Shuman AG, Terrell JE, Light E, Wolf GT, Bradford CR, Chepeha D, Jiang Y, McLean S, Ghanem TA, Duffy SA. Predictors of Pain Among Patients With Head and Neck Cancer. Arch Otolaryngol Head Neck Surg. 2012;138(12):1147-1154. doi:10.1001/jamaoto.2013.853
Author Affiliations: Department of Otolaryngology–Head and Neck Surgery, University of Michigan Medical School, Ann Arbor (Drs Shuman, Terrell, Wolf, Bradford, Chepeha, McLean, and Duffy); Head and Neck Service, Department of Surgery, Memorial Sloan–Kettering Cancer Center, New York, New York (Dr Shuman); Biostatistics Core, University of Michigan Comprehensive Cancer Center, Ann Arbor (Ms Light); School of Nursing, University of Michigan, Ann Arbor (Ms Jiang and Dr Duffy); Department of Otolaryngology–Head and Neck Surgery, Henry Ford Health System, Detroit, Michigan (Dr Ghanem); and Ann Arbor Veterans Affairs (VA) Center for Clinical Management Research, VA Ann Arbor Healthcare System, Ann Arbor (Dr Duffy).
Objective To determine predictors of pain 1 year after the diagnosis of head and neck cancer.
Design Prospective, multisite cohort study.
Setting Three academically affiliated medical centers.
Patients The study population comprised 374 previously untreated patients with carcinoma of the upper aerodigestive tract.
Main Outcome Measures Participants were surveyed before treatment and 1 year thereafter. Multivariate analyses were conducted to determine predictors of the 36-Item Short-Form Instrument (SF-36) bodily pain score 1 year after diagnosis.
Results The mean SF-36 bodily pain score at 1 year was 65, compared with 61 at the time of diagnosis (P = .004), and 75, the population norm (lower scores indicate worse pain). Variables independently associated with pain included pretreatment pain score (P < .001), less education (P = .02), neck dissection (P = .001), feeding tube (P = .05), xerostomia (P < .001), depressive symptoms (P < .001), taking more pain medication (P < .001), less physical activity (P = .02), and poor sleep quality (P = .006). The association between head and neck cancer pain and current smoking and problem drinking did not reach significance (P = .07 and P = .08, respectively).
Conclusions Aggressive pain management may be indicated for patients with head and neck cancer who undergo neck dissections, complain of xerostomia, require feeding tubes, and have medical comorbidities. Treatment of modifiable risk factors such as depression, poor sleep quality, tobacco use, and alcohol abuse may also reduce pain and improve quality of life among patients with head and neck cancer.
The Institute of Medicine considers effective analgesia a moral imperative and has recently challenged medical providers to better understand, prevent, and treat pain.1 Patients with head and neck cancer frequently experience pain, even long after the completion of treatment.2 The prevalence of bodily pain among patients with head and neck cancer has been estimated at 70%, considerably higher than what has been reported in other patients with cancer.3 Pain experienced 1 year after diagnosis strongly predicts long-term quality of life4 and is associated with poorer survival among head and neck patients, particularly those with advanced disease.5 Bodily pain is also a robust predictor of disability in patients with head and neck cancer.6
Many studies have investigated the factors associated with bodily pain among patients with cancer, which include age, underlying physical and mental health, and depression.7,8 Specific to head and neck cancer, medical comorbidities, chemotherapy, and presence of a feeding tube are associated with significant bodily pain.9 Tumor site and stage have been correlated with pain reported by patients with oral cavity cancer.10
Tobacco use has been temporally associated with pain severity,11,12 and problem drinking has been associated with pain.13 Diet and physical activity also influence physical pain among patients with cancer.14,15 While increased physical activity16 and fruit and vegetable intake14 have been shown to improve quality of life among patients with head and neck cancer, no studies have shown a relationship between these factors and pain. Moreover, bodily pain predicts poor sleep quality and is significantly associated with insomnia. 17,18
While effective pain treatments exist, identifying those most at risk for pain could facilitate targeted assessment and intervention. Thus, the objective of this study was to determine the variables associated with pain among patients with head and neck cancer 1 year following diagnosis.
