Fax: (011) 0031887555850.
The prognostic role of comorbidity in salivary gland carcinoma
Article first published online: 14 AUG 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 7, pages 1572–1579, 1 October 2008
How to Cite
Terhaard, C. H. J., van der Schroeff, M. P., van Schie, K., Eerenstein, S. E. J., Lubsen, H., Kaanders, J. H. A. M., Smeele, L. E., Burlage, F. R., van Den Ende, P. L. and Baatenburg de Jong, R. J. (2008), The prognostic role of comorbidity in salivary gland carcinoma. Cancer, 113: 1572–1579. doi: 10.1002/cncr.23771
- Issue published online: 17 SEP 2008
- Article first published online: 14 AUG 2008
- Manuscript Accepted: 25 APR 2008
- Manuscript Revised: 24 APR 2008
- Manuscript Received: 28 FEB 2008
- malignant salivary gland tumors;
- overall survival;
- disease-free survival;
- prognostic factors;
- head and neck cancer;
- multivariate analysis
Patients with head and neck cancer are prone to develop significant comorbidity mainly because of the high incidence of tobacco and alcohol abuse, both of which are etiologic and prognostic factors. However, to the authors' knowledge little is known regarding the prognostic relevance of comorbidity in patients with salivary gland cancer.
A retrospective cohort of 666 patients with salivary gland cancer was identified within the Dutch Head and Neck Oncology Cooperative Group database. For multivariate analysis, a Cox proportional hazards model was used to study the effect of comorbidity on overall survival and disease-specific survival.
According to the Adult Comorbidity Evaluation-27 (ACE-27) index, 394 patients (64%) had grade 0 comorbidity, 119 patients (19%) had grade 1 comorbidity, 71 patients (12%) had grade 2 comorbidity, and 29 patients (5%) had grade 3 comorbidity. In multivariate analysis for overall survival, the ACE-27 comorbidity grade was a strong independent prognostic variable. The hazards ratio (HR) of death, including all causes, was 1.5 (95% confidence interval [CI], 1.1-2.1) for patients with ACE-27 grade 1 comorbidity versus grade 0 comorbidity (P < .007). The HR was 1.7 (95% CI, 1.2-2.5) for grade 2 comorbidity (P = .003) and 2.7 (95% CI, 1.5-4.7) for grade 3 comorbidity versus grade 0 comorbidity (P = .001). In the current analysis, ACE-27 comorbidity grade was not an independent prognostic factor for disease-free survival.
To the authors' knowledge, this is the first study concerning the prevalence and relevance of the prognostic comorbidity variable ACE-27 grade in patients with salivary gland cancer. Overall survival, but not disease-free survival, was correlated strongly with ACE-27 grade. Compared with other studies that investigated the effect of comorbidity on patients with head and neck cancer, patients with salivary gland cancer had less comorbidity. Their comorbid status appeared to be reasonably comparable to that of patients with other nonsmoking- and nonalcohol-related cancers. Cancer 2008. © 2008 American Cancer Society.
Patients with head and neck cancer are prone to have significant comorbidity mainly because of the high incidence of tobacco and alcohol abuse, both of which are etiologic and prognostic factors. The associated medical problems are most pronounced in the respiratory and cardiovascular systems. These conditions have direct impact on the care of the patient, selection of initial treatment, and evaluation of treatment effectiveness.1–3 It has been demonstrated that comorbidity is an important prognostic factor for overall survival4 and, in some studies, for locoregional control5 in patients with malignant tumors of the head and neck area. Comorbidity may influence the choice of treatment, eg, leading to a more conservative approach to prevent possible postoperative complications. Several comorbidity indices have been developed, such as the Kaplan-Feinstein Index6 and the Adult Comorbidity Evaluation-27 (ACE-27), which is a modification of the Kaplan-Feinstein Index.6 In recent years, the ACE-27 has become a common method for scoring comorbidity. The ACE-27 adds several important comorbid conditions, such as acquired immunodeficiency syndrome and diabetes mellitus. Modification of the Kaplan-Feinstein Index into the ACE-27 was done based on expert opinion and a review of the medical literature.1 The ACE-27 grades specific comorbid conditions into 1 of 4 levels of comorbidity—none, mild, moderate, or severe—according to the severity of individual organ decompensation and prognostic impact.7 The overall comorbid score, or ACE-27 ranking, is based on the highest ranked single ailment. Patients with 2 or more moderate ailments in different organ systems or disease groupings are graded as severe. The ACE-27 is a comprehensive tool that has been modified previously and validated8 and is of common use in head and neck cancer literature. Whether comorbidity is a prognostic factor in salivary gland carcinomas is not known: Because of the absence of an etiologic correlation between salivary gland carcinomas and tobacco and alcohol abuse, a prognostic significance in theory may be questionable. The objective of this study was to evaluate the prevalence and the prognostic importance of comorbidity in a group of patients who were treated for salivary gland cancer. To our knowledge, in the literature, no such study has been published to date.
