increased odds of death.
Smoking cessation was initially part of the model; however, there was a significant association between smoking status and compliance with PTS seen with χ2 analysis (P = .003) (Table 3). When excluding smoking and including compliance in the model, and comparing it to the model with the exclusion of compliance with PTS and the inclusion of smoking, the former was more robust (Table 4).
Posttreatment surveillance is an important component in the treatment of patients with HNSCC. Compliance with this aspect of cancer care was found to be associated with several variables. Census tract income level does seem to influence compliance. Patients in the upper income level were more likely to remain compliant with their PTS, while those in the middle income levels were more likely than expected to miss appointments. Patients in the moderate and lowest income levels were found to be between the upper and middle groups in terms of their posttreatment surveillance compliance. It is important to note that only 9 patients were in the lowest income group, so it is difficult to draw any conclusions about this group. While patients in the upper income level would be expected to have high compliance rates, it is interesting that patients in the middle income level were more likely to miss appointments. This could be the result of a portion of patients in the middle income level not having the same socioeconomic status as the surrounding population in that area.
Distance from KUMC was an important factor in compliance with PTS. Given that some patients had to travel more than 200 miles for follow-up, it is not surprising that some were unable to be fully compliant. Patients living further away may have elected to miss follow-up appointments owing to the long distance or they may have elected to receive follow-up closer to home. This difference could not be determined based on the study design.
Quiz Ref IDPatients who were compliant with their PTS were significantly more likely to quit tobacco products, and those who quit had improved survival. This shows that surveillance can have an added benefit of tobacco cessation. Tobacco cessation was discussed at every follow-up visit, and the treating physician noted the patients’ current smoking status. A decrease in consumption was encouraged, and progress in reduction of use was praised. The harm of continued tobacco use, such as the increased risk of second primaries, was reiterated to patients who continued to consume tobacco products. Cessation was self-reported, however, and patients may have denied continued smoking in order to provide the physician with the desired response. As such, we do not currently use testing of any form that would determine, with certainty, tobacco cessation. We believe patients who attend follow-up appointments will increase their chances of quitting. Given that the relative risk of a second primary tumor was 2.75 times for patients who continued to smoke following treatment, this could have had a significant impact on survival.8
The second purpose of this study was to identify factors influencing survival in this patient population. This study has shown that compliance with PTS has clear correlations with survival, as discussed herein. As expected, continued tobacco use was found to be associated with survival. Compliance contributed significantly to the Cox hazard model for overall survival, but differences between the categories of compliance were not statistically significant. The interaction between smoking cessation and compliance may play a role in this finding.
Given the overall significant contribution to the model, the lack of statistically significant differences between the compliance categories could be an area for future research. Patients who were fully compliant seemed to have the worst survival in the first few critical years following treatment. This seemed to improve after 30 months, when survival for patients who had missed 3 or more appointments had decreased. The fact that compliance with PTS at the highest risk time period for recurrence did not improve survival is troubling; a beneficial effect is seen in later years of follow-up.
Living more than 200 miles from KUMC was found to be negatively associated with overall survival. Surprisingly, the opposite result has been found in another study6 examining treatment distance and its effect on survival. Logically, one would assume that previous studies would have obtained results similar to ours in this respect. Lack of access to medical services, as well as the travel time required for follow-up visits, may explain why a greater distance from KUMC was associated with worse survival.
Census tract income level was significantly associated with overall survival in both Kaplan-Meier analyses and Cox proportional hazard models. Residing in an upper income census tract provides a protective effect as it relates to 5-year survival in this patient population. Those patients living in tracts classified as middle or moderate income follow the expected stratification according to their mean survival. The small sample size of the lowest income census tract may influence our findings, but they were found to have the highest mean survival time among the groups.
Thus far, we have established that patients should be considered as high risk if they have incomplete PTS, continue to use tobacco after treatment, live in middle or moderate income census tracts, or live more than 200 miles from the treatment center. Some studies5,9 have found that compliance with PTS does not correlate with improved survival. However, given the benefits of PTS discussed herein, including smoking cessation and early identification of recurrence, compliance seems to provide tangible benefits to patients. Patients with the higher-risk characteristics identified herein may benefit from a more aggressive PTS schedule.