This investigation was conducted as part of a prospective cohort study involving patients enrolled in the University of Michigan Head and Neck Cancer Specialized Program of Research Excellence. This substudy was performed to determine variables associated with pain 1 year after diagnosis among patients with head and neck cancer. The dependent variable was the 36-item Short-Form Instrument (SF-36) bodily pain score at 1 year after diagnosis. The independent variables included demographic characteristics, clinical factors, and health behaviors. Institutional review board approval was obtained at each study site.
Subjects were recruited from the waiting rooms in head and neck cancer clinics within 3 institutions: the University of Michigan Health System, the Veterans Affairs (VA) Ann Arbor Healthcare System, and the Henry Ford Health System (HFHS). Patients with squamous cell carcinoma of the upper aerodigestive tract mucosa (excluding skin and esophagus) were eligible for enrollment. Patients who were younger than 18 years, pregnant, non-English speaking, psychologically or cognitively unable to complete the survey, or with different tumor sites and/or histologic subtypes were excluded. Research assistants recruited patients after obtaining informed consent and administered a written survey. Clinical data were collected directly from the surveys and were confirmed and/or abstracted from the medical record. The survey was repeated 1 year after diagnosis. Participants received $10 compensation after the completion of each questionnaire. Of 811 patients recruited to the study, 559 had 1-year SF-36 scores, and the 374 with complete data were included in the final multivariate analysis.
The SF-36 is a comprehensively validated index used to assess quality of life.19,20 It has been accepted as one of the most widely used and exhaustively studied survey-based, patient-assessed health outcome measures.21 It has been successfully used to study patients with head and neck cancer and independently predicts survival within this cohort.22,23 The SF-36 bodily pain score is an independently validated subscale that has high fidelity and reproducibility in describing patient-reported pain24; based on a scale of 0 to 100, the population norm (in a cohort with a mean age of 58 years) was 75 (lower scores indicate worse pain).25
Demographic information included patient age at the time of diagnosis (in years), sex, race, marital status (yes/no), and education level (high school or less vs some college or more). Because race has been shown to affect outcomes among patients with head and neck cancer, respondents were asked to self-identify their ethnicity/race using the US census 2-category question about Hispanic/Latino origins and a 5-category question about race (of which respondents could select multiple categories).26 Because there were few respondents in many of the categories, race was collapsed into white and nonwhite for analysis.
Cancer descriptors included the primary tumor site and disease stage according to the American Joint Committee on Cancer (AJCC) (sixth edition).27 For the purposes of statistical analysis, tumor sites were grouped in aggregate (into pharynx, larynx, and oral cavity [including other/sinus]), and stage was separated into stages I/II vs stages III/IV.
Treatment variables included whether subjects received (yes/no) radiation, chemotherapy, primary site surgery, and/or neck dissection. Also assessed was the presence (yes/no) of an indwelling feeding tube and/or tracheotomy at 1 year after diagnosis. Neck dissections were defined as any systematic cervical nodal extirpation regardless of level(s) dissected or additional structure(s) removed.
Xerostomia was described by a question taken verbatim from the Head and Neck Quality of Life (HNQOL), a validated, disease-specific quality of life instrument for patients with head and neck cancer.28 Medical comorbidities were assessed by medical chart abstraction using the validated Adult Comorbidity Evaluation-27 (ACE-27) index and classified into 2 groups—moderate to severe comorbidity vs no to mild comorbidity.29 Depression was measured using the validated 5-item Geriatric Depression Scale-Short Form (GDS-SF): a score of 4 or higher on this scale indicates probable depression.30 Pain medicine use by subjects at 1 year was assessed with 1 question measured on a 5-point Likert scale from 1 (never) to 5 (always).
