Is suicide a significant contributor to mortality in head and neck cancer - A surveillance, epidemiology, and end results database study

INTRODUCTION

A diagnosis of cancer is a major stressor event for the patient and family, uncontrolled pain and physical symptoms may lead to suicidal thoughts. The psychological distress, feeling of lack of autonomy, independence/hope, feeling of helplessness, impending morbidity associated with cancer and its treatment, associated economic crisis, familial outlook, decreased quality of life, and depression (psychosocial factors) are factors that play a crucial role in such suicide ideation and completion. Age, gender, treatment type, and stage of cancers (individual factors) may also play a predominant role in determining the suicidal risk. Cumulatively, these factors increase the risk of suicide among People Living With Cancer (PLWC) as compared to the general population.[1] It has been reported that about 17.7% of American PLWC harbor suicidal ideation. Affective illness, alcoholism, and depression are some of the important determinants of suicide even in the physically healthy population.[2,3] Apart from cancer-specific morbidities, psychosocial and psychosomatic effects of cancer, organic mental syndromes, and/or exacerbation of pre-existing psychopathologic abnormalities are also reported to predispose to suicides or suicidal tendencies.[2-4]

There are reports of higher incidence of suicide among mid-facial, head and neck cancers (HNC, including laryngeal and thyroid tumors), as compared to the general population and PLWC.[5-11] Suicide incidence is noted to be very high for larynx and to some extent thyroid cancers and the prevalence influenced by the treatment and staging. The concepts of treatment/staging are constantly evolving with time and repeatedly undergo periodic changes. Irrespective of the evolving concepts of treatment and staging, HNC suicides are reported to be increasing in North America.[5,7] The influence of grade of HNC at diagnosis on survival outcome also has not been studied. Furthermore, there are very few reports addressing suicides in non-laryngeal, non-thyroidal HNC in the recent literature.[5,7] Reports of survival patterns between grades of tumors, number of malignant tumors of non-laryngeal, and non-thyroidal HNC are very few.

There is a recent report of depressive factors influencing survival among patients with HNC.[12] Tongue cancer is emerging as a cause of concern among Indians, and there is no reliable data on the extent of suicides and depression among this subset of patients.[13,14] Lessons from existing large scale data would help Indian oral cancer and mental health-care personnel to address the issue of oral, HNC survival, and suicides effectively. Hence, using a large, publicly available American database, an attempt is made to address the lacunae and describe the survival pattern in non-laryngeal, non-thyroidal HNC with emphasis on suicide.

MATERIALS AND METHODS

RESULTS

Individual depersonalized data collected from the SEER-18 registry from 1973 to 2014 as of November-2016 and released in April-2017 (inclusive of adjustments due to Hurricane Katrina) and linked to all states and counties were repeatedly run to collect maximum data. The entire study population with HNC, as defined in the material and method section was classified as that is alive, committed suicide, dead (not first tumor), dead due to HNC, and dead due to other conditions.

In all, data of 218,048 USA residents in the SEER database with HNC (having substantial details) were collated. Of these, 774 people committed suicide during 1973–2014, 81779 (37.5%) were alive, 63262 (29%) dead due to HNC, 28120 (12.9%) dead in not first tumor category, and 44113 (20.2%) dead due to other causes. The average survival period was 62.83 ± 73.87 (95% Confidence interval = 62.52–63.14) months, median of 34 months and the range was 0–503 months, with an interquartile range of 79 (11–90) months. Of all HNC cases, tongue was the most common site (25.6%, n = 55878) followed by gum and other mouth (14.7%, n = 32086). Majority of the cases were squamous cell carcinoma (SCC, 83%, n = 181000).

The demographics of age group, gender race, insurance subscription, site of HNC, grade of HNC, year of occurrence, and marital state at the time of diagnosis were compared with the present status of the patients. The distribution of the same is given in Table 1. All the parameters considered were statistically significant differences between the patient survival outcomes (P = 0.000).

An attempt was made to compare the survival period in terms of the patient outcome. After excluding the 2009–14 periods for reasons mentioned earlier, it was observed that 154,743 HNC patients formed the study group, out of which survival details of 830 patients were not available. Of the remaining 153,913 HNC patients, 79,514 died within 5 years. It included 416 (0.5%) suicides, 17,092 (21.5%) death (not first tumor), 43,299 (54.5%) death due to HNC, and 18,707 (23.5%) death due to other causes. The difference in other predictors is shown in Table 2.

The mean survival period in the 5-year (60 months, 1974–2008) period did not change over the time phases since 1974. [Figure 1] Mean 5-year survival in cases of suicide, was progressively decreasing with time frame while death due to other causes occurred at a relatively less time in 2003–08 as compared to 1973–78. Of interest is the nearly stable survival, by volume in death due to HNC in the entire span of time 1973–2008 but presented with a difference when expressed as a percentage in 5-year survival [Figure 2].