What type of surveillance schedule would benefit these high-risk patients? One could argue that decreasing the time between surveillance visits by half or adding an additional visit in each year of follow-up would be appropriate. However, increasing the number of visits can cause further issues. Almost half of the patients had to travel more than 100 miles for follow-up, and it is also clear that survival is influenced by the distance patients travel to KUMC. Requiring patients from distant locations to have more frequent follow-up visits places additional stress on those patients, including more travel time and increased cost. Closer follow-up should then be determined on an individual basis in high-risk patients. The factors found in our model—advanced stage, age, and income tract level—could be used to advise patients on the most appropriate follow-up schedule.
Finally, how do we increase compliance in PTS? Emphasizing to patients the importance of surveillance on survival and early detection of recurrence is crucial. Maintaining good communication with the patient, such as ensuring that any missed appointments are rescheduled in a timely fashion, is also beneficial. Smoking cessation should be discussed, but in a productive manner that does not discourage patients from attending future follow-up visits. Unfortunately, there are few other incentives available to improve compliance in unmotivated patients.
There are some limitations to this study. This was a retrospective collection of data, which contributes to a lack of complete data for all patients. For example, we were unable to determine the reason for loss of follow-up among some patients. Patients who lived a long distance from KUMC may have received follow-up closer to home, electing to not make the long trip to the treating institution. However, we were unable to differentiate those patients who received local follow-up from those who truly just missed follow-up appointments. Despite the large population of patients studied, several subgroups were small (tumor site, use of chewing tobacco), which limits the power of some of our findings.
Another limitation is that we did not have the human papillomavirus (HPV) status of patients with oropharyngeal squamous cell carcinoma (SCC) in our cohort. Unfortunately, routine testing was not done during that time period. HPV-positive patients have a better prognosis and generally higher socioeconomic status then the HPV-negative patients, but we were unable to determine the impact of HPV status on the population of patients with oropharyngeal SCC in our study.10
The nature of the study introduces a selection bias that is impossible to control. Given the multiple variables influencing survival, it is not surprising when these variables are associated with one another. However, our hazard model shows that age, compliance with PTS, advanced stage, and census tract income levels all independently affect survival in this patient cohort. We believe this patient cohort is a typical representation of the population with head and neck cancer, both in demographics and compliance, and the results are widely applicable to this general population. We made every possible effort to remove any selection bias by including the full sample of patients who met our inclusion and exclusion criteria.
Patients had improved overall survival with full compliance of PTS as well as tobacco cessation. Compliance levels were significantly associated with smoking cessation and patients traveling greater than 200 miles for follow-up. High-risk patients included those who had not attended all follow-up appointments, continued tobacco use, lived in middle or moderate census tract income levels, or lived more than 200 miles from KUMC.
Corresponding Author: Michael W. Deutschmann, MD, FRCSC, Department of Otolaryngology–Head and Neck Surgery, University of Kansas Medical Center, 3901 Rainbow Blvd, MS 3010, Kansas City, KS 66160 (firstname.lastname@example.org).
Submitted for Publication: December 30, 2014; final revision received March 19, 2015; accepted March 19, 2015.
Published Online: May 7, 2015. doi:10.1001/jamaoto.2015.0643.
Author Contributions: Dr Deutschmann had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sykes, Shnayder.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Deutschmann, Sykes, Harbison, Cabrera-Muffly.
Critical revision of the manuscript for important intellectual content: Deutschmann, Sykes, Cabrera-Muffly, Shnayder.
Statistical analysis: Deutschmann, Sykes, Harbison, Cabrera-Muffly.
Administrative, technical, or material support: Deutschmann, Sykes, Harbison, Cabrera-Muffly.
Study supervision: Shnayder.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the joint American Head and Neck Society and International Federation of Head & Neck Oncologic Societies Meeting; July 26-30, 2014; New York, New York.