Subjects were asked about their prior and current tobacco use; respondents who reported any tobacco use within the past month were considered current smokers; those who reported use prior to 1 month ago were considered former smokers. The 10-item Alcohol Use Disorders Identification Test (AUDIT) was used to assess the amount of alcohol intake and related problems; a score of 8 or higher on this test indicates a high risk of alcohol-related disorders.31 Selected questions from the validated Willett food frequency questionnaire were used; respondents reported the average number of servings of fruit, vegetables, and fried foods consumed over the past year.32 The validated Physical Activity Scale for the Elderly (PASE) was used to measure activity; scores ranged from 0 to 400 or more (higher scores indicate more activity), and the population mean score for people aged 65 to 100 years was 103.33 Sleep quality was assessed using the validated Medical Outcomes Study sleep measure (MOS-Sleep); lower scores indicate poorer sleep.34,35
Descriptive statistics were calculated for all measures. Frequencies and percentages are presented for categorical variables, and means and standard deviation are presented for continuous measures. A correlation matrix was used to determine the co-linearity between variables (not shown). Bivariate analyses (analysis of variance and the Spearman ρ statistic) were used to test correlations between SF-36 pain index and each independent variable. Based on the bivariate analyses, ordinary least squares multiple linear regression was conducted to determine independent variables associated with the dependent variable at 1 year after diagnosis. All analyses were conducted using SAS 9.3 software (SAS Institute Inc).
The demographic and health characteristics of the sample are summarized in Table 1. The mean age of participants was 58 years. Most patients were male (78%), white (91%), and married (60%), and just over half had some college education or more (56%). The majority of primary sites were pharyngeal (53%), and advanced-stage disease was common (75% with stage III or IV). Approximately 84% had radiation therapy, 62% had chemotherapy, 38% had primary site surgery, and 43% underwent neck dissection. At 1 year after treatment, 18% of patients had an indwelling feeding tube and 4% had a tracheotomy. Xerostomia was frequent, with 44% of patients complaining of “a lot” or “extreme” symptoms at 1 year. Approximately 39% of patients screened positive for depression. Most patients (54%) never or rarely used pain medicine at 1 year. Approximately one-quarter of patients (22%) were current smokers, and only 12% were problem drinkers.
The mean SF-36 pain score was lower (worse) before treatment than 1 year later (61 vs 65; P = .004). Factors associated with 1-year pain on bivariate analysis are summarized in Table 2 and included marital status, educational level, cancer site, primary site surgery, neck dissection, feeding tube, tracheotomy, xerostomia, depression, smoking, problem drinking, vegetable and fruit intake, physical activity, and sleep.
The multivariate analysis is reported in Table 3. Multivariate analysis indicated that pretreatment pain (P < .001), less education (P = .02), neck dissection (P = .001), presence of a feeding tube (P = .05), xerostomia (P < .001), depressive symptoms (P < .001), taking more pain medication (P < .001), less physical activity (P = .006), and poor sleep quality (P = .02) were independently associated with worse pain 1 year after diagnosis. Current smoking and problem drinking were not statistically significant in the multivariate analyses (P = .07 and P = .08, respectively). While cancer site, primary site surgery, tracheotomy, comorbidities, and vegetable and fruit intake were significant in the bivariate analyses, they were no longer significant in the multivariate analyses.
Alternative regression models were created to control for potential confounders and statistical limitations (data not shown). Models that did not incorporate pretreatment pain scores and those that excluded analgesic use both yielded similar results to the reported data. Cancer stage restratified into stages I/II/III vs stage IV, inclusion of tumor N category, inclusion of recurrent disease, and chemoradiation (as a yes/no variable) also did not change the results.
Bodily pain scores among patients with head and neck cancer are better at the 1-year follow-up compared with the pretreatment value. The improvement at 1 year is intuitive, given the pain inherent to an untreated tumor burden at the time of diagnosis and the likelihood that the acute stigmata of cancer treatment have resolved by 1 year. Nonetheless, the mean 1 year pain score is worse than scores seen in the general population of similar age and is predicted by pretreatment pain.19 Our prior work with patients with head and neck cancer has shown that a 10-point decrement in SF-36 bodily pain scores is associated with a 20% increased risk of disability.6 In another study, the incidence of oral pain among 5-year head and neck cancer survivors was 43%.36 Our data corroborate that patients with head and neck cancer continue to experience pain 1 year after diagnosis and that dedicated, evidence-based strategies for effective analgesia are likely to be critical in order to optimize our patients' qualities of life.
