Graphic presentation of diagnostic delay hypotheses.
Carvalho AL, Pintos J, Schlecht NF, Oliveira BV, Fava AS, Curado MP, Kowalski LP, Franco EL. Predictive Factors for Diagnosis of Advanced-Stage Squamous Cell Carcinoma of the Head and Neck. Arch Otolaryngol Head Neck Surg. 2002;128(3):313-318. doi:10.1001/archotol.128.3.313
To identify the predictive factors (with emphasis on diagnostic delay) associated with the diagnosis of an advanced–clinical stage head and neck cancer.
Cross-sectional study of patients with head and neck cancer originally recruited for a case-control study.
Three referral oncological centers in metropolitan areas in southern Brazil: São Paulo, Curitiba, and Goiânia.
The study population comprised 679 patients recently diagnosed as having a previously untreated head and neck squamous cell carcinoma.
Main Outcome Measure
Diagnosis of advanced disease (clinical stage III-IV) head and neck cancer.
Patients with laryngeal and hypopharyngeal cancers were more likely to be diagnosed as having advanced disease than those with lip, oral, and oropharyngeal cancers (88.0% vs 74.6%) (P<.001). Patient delay was inversely associated with clinical stage at diagnosis in patients with the same cancers, while professional delay was directly associated with a higher risk of advanced clinical stage at diagnosis (P = .001 and P = .006, respectively). In the analysis of laryngeal and hypopharyngeal cancer, both patient and professional delays were associated with advanced disease, with patient delay being a stronger predictive factor than professional delay.
Clinical stage at diagnosis was associated with sociodemographic characteristics, patient delay, and professional delay. Our results indicate that continued educational programs for the population and health care professionals regarding the identification of early symptoms of head and neck cancers are warranted.
THERE IS general consensus that clinical stage at the time of diagnosis is the most important predictor of recurrence and death in patients with cancers of the head and neck.1- 6 Clinical stage at the time of diagnosis is influenced by several clinical and sociodemographic variables, including patient delay in consulting a health care professional and professional delay in diagnosing and treating the disease.7- 9
Patient delay in seeking professional health care and professional delay regarding cancer diagnosis have been found to predict clinical stage of oral cancer at diagnosis,10- 12 with direct consequences on the complexity and cost of treatment and on disease prognosis. Patient delay in consulting a health care professional can be attributed to 2 major factors: patient delay in recognizing the signs or symptoms of cancer or difficulties in accessing professional care.13,14 Professional delay can be attributed to failure in recognizing the signs and symptoms suggestive of cancer.15,16 We evaluated the relationship between sociodemographic and clinical factors on clinical stage in a large case-control study in Brazil to identify the predictive factors (with emphasis on diagnostic delay) associated with the diagnosis of an advanced head and neck cancer.
Between February 1986 and January 1989 we conducted a case-control study of head and neck cancers in 3 metropolitan areas of southern Brazil: São Paulo, Curitiba, and Goiânia. All patients presenting with a diagnosis of primary squamous cell carcinoma of lip, oral cavity, oropharynx, larynx, and hypopharynx (International Statistical Classification of Diseases, 10th Revision codes C00-C06, C09-C10, C12-C13, and C32) were eligible for this study. All cases were confirmed histopathologically as squamous cell carcinoma. With the exception of the head and neck surgery service in São Paulo, which was responsible for approximately 20% of all incident cases in the city, patient accrual in the 2 other centers approached 100% of all incident cases in each area for the study period. Data were collected prior to any medical treatment by interview using a 40- to 60-minute structured questionnaire. Interviews elicited detailed information on socioeconomic and demographic variables, dental health, history and habit of tobacco smoking, and lifetime alcoholic beverage consumption. Questionnaires also included information on the date of recognition of first signs or symptoms of the disease and which health professional (dentist or physician) was visited by the patient before being referred to the hospital for definitive diagnosis and treatment. Dates of consultation with each professional were also recorded.
Tumor sites were classified into 2 different groups: tumors visible with the naked eye in a clinical examination, such as cancers of the oral cavity, including the lip, mouth, and oropharynx (group 1) and tumors visible only with special equipment such as a laryngoscope or endoscope, which included tumors of the larynx and hypopharynx (group 2).