MATERIALS AND METHODS
On the basis of 565 patients in a dataset from the Dutch Head and Neck Oncology Cooperative Group (NWHHT) concerning general results in patients with salivary gland cancer,9 the role of radiotherapy10 and the importance of facial nerve palsy in parotid cancer11 have been published. An update of the NWHHT salivary gland cancer database has been performed, including the variable comorbidity. In this article, we present the results from those 565 patients and from an additional 101 patients with malignant salivary gland tumors, including up-to-date follow-up data and information on comorbidity. Thus, the dataset has been extended to 666 patients who were treated between 1985 and 1994. The median follow-up of patients who were alive at last follow-up was 125 months.
The mean patient age was 59 years (range, 8-100 years). Fifty-two percent of patients were men. Pain was a complaint in 27% of patients. Clinical tumor invasion of the skin was noted in 7% of patients. The salivary gland cancer was located in the parotid gland in 56% of patients, in the submandibular gland in 13% of patients, in the oral cavity in 26% of patients, and in the pharynx/larynx in 5% of patients. Facial nerve function was intact in 79% of patients with parotid gland cancer, partially impaired in 13% of patients; and, in 8% of patients, complete facial paralysis was noted before treatment.
Comorbidity Scoring Using the ACE-27 Index
In this analysis, we used the ACE-27 index for scoring comorbidity. Data about comorbidity were derived from the medical record in 92% of patients. The number of patients with ACE-27 grade 0, 1, 2, and 3 comorbidity was 394 patients (64%), 119 patients (19%), 71 patients (12%), and 29 patients (5%), respectively.
The TNM staging was used for both major and minor salivary gland cancers according to the 2002 American Joint Committee on Cancer (AJCC) Cancer Staging Manual.12 Four percent of patients had M1 disease, and 15% had positive lymph nodes at first presentation. Distribution of T classification was 24%, 40%, and 35% for patients with T1, T2, and T3/T4, respectively. According to the 2002 AJCC manual, staging for facial nerve tumor invasion of the skin, mandible, ear canal, and/or facial nerve is staged as T4a. Table 1 shows the distribution of clinical disease classification according to the 2002 AJCC manual.
|I||Tumor ≤2 cm (T1), N0, M0||19|
|II||Tumor >2 cm to ≤4 cm (T2), N0, M0||29|
|III||Tumor >4 cm (T3), N0, M0 or T1-T3, N1, M0||24|
|IVa||T1-T3, N2, M0 or T4a,* N0-N2, M0||22|
|IVb||T4b,† N0-N2, M0 or any T, N3, M0||1|
|IVc||Any T, any N, M1||4|
Surgery alone was undergone by 141 patients (22%) and was combined with postoperative radiotherapy in 444 patients (67%). Radiation alone was received by 47 patients (7%), and no treatment or chemotherapy alone was received by 34 cases (5%).
Histology was reviewed and classified according to the 1972 World Health Organization classification. Adenoid cystic carcinoma was diagnosed most frequently (27%), followed by mucoepidermoid carcinoma (16%), acinic cell carcinoma (14%), carcinoma expleomorphic adenoma (8%), undifferentiated carcinoma (7%), and squamous cell carcinoma (5%). Twenty-one percent of tumors were diagnosed as adenocarcinoma, not otherwise specified. Sixteen patients (2%) were missing information on histology.
Including data derived from the resection specimen of the primary tumor (n = 585) and neck dissection (n = 228), the distribution according to pathologic AJCC stages I, II, III, and IV was 27%, 21%, 16%, and 35%, respectively. Resection of the tumor was incomplete in 36% of patients and close (<5 mm) in 21%. Bone and skin were invaded in 8% and 5% of patients, respectively. Perineural invasion was diagnosed in 30% of patients; however, information about perineural invasion was lacking in 21% of patients.