Lower educational level was associated with more pain, corroborating data suggesting that educational level also predicts quality of life among patients with nasopharyngeal cancer and overall disability in a population of treated patients with head and neck cancer.6,37 Interestingly, none of the other demographic characteristics evaluated were independently associated with bodily pain scores. Our findings support the notion that physical and psychological factors influence pain to a greater extent than demographics38 and are in accordance with a large study that found that no demographic factors predicted pain among a diverse cohort of patients with cancer.39
While cancer site, stage, and treatment would seemingly affect pain, these were surprisingly not significant in the multivariate analyses. Instead, the adverse consequences of treatment, which were studied as independent variables as well, were more indicative of pain at 1 year. For example, xerostomia was closely associated with pain, but radiation therapy was not similarly related; thus we postulate that salivary gland-sparing regimens delivered with intensity-modulated radiation treatment (IMRT) might limit xerostomia and resultant pain perception.40 Data suggesting that IMRT broadly improves quality of life when compared with conventional radiation delivery algorithms confirm this conjecture.41,42 Unfortunately, specific data on radiation fields was not collected and could not be included in the analyses.
The presence of a feeding tube at 1 year strongly influences quality of life and was significantly associated with pain in the multivariate analysis.9 In many cases, consequences of treatment (regardless of modality) influence posttreatment function related to eating and drinking. The presence of a tracheotomy at 1 year, which also indirectly reflects tissue effects of cancer-directed treatment, was not significantly associated with pain, possibly owing to the low incidence of tracheotomies at 1 year. While primary site surgery was not associated with pain, the one treatment that was significantly associated with pain was neck dissection. This finding supports prior data that highlight the tangible, long-term functional consequences of neck dissections, particularly in posttreatment settings.43 Patients undergoing neck dissection were likely to also have been treated with radiation, which might also contribute to pain. We did not compare pain scores for neck dissection alone with neck dissection prior to or following radiation therapy, nor did we specifically control for the type of neck dissection. While the era of “radical” neck surgery has been supplanted by movements toward “selective” operations, these observations confirm that neck dissections, even those performed by experienced surgeons and that aim to limit morbidity, are not always innocuous.44
Medical comorbidities were not associated with pain, which does not corroborate findings in the general population of patients with cancer; however, 70% of the population had no or mild comorbidities. A relationship exists between medical comorbidities and many quality-of-life–related metrics in patients with head and neck cancer.45,46 Among patients with head and neck cancer along with multiple medical problems, amelioration of symptoms may require diligent titration of pharmacologic regimens and close collaboration with other clinicians.
As has been seen in other studies of patients with cancer, depression is highly correlated with pain.47 Depressed patients are more sensitive to somatic discomfort, which also engenders depressive symptoms, particularly among patients with cancer.48 Depression is also associated with health behaviors, some of which were associated with pain and which must be considered in treating pain.49,50 Smoking, alcohol abuse, and depression are interrelated, and treating depression may enhance substance abuse cessation efforts. Also, depression has been shown to be correlated with sleep disturbances.17 The data suggest that clinicians should seek to diagnose and treat depression among patients with head and neck cancer complaining of pain and also that depressed patients may require aggressive analgesia in addition to treating their mood and/or sleep disorders.
The smoking status of patients with cancer is closely linked with quality of life, and we identified a trend toward more pain in current tobacco users.51 The mechanisms by which tobacco influences head and neck cancer pain are undoubtedly multifactorial, involving central nociceptive processing mechanisms, psychological factors, and direct mucosal irritation, among others.12 Maladaptive strategies in which tobacco is relied on for pain relief are prevalent; educational and pharmaceutical initiatives in which tobacco is recast as an accomplice to pain may improve success rates in achieving abstinence.52
The association between problem drinking and with head and neck cancer pain did not reach significance (P = .08). In general, patients with better function are more likely to drink alcohol socially, but alcohol may also be used for its analgesic properties among those in pain. Self-medication of pain with alcohol has been associated with pain frequency, depression, and use of pain medications.53 Moreover, problem drinkers have a higher incidence of tobacco abuse and are less likely to quit.54 Clinicians should individualize counseling and interventional strategies for patients with problem drinking, being mindful of its complex relationship with clinical factors and other health behaviors.55
There was an association between increased physical activity and less pain. Directed rehabilitative interventions also appear to improve somatic function in patients with head and neck cancer, specifically regarding dysphagia treatment and prevention and shoulder-neck range of motion after surgery.56,57 Poor sleep was closely associated with worse pain, which is supported by prior research,58 since sleep disturbances are likely to be caused by severe pain. Our findings support the need to screen for sleep disturbances among patients with bothersome pain and also suggest that undertreated pain may contribute to seemingly idiopathic insomnia.