Patient delay was defined as the time between date of recognition of first sign or symptom of the disease by the patient and date of first visit to a professional who was qualified to refer the patient for definitive diagnosis and treatment: physician or dentist for group 1 and physician for group 2. Professional delay was defined as the time between this consultation and the date of histopathological diagnosis.
Clinical stage was classified according to the International Union Against Cancer (UICC) criteria17 and grouped as early clinical stage (clinical stage I-II) or advanced clinical stage (clinical stage III-IV). Neck mass was one of the signs recorded and used in the analysis regarding number of symptoms for both groups. However, this sign was not used as a variable in the analysis because almost all patients in these groups had clinically metastatic disease, which were therefore classified as advanced clinical stage.
Statistical analysis was performed using the SPSS statistical software (SPSS Inc, Chicago, Ill). Means and proportions were calculated to describe the distribution of characteristics in the patient population. Odds ratios (ORs) for diagnosis of advanced disease (clinical stage III-IV) and their respective 95% confidence intervals (CIs) were estimated using multivariate logistic regression. Trends for linear association were assessed by including the categorical factors as ordinal variables in the regression models. The regression models were adjusted for the variables found to be important in predicting patient or professional delay. The variables adjusted for in model 1 included age, sex, education, and monthly income, and in model 2, number of symptoms, tobacco and alcohol consumption, and dental health were also adjusted for. Adjustment for patient and professional delays were included in both models. The relationship between patient delay and professional delay time was examined using Pearson product moment correlation after transforming both variables into a logarithmic scale to correct for the skewness in the distributions.
Of the 679 patients initially recruited, 3 were excluded owing to their inability to remember the onset date of their signs or symptoms. The distribution of patients according to metropolitan area was as follows: 204 patients (30.2%) from São Paulo; 310 (45.9%) from Curitiba, and 162 (23.9%) from Goiânia. Of the 676 patients, 595 (88.0%) were male adolescents or men, 569 (84.2%) were of European descent, and ages ranged from 15 to 82 years (median, 58 years). There were 60 patients (8.9%) with cancer of the lip, 219 (32.4%) with cancer of the oral cavity, 138 (20.4%) with cancer of the oropharynx, 96 (14.2%) with cancer of the hypopharynx, and 163 (24.1%) with cancer of the larynx. Patient delay ranged from 1 week to 96 months (median, 3 months). Professional delay ranged from 1 day to 31 months (median, 1 month). Patient and professional delays were inversely correlated (Pearson correlation coefficient, −0.361; P<.001), ie, longer patient delay was linked with shorter professional delay, and vice versa.
Of the 676 patients analyzed, 535 (79.1%) were diagnosed as having an advanced–clinical stage cancer. Patients in group 2 were more likely to be diagnosed as having an advanced–clinical stage cancer than patients in group 1 (88.0% and 74.6%, respectively; P<.001).
Table 1 gives the univariate analysis for the association of selected study variables with the diagnosis of advanced clinical stage cancer. Table 2 gives the association between patient and professional delays with advanced disease among group 1 patients. Both patient and professional delays were associated with advanced clinical stage. While professional delay was positively associated with advanced-stage disease, there was a negative association between patient delay and stage III and IV cancer. When adjusting for age, sex, education, monthly income, number of symptoms, dental health, and tobacco and alcohol consumption, patients who waited 3 or more months before consulting a health professional were less likely to be diagnosed as having advanced disease compared with patients who consulted within the first month (OR, 0.46; 95% CI, 0.2-0.9). On the other hand, patients who experienced a professional delay of more than 3 months had a higher risk of advanced disease compared with a professional delay of less than 1 month (OR, 1.69; 95% CI, 0.7-4.3).
Table 3 gives the association of patient and professional delays for group 2 patients. For this group, both patient and professional delays were positively associated with advanced-stage cancer in the fully adjusted statistical model (model 2, Table 3), although professional delay was not a statistically significant predictor. For patient delays of more than 3 months, the OR for advanced disease was 3.37 (95% CI, 1.0-11.8), and for professional delay of more than 3 months, the OR was 1.67 (95% CI, 0.5-5.9), compared with a delay of less than 1 month. It is interesting that the total delay (time of patient delay plus professional delay) for groups 1 and 2 was not associated with advanced clinical status at diagnosis.