For univariate analysis, version 10.0 of the SPSS/PC software program was used. Statistical significance was calculated by using the chi-square test and the Mann-Whitney test. To determine independent risk factors for ACE-27, binary regression analysis was performed. For actuarial curves, the method of Kaplan Meier was used along with the log-rank test for computing statistical significance. For multivariate analysis, a Cox proportional hazards model was used. A comparison between the occurrence of comorbidity in patients with salivary gland carcinomas and in patients with other head and neck cancers is presented in Table 2. Statistical significance was calculated by using the Kruskal-Wallis test, because there is an ordering in the ACE-27 scores.
|Study||Type of Tumor||No. of Patients (%)||Chi-square*|
|Total No.||ACE Grade 0||ACE Grade 1||ACE Grade 2||ACE Grade ≥2||ACE Grade 3|
|Ferrier 200513||HNSCC: Sinus, lip, oral cavity, nasopharynx, oropharynx, hypopharynx, larynx||117||48 (41.0)||35 (29.9)||—||34 (29.1)||—||22.6†|
|Borggreven 200314||HNSCC: Oral cavity, oropharynx||100||17 (17)||36 (36)||34 (34)||—||13 (13)||83.4†|
|Rogers 200615||HNSCC: Hypopharynx, larynx, oropharynx||157||74 (47)||57 (36)||—||26 (17)||—||21.6†|
|Sanabria 20074||HNSCC: Nasopharynx, nose/paranasal, hypopharynx, oropharynx, larynx, oral cavity||309||77 (25)||141 (46)||48 (15)||—||43 (14)||136.9†|
|Piccirillo 20047||HNSCC: Lip, oral cavity, nasopharynx, oropharynx, hypopharynx, larynx, thyroid gland||341||188 (55)||82 (24)||53 (16)||—||18 (5)||8.0†|
|Current study||Salivary gland||613||394 (64)||119 (19)||71 (12)||—||29 (5)|
Correlation Between ACE-27 Comorbidity and Possible Prognostic Factors
ACE-27 grade was correlated significantly with age (P< .001) (see Fig. 1). ACE-27 comorbidity grades 1 through 3 were diagnosed in 30% of women versus 41% of men (P = .02). The ACE-27 grade was not correlated with pain, facial nerve function, tumor localization, or clinical AJCC 2002 stage. Among the patients who had clinical skin invasion, 52% had ACE-27 comorbidity grades 1 through 3 compared with 34% among patients without clinical skin invasion (P = .02). For all histologic subtypes, the distribution according to ACE-27 grade was equal, except for squamous cell carcinoma. Approximately 66% of patients with squamous cell carcinoma had ACE-27 comorbidity grades 1 through 3 compared with approximately 33% of all other patients. Patients who had ACE-27 grade 3 comorbidity received radiotherapy alone significantly more often compared with other patients. Conversely, if patients underwent surgery, then postoperative radiotherapy was received less frequently in this group (see Fig. 2). The mean delay between surgery and postoperative radiotherapy was equal for patients with ACE-27 grade 0 through 2 comorbidity; however, the delay was significantly prolonged for patients with ACE-27 grade 3 comorbidity (6.5 weeks vs 9 weeks, respectively; P = .005).
No correlation was observed between the ACE-27 comorbidity grade and any histopathologic variable derived from the resection specimen. The completeness of resection and the ACE-27 grade or age were not related. ACE-27 grade was related to 2002 AJCC pathologic stage: ACE-27 grade 0 comorbidity occurred in 79%, 64%, 61%, and 59% of patients with AJCC stage I, II, III, and IV, respectively (P = .001). Independent risk factors for ACE-27 (grade 0 comorbidity vs grade 1-3 comorbidity) were age, with an increased relative risk (RR) of 1.05 (95% confidence interval [CI], 1.04-1.06) per year, and sex, with an RR for men versus women of 1.7 (95% CI, 1.2-2.5).
The actuarial 10-year disease-free survival rate for grade 0, 1, 2, and 3 comorbidity was 59%, 47%, 43%, and 40%, respectively (P = .003; log-rank test) (see Fig. 3). Because disease-free survival was determined by 3 different events (local control, regional control, and distant metastases), these 3 events were analyzed separately. No significant difference was observed between ACE-27 comorbidity grade and actuarial regional control or distant metastases-free survival. The 10-year actuarial local recurrence-free survival rate was 81%, 68%, 75%, and 65% for grade 0, 1, 2, and 3 comorbidity, respectively (P = .005; log-rank test).