Close

Figure 1

Mean 5-year survival in head and neck cancer patients in different time phases.

Export to PPT

Close

Figure 2

Number of head and neck cancer and 5-year survival in different time phases.

Export to PPT

Kruskal–Wallis test revealed that pooled (overall) and 5-year survival period differed between the 5-outcome groups. The survival period varied significantly between the 5-groups (P = 0.000). Death due to HNC occurred at a relatively lower mean months in pooled and 5-year survival analysis. This was closely followed by death in not first tumor group trailed by suicide and death due to other causes. The difference was statistically significant (P = 0.000). ANOVA test showed that death due to “not first oral tumor” had highest mean number of tumors (2.55 ± 0.93) followed by patients who committed suicide had a mean (1.45 ± 0.79) of malignant tumors while those who were alive had 1.31 ± 0.67 tumors, dead due to HNC had 1.31 ± 0.67, and those dead due to other causes 1.39 ± 0.63. The difference was statistically significant (P = 0.000) [Table 3].

Kaplan–Meier survival statistics were performed to estimate the 5-year (60 months) survival characteristics of the study population in the time frame of 1973–2008. The endpoint was demise due to any cause which was stratified based on survival outcome, as described earlier. A log-rank test was run to determine if there were differences in the survival distribution for the different outcomes in death (suicide, death due to HNC, death due to other causes, and not primary tumor) as strata and grade of tumors as a factor. The survival distribution was statistically significant difference, χ²(2) = 99.192, P < 0.0005. [Figure 3]. The characteristics are given in Table 4. In the case of suicides, it was observed that about 550% of cases succumbed before the 30 months. Anaplastic group (Group IV) had a rapid drop in the slope in the graph as compared to other grades of HNC. In death due to HNC and not a first tumor, there was no remarkable difference between the grades of the tumor while the slope of the graph was uniform for all tumor grades in the group of patients that succumbed due to other causes.

Close

Figure 3

Kaplan–Meier curve for 5-year survival outcome of head and neck cancer as compared by the grades of tumor in the study group. (a) Death due to head and neck cancer (b) Death due to - not first tumor (c) death due to other causes (d) death due to suicides.

Export to PPT

Multiple regression analysis was done to predict 5-year survival (in months) of all HNC in 1973–2008, both years inclusive from race, side laterality, tumor grade, marital status at diagnosis, age group, gender, and site among all HNC patient. There was linearity as assessed by partial regression plots and a plot of studentized residuals against the predicted values. Independence of residuals was seen with Durbin Watson statistics (0.1590). There was homoscedasticity, as assessed by visual inspection of a plot of studentized residuals versus unstandardized predicted values. There was no evidence of multicollinearity, as assessed by tolerance values >0.1. There were no studentized deleted residuals ≥±3 standard deviations, no leverage values ≥0.2, and values for Cook’s distance >1. The assumption of normality was met (Q-Q Plot). The multiple regression model statistically significant predicted survival, F (11, 18424) = 203.818, P < 0.0005, adjusted R² = 0.12. All variables added statistically significant to the reduction, P ≤ 0.05. Regression coefficients and standard errors can be found in Table 5.

DISCUSSION

The purpose of this study was to describe the suicidal and survival characteristics of HNC patients. Previously, such an attempt was made using data till 2011.[5] Knowledge and insight would help policy framers to formulate guidelines for maxillofacial oncology team to sensitize, educate, and identify the PLWC who might have suicidal ideation. Suicide among Indian PLWC is poorly documented, and there has been a report of increasing trends of tongue cancer, especially in non-habit, elderly women.[13,14] Hence, knowledge from this would help Indians to frame better suicide screening among PLWC.

An attempt was made to decipher the influence of grade of HNC at diagnosis on survival outcome of non-laryngeal, non-thyroidal HNC. An attempt was also made to correlate the grades of tumors, number of malignant tumors of non-laryngeal, and non-thyroidal HNC with survival outcomes as defined earlier.

Relatively more percentage of Caucasians (93% of all suicides) committed suicide as compared to 78.7% due to HNC, 7% of Afro-Americans committed suicide, and 14% succumbed to HNC. Males (91.7%) were more likely to commit suicide. This is consistent with other studies and trends too.[5,7] Proportionally, more suicide was associated with moderately differentiated HNC (31.2–36.5%). About 80% of the study population lay in between 45 and 84 years of age. It was observed that a significant number of suicides in HNC occurs in the initial 6 months of diagnosis. Nearly 55% of all HNC suicides occurred within the first 3 years of diagnosis. Tongue and tonsil are the most common sites involved in HNC suicide while the site is often an unpaired HNCs (75.9%). SCC is the most predominant HNC type. All the findings are consistent with earlier reported literature from several parts of the world.[2,5,10]