Clinical stage is the most important prognostic factor for patients with head and neck cancer.1- 6 Various factors have been investigated as predictors for a diagnosis of an advanced–clinical stage cancer.7- 9 However, there is controversy in the medical literature regarding the association between patient and professional delays with clinical stage at diagnosis. Although it seems logical that longer patient and professional delays would be associated with a diagnosis of an advanced-stage disease, not all studies of this issue arrived at similar findings. Some studies observed that variables usually associated with patient and professional delays, such as sex, dental status, alcohol consumption, and socioeconomic status, are associated with clinical stage at diagnosis.7,8 Most of the studies, however, could not find a direct association between diagnostic delay and clinical stage. Nonetheless, it is difficult to reject the hypothesis that diagnostic delays contribute to a poor prognosis.1,11
Rubright et al8 found an association between older age and poor dental status with diagnosis of an advanced– clinical stage cancer; however, they did not observe an association between diagnostic delay and clinical stage. Elwood and Gallagher7 observed that high socioeconomic status, low level of alcohol consumption, and regular dental care were associated with diagnosis of an early-stage cancer. They also observed that diagnostic delay was different for oral cancer compared with other head and neck tumors and failed to find an association between diagnostic delay and clinical stage. In agreement with these studies, Wildt et al10 did not find a correlation between patient or professional delays and clinical stage. Two other studies did not observe an association between diagnostic delays and tumor size.18,19 On the other hand, Gorsky and Dayan,9 found a positive association between diagnostic delay and clinical stage in patients with cancer of the lip.
Several factors may explain, at least partially, the weak or null association seen between diagnostic delay and clinical stage. While professional delay can be measured in a relatively accurate way, patient delay is prone to measurement error. The date of onset of symptoms is based on perception, which is highly subjective and may be influenced by several social and cultural factors. Patients often neglect the importance of feeble symptoms, but when they seek medical care for early symptoms, the condition is often labeled as benign and the diagnosis is delayed.20 Patients with fast-growing tumors can be diagnosed relatively rapidly but often after an advanced clinical stage develops given the nature of the disease. On the other hand, patients with slow-growing tumors usually experience long diagnostic delays and are diagnosed as having early-stage disease (Figure 1).11
In our study, we observed that the factors associated with an advanced oral cancer (group 1) are different from those associated with hypopharyngeal and laryngeal cancer (group 2). For oral cancer, female sex and dental status (use of non–well-fitted dentures) were associated with a lower risk of being diagnosed as having advanced-stage disease, while patients who were single, who were current smokers, and those with poor dentition presented a higher risk. For patients with laryngeal or hypopharyngeal cancers, older age was associated with a lower risk of being diagnosed as having advanced-stage disease. Patients with more than 1 symptom were at higher risk of advanced cancer in both groups compared with patients with only 1 symptom. Obviously, different tumor sites will manifest with different symptoms. Tumor symptomology may play an important role in the diagnostic process, given its influence on attention by the patient and professional care.7
In the analysis of diagnostic delay as a predictor of advanced disease, we found that oral cancers were influenced in a different way than laryngeal and hypopharyngeal cancers. For patients with cancer of the lip, mouth, and oropharynx, longer patient delay was associated with lower risk of being diagnosed as having an advanced clinical stage. These results favor the tumor biology hypothesis (Figure 1), according to which patients with slow-growing tumors have a longer delay and are more likely to be diagnosed as having early disease than patients with fast-growing tumors. On the other hand, we found that professional delay was positively associated with clinical stage at diagnosis. This finding favors the logical hypothesis (Figure 1). For patients with laryngeal and hypopharyngeal cancers, both patient and professional delays were associated with advanced clinical stage, also favoring the logical hypothesis. It should be noted that the association of patient delay with advanced disease was substantially stronger than that of professional delay. Two additional factors may contribute to bias the associations of disease burden with patient and professional delays, but their role is difficult to establish in studies such as the present one. The first of these factors is the contribution of the overall health system delay, which includes the professional referral. The patient may seek prompt medical attention for his or her symptoms and the first professional may immediately recognize the sign