In a multivariate analysis with disease-free survival as the endpoint, variables that were significant in the univariate analysis were included. Inclusion of clinical AJCC 2002 stage resulted in loss of significance of the variables tumor classification, lymph node status, metastasis stage, facial nerve invasion, and skin invasion. Independent factors were clinical AJCC 2002 stage, pain, sex, and age. In this analysis, ACE-27 comorbidity grade was not an independent prognostic factor for disease-free survival. Because treatment performed may influence disease-free survival, this was also included in the analysis. Treatment was an independent factor. Compared with surgery alone, combined surgery and radiotherapy resulted in significantly better locoregional control, and radiation alone in resulted significantly worse results consistent with our previous report.10 For the patients who underwent surgery, the relevance of the histologic parameters derived from the specimen also was analyzed. The status of the resection margins (patients with incomplete or close margins fared worse) and the pathologic lymph node status also were independent factors for disease-free survival.
The 10-year actuarial overall survival rate was 62%, 38%, 25%, and 14% for ACE-27 grade 0, 1, 2, and 3 comorbidity, respectively (log-rank test; P < .0001) (see Fig. 4). In a multivariate analysis of overall survival, ACE-27 comorbidity grade was a strong independent prognostic variable. The hazard ratio (HR) of death, including all causes, was 1.5 (95% CI, 1.1-2.1) for ACE-27 grade 1 comorbidity versus grade 0 comorbidity (P < .007). The HR was 1.7 (95% CI, 1.2-2.5) and 2.7 (95% CI, 1.5-4.7) for grade 2 comorbidity and for grade 3 versus grade 0 comorbidity, respectively (P = .003 and P = .001, respectively). In addition to ACE-27 grade, pain, age, clinical AJCC 2002 stage, localization of the salivary gland tumor (patients with oral cavity tumors had the best survival), and histologic subtype (patients with acinic cell tumors fared best) were independent factors for overall survival. For the patients who underwent surgery, the significant histologic variables derived from the specimen were pathologic lymph node status, bone invasion, and pathologically verified invasion of the skin.
Comparison of Comorbidity in Salivary Gland Carcinoma Versus Head and Neck Squamous Cell Carcinoma
The results from a comparison between comorbidity in patients with salivary gland carcinomas and comorbidity in patients with other head and neck cancers are presented in Table 2. The chi-square test statistics from the reports by Ferrier et al,13 Borggreven et al,14 Rogers et al,15 Sanabria et al,4 and Piccirillo et al7 were 22.6 (P < .0005), 83.4 (P < .0005), 21.6 (P < .0005), 136.9 (P < .0005), and 8 (P = .046), respectively. Because, in our dataset, the only 2 factors that were related significantly to the ACE-27 comorbidity grade after adjusting for all other factors were age and sex, we checked to determine whether the distribution of age in the articles mentioned above were in reasonable concordance with our data. We also examined the sex distribution over the different articles and observed no significant differences (data not shown). The significant test statistics suggest that the comorbidity distribution is different in patients with salivary gland cancer compared with the comorbidity distribution in patients with head and neck squamous cell carcinoma. The distribution shown in Table 2 indicates that patients with salivary gland cancer have less comorbidity.
Comparison of Comorbidity in Salivary Gland Carcinoma Versus Other Nonsmoking- and Nonalcohol-related Cancer
We know of only 1 article in which the ACE-27 score was determined for patients with nonhead and neck cancers.1 We compared the ACE-27 scores from our cohort of salivary gland cancer patients with the scores from patients who had prostate cancer patients to compare with another nonsmoking- and nonalcohol-related form of cancer. The results are presented in Table 3. The Kruskal-Wallis test indicated that there was a significant correlation (chi-square statistic, 14.6; P = .002). This implies that the comorbidity distribution does deviate from the distribution among patients with prostate cancer, as it does in Table 2.