While one might postulate instead that efficacious medication use should decrease pain, our data reinforce that patients with head and neck cancer who use analgesics still report pain, suggesting that their regimens may not be adequately effective. While we did not specifically address analgesic efficacy or regimens, data suggest that relatively simple interventions, such as prescribing medications on a schedule rather than as needed, more effectively manage postsurgical pain in patients with head and neck cancer.59 The World Health Organization pain ladder offers a practical, stepwise approach to managing cancer pain that is readily applicable to patients with head and neck cancer.60 Patients with head and neck cancer are at risk for cancer recurrence and for second primaries; pain is a frequent harbinger of such phenomena, and clinicians must remain ever vigilant during continued oncologic surveillance.
The SF-36 bodily pain index does not distinguish head and neck symptoms; thus, pain scores may or may not relate directly to patients' cancer and its treatment. In quality-of-life research in head and neck cancer, no single metric can effectively encapsulate the patient experience or perspective.61 Given that quality of life among patients with head and neck cancer may decline over many years, 1-year pain scores do not necessarily encompass the evolution of longer-term symptoms.62 Despite extra efforts to recruit minorities, the study overrepresented white men, potentially limiting the ability to extrapolate our findings in other populations.
Many patients with head and neck cancer experience pain 1 year after diagnosis. Pain, more than any other symptom or disease, is the single greatest source of financial expense and patient complaints in America.63 Reducing the potential for pain from neck dissection, minimizing xerostomia, and using evidence-based strategies for effective analgesia are opportunities worth considering in order to limit pain and improve quality of life for these patients. An understanding of the complex relationships among cancer treatment, health behaviors, and pain is instructive to clinicians faced with preventing, anticipating, counseling, and treating bothersome symptoms in a challenging patient population.
Correspondence: Sonia A. Duffy, PhD, RN, School of Nursing, University of Michigan, 2215 Fuller Rd, PO Box 130170, Ann Arbor, MI 48113-0170 (firstname.lastname@example.org).
Submitted for Publication: July 6, 2012; final revision received August 23, 2012; accepted September 18, 2012.
Published Online: November 19, 2012. doi:10.1001/jamaoto.2013.853
Author Contributions: Dr Duffy had full access to all 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: Shuman, Terrell, Wolf, Chepeha, and Duffy. Acquisition of data: Terrell, Light, Wolf, Chepeha, McLean, Ghanem, and Duffy. Analysis and interpretation of data: Shuman, Terrell, Light, Wolf, Bradford, Chepeha, Jiang, and Duffy. Drafting of the manuscript: Shuman, Light, Wolf, Chepeha, Jiang, and Duffy. Critical revision of the manuscript for important intellectual content: Shuman, Terrell, Light, Bradford, Chepeha, McLean, Ghanem, and Duffy. Statistical analysis: Shuman, Light, and Jiang. Obtained funding: Terrell, Wolf, and Duffy. Administrative, technical, and material support: Terrell, Wolf, Chepeha, and Duffy. Study supervision: Terrell, Bradford, and Duffy.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the National Institutes of Health through the University of Michigan's Head and Neck Specialized Program of Research Excellence (SPORE) grant (P50 CA097248) titled “The Molecular Basis of Head and Neck Cancer: Biology, Treatment and Prevention.”
Previous Presentation: This study was presented at the American Head and Neck Society meeting; July 24, 2012; Toronto, Ontario, Canada.
Additional Contributions: We would like to express our gratitude to the many patients with head and neck cancer who participated in this study. Many thanks to all clinicians, researchers, and ancillary staff members associated with the care and study of the patients involved with the University of Michigan Head and Neck Specialized Program of Research Excellence, including Tom Carey, PhD; Theodoros Teknos, MD; Nisha D’Silva, DDS, PhD; Avraham Eisbruch, MD; Theodore Lawrence, MD, PhD; Sofia Merajver, MD, PhD; Jacques Nor, DDS, PhD; Mukesh Nyati, PhD; Laura Rozek, PhD; Shaomeng Wang, PhD; and Frank Worden, MD.