Suicides represents a small proportion of HNC patient deaths in this sample (0.4%). It has been reported up to 17.7% of all patients with cancer have suicidal ideation. The 0.4% suicide is only a reflection of completion of suicide and is not a measure of suicide ideation or attempts. Persistence of risk of suicide, even after 60 months of initial diagnosis highlights the need for continuous monitoring in this vulnerable population.[2-4] As seen in Figure 1, the mean 5-year survival outcome has not changed during the time phase studied. The “death due to other causes” was progressively decreasing while other outcomes were nearly stable. The number of HNC being diagnosed is increasing over the time phases. As observed in Figure 2, the number of PLWC being alive at 5-year period was increasing while the “death due to other causes” was significantly decreasing while death due to cancer is nearly stable. The advances in medical care for non-cancer related ailments could be the suggested reason, and further factors behind this phenomenon need to be explored.

The survival period, especially the Inter Quartile Range denotes that at least 25% of the suicide category had lived only 11 months after diagnosis while 75% of them live up to 7.5 years after the diagnosis. Similarly, 75% of those who are alive live up to 10 years. 75% of the people who are dead due to HNC live only up to 2 years 9 months since diagnosis. Survival is also influenced by the grade of tumor at presentation [Table 3, Figures 1-4]. The survival trend difference between the grades of the tumor (P = 0.000) and warrants further investigation.

The number of tumors at initial presentation is also significant. As observed in the Table 2, the number of tumors indicates that those HNC who had committed suicide had relatively higher mean number of tumors. To the best of our knowledge, there is pertinent literature to support or refute this claim. The relationship between 5-year survival and grade of tumor (at presentation), as seen in Figure 3, indicates that the grade of the tumor has a definite role to play in 5-year survival, especially with suicides.

Observations of increased prevalence of depression in several types of cancers including HNCs, in higher frequency than general populations, have been reported.[16,17] Depression is a known to be associated with high risk of suicide, even in normal population. Role of neurotransmitter, particularly of serotonin and noradrenaline in depression and suicide have been documented.[18-20] Conventional history and mental health assessment interviews are the current methods to prevent the suicide in HNCs.[21,22] Often the HNCs team, especially from the first contact, surgeons, often feel not sufficiently trained to screen for suicidal tendency and by reflex, they often refer it to the mental health team. Non-compliance to such referrals is not uncommon and may compound to the magnification of the issue.[23]

The statistical outcome of the multiple regression analysis points out several factors, the adjusted R² being 0.12 (for all HNCs) and indicates that sum of all considered independent variables into this regression model (race, side laterality, tumor grade, marital status at diagnosis, insurance status, age group, gender, and site) explained only up to 12% of the variability of the considered dependent variables. Therefore, many other factors such as morbidity associated with treatment/HNC, concomitant substance abuse, treatment approach, and pre-existing mental illness besides numerous other factors may contribute to the occurrence of suicides in patients with HNCs.

The study also suffers from certain limitations inherent to observational data of SEER. Missing data, reproducibility, reproducibility of data collected from database, representative sampling, etc., pose methodological issues.[24,25] Data collected is based on queries posed in the software, and even alteration of one criterion could change the number of cases in the net result. The relying on the cause of death classification and chance of misclassification of death, including inability to differentiate between treatment-related mortality and death from underlying comorbid conditions or missing out of secondary tumors. In addition, important covariates such as comorbidity, pre-existing mental illness, and substance abuse including but not limiting to tobacco and alcohol, HPV, education, and socioeconomic factors were not factored in this study. Limitation of not including TNM staging parameters, treatment factors, and duration of therapy could have a direct bearing on the results. In the present study, the results of Kaplan–Meier statistics and multiple regression analysis have to be interpreted with caution.[26]

Clinical Implication: Globally studies indicate that non-communicable diseases (NCDs) such as mental illness, substance abuse, and cancer are one of the major causes of mortality and morbidity. Oral diseases have emerged as the single most common NCD, which needs to be addressed.[27-29] Impact of mental illness, substance abuse with respect to oral diseases, including HNC has not been adequately studied. This is in spite of the fact that oral diseases, is a part of NCD and has a major impact on quality of life. The results of this study indicate that there should be a multi-disciplinary approach to provide much-needed relief to, especially, when the patients with oral diseases harbor suicidal ideation.

CONCLUSION

The number of HNC tumors has an influence on 5-year survival rate. The 5-year survival characteristics of HNC in patients in the SEER database for the period 1974–2008 is presented. It is observed that suicide risk in PLWC HNC is higher and the grade has a bearing on the 5-year survival. The risk of suicide is high in the initial period of diagnosis and persists even after 10–30 years of diagnosis. Hence, PLWC should be screened by competent mental health professionals, when some signs of negative approach or feelings are expressed by patients. With newer classifications of HNC being proposed based on treatment, importance to mental well-being also should be considered. With increasing survivorship after HNC cancer, there is a need for more structured and evidence-based mental health care that has been already called for.