|Study||Type of Tumor||No. of Patients (%)||Chi-square*|
|Total No.||ACE Grade 0||ACE Grade 1||ACE Grade 2||ACE Grade 3|
|Piccirillo 20001||Prostate cancer||1110||698 (63)||283 (25)||101 (9)||28 (3)||14.6†|
|Current study||Salivary glands||613||394 (64)||119 (19)||71 (12)||29 (5)|
To our knowledge, very few reports published to date have provided the results of multivariate analysis on prognostic indicators performed on a large group of patients that included tumors of all major and minor salivary glands.16–18 Most reports deal exclusively with either tumors of the major salivary glands (submandibular and parotid glands combined), or tumors of the minor salivary glands,19–24 or tumors of the parotid gland only.25–34 We performed an update of the database of the NWHHT concerning all salivary gland cancers irrespective of site and histology. This analysis, which was based on data from all head and neck centers in the Netherlands, included comorbidity scores. Compared with the report published in 2004,9 the current database has been extended from 565 patients to 666 patients with malignant salivary tumors who were treated between 1985 and 1994. The median follow-up of the patients who remained alive at last follow-up was extended from 99 months to 125 months. In particular, this appears to be important for the patients with adenoid cystic carcinomas (27%; 180 of 666 patients), which can behave in an indolent way and, thus, require a long follow-up. Overall, the extension of the follow-up did result in only a few instances of disease recurrence 10 years after treatment.
In the current analysis, we used the ACE-27, a commonly used tool in head and neck cancer literature. If it is used retrospectively, then the information needed to score the ACE-27 is collected through, for example, the medical chart, nursing notes, referral letters from general practitioners, anesthetic sheets, and laboratory investigations. Comorbidity scoring with the ACE-27 using trained cancer registrars in the US reportedly has produced valid information.8 Paleri and Wight35 studied the applicability of the ACE-27 by using notes extraction in a cohort of patients in the UK with head and neck cancer. Those authors concluded that, along with difficulties regarding (confidential) human immunodeficiency virus status, retrospective data collection and the completion of a comorbidity index is feasible. In another study by Paleri and Wight,36 they compared the information obtained from a review of notes alone with information that was available after a structured patient interview. They observed that a retrospective review of notes was an accurate and reliable technique for grading comorbidity. However, in a feasibility study of the ACE-27 among patients with head and neck cancer, Rogers et al37 identified several problems regarding the retrospective use of the ACE-27 index. Not all items mentioned in the ACE-27 are recorded in the actual charts and, if mentioned, are not always detailed enough. Items in the ACE-27 index such as immunologic disorders are included infrequently when taking a patient's history. In addition to the comorbid conditions that are included in the ACE-27 index, there are other theoretic prognostic comorbid conditions, such as anemia and hypercholesterolemia.
In our study, ACE-27 scores were related significantly to age and sex. The first appears to be a consequence of deteriorating health in the elderly, and the latter appears to be a presumed difference in lifestyle. Comorbidity in our study was independent of histologic tumor type, except for squamous cell carcinoma, which has a significant correlation with comorbidity and, thus, reflects the known relation between comorbidity and head and neck squamous cell carcinoma. On the basis of these findings, it may be argued that squamous cell carcinomas are a different entity among all salivary gland carcinomas.
Comorbidity, and specifically comorbidity scored by the ACE-27 index, in head and neck cancer is reportedly a significant prognostic factor.1, 4, 13–15, 38 Comparison of comorbidity in salivary gland carcinoma and the head and neck cancer studies mentioned above indicated that there were consistent and significant differences. Recently, in a retrospective cohort study of 310 patients with head and neck cancer aged >70 years (average age, 76 years) who had a median follow-up of 22.5 months, Sanabria et al4 reviewed the effect of comorbidity on overall and cancer-specific survival. They reported an HR of 1.72 (P = .002) in a multivariate analysis of moderate comorbidity and severe comorbidity (ACE-27 grade 2 and 3 comorbidity) compared with no comorbidity and mild comorbidity (ACE-27 grade 0 and 1 comorbidity) on overall survival. In the current study, multivariate analysis indicated that the prognostic effect of comorbidity on overall survival also was strong. The HR for ACE-27 grade 1, 2, and 3 comorbidity was 1.5, 1.7, and 2.7, respectively (reference, ACE-27 grade 0 comorbidity). A different measure of survival was used by Allareddy and Konety in 200639 in a retrospective analysis from the US of a nationwide inpatient sample for the years 2000 through 2003 consisting of 24,803 patients with head and neck cancer, including salivary gland cancers. Instead of overall survival, as a proxy for assessing clinical outcomes, those authors used in-hospital mortality as the primary outcome (the overall in-hospital mortality rate was 5.18%; n = 1284 patients). Specific comorbid conditions, such as congestive heart failure, renal failure, liver disease, and coagulopathy, had increased risks of in-hospital mortality. That analysis was done on the basis of a multivariate analysis that included the variables sex, comorbid conditions, complications, health insurance status, hospital bed size, hospital localization, and hospital teaching status. The analysis did not include, for example, tumor size and, thus, was uncorrected for disease stage. Noteworthy is a study done by Park et al40 indicating that prediagnosis smoking and heavy alcohol use not only are precursors for comorbidity and etiologic factors in head and neck cancer but also are significant prognostic factors on alone. Several other authors have reported that therapy is dependent on comorbidity41, 42 In our cohort, patients with grade 3 comorbidity received significantly less than gold-standard treatment (surgery or surgery with postoperative radiotherapy) (see also Fig. 2), had a longer delay between surgery and postoperative radiotherapy, and received more primary radiotherapy.
In our cohort, comorbidity was not a relevant prognostic factor for disease-free survival. Neither univariate analysis nor multivariate analysis revealed a significant difference between ACE-27 comorbidity grade and actuarial regional control or distant metastases-free survival. This concurs with the report by Rogers et al in head and neck squamous cell carcinoma,15 who observed no significant correlation between comorbidity and disease-free survival. Sanabria et al4 reported a nonsignificant HR (P = .11) in multivariate analysis for the ACE-27 index grade in relation to cancer-specific survival. An observational prospective cohort of 17,712 patients with cancer in the US, including disease locations in head and neck and also the prostate, lung, breast, digestive system, and gynecologic and urinary systems, revealed that 5058 patients (28.5%) developed disease recurrence.7 The odds ratios of developing recurrence for increasing levels of comorbidity, after adjusting for extent of disease and treatment, with the category ‘none’ as the referent, were 1.18 (95% CI, 1.07-1.30) for mild comorbidity, 1.37 (95% CI, 1.22-1.53) for moderate comorbidity, and 1.54 (95% CI, 1.31-1.80) for severe comorbidity.
We compared comorbidity in patients with salivary gland cancer with comorbidity in patients with other head and neck cancers. Because these data were not matched on age or sex, a valid comparison may be hindered. In analyzing at the comorbidity distribution, it appears that patients with salivary gland cancer have significant less comorbidity. A second analysis in which we tried to compare the comorbidity in our database and the comorbidity in cohort of patients with prostate cancer indicated that there were many similarities, although the distributions were not statistically similar.
In conclusion, to our knowledge, this is the first study concerning the prevalence and relevance of the prognostic comorbidity variable ACE-27 in patients with salivary gland cancer. For this selected group of patients, the ACE-27 index was correlated with age and sex. The treatment performed was influenced by ACE-27 comorbidity only in patients with grade 3 comorbidity. Overall survival was correlated strongly with the ACE-27 index; however, disease-free survival was not. Comparison with other studies in which the effect of comorbidity on head and neck cancer was studied indicated that patients with salivary gland cancer have less comorbidity. Their comorbid status appeared to be reasonably comparable to that among patients with other nonsmoking- and nonalcohol-related cancers. In a future article, we plan to use the current data to construct predictive models for overall survival.
- 8Inclusion of comorbidity into oncology data registries. J Registry Manage. 1999; 26: 66–70., , , , .
- 9Salivary gland carcinoma: independent prognostic factors for locoregional control, distant metastases, and overall survival: results of the Dutch Head and Neck Oncology Cooperative Group. Head Neck. 2004; 26: 681–692; discussion 692–693., , , et al.
- 12American Joint Committee on Cancer. Major salivary glands (parotid, submandibular, and sublingual). In: GreeneFL,PageDL,FlemingID, et al, eds. AJCC Cancer Staging Manual,6th ed. New York, NY: Springer-Verlag; 2002: 53–58.
- 33Determinants of survival in parotid gland carcinoma: a population-based study. Am J Otolaryngol Head Neck Med Surg. 2005; 26: 39–44., .
- 36A cross-comparison of retrospective notes extraction and combined notes extraction and patient interview in the completion of a comorbidity index (ACE-27) in a cohort of United Kingdom patients with head and neck cancer. J Laryngol Otol. 2002; 116: 937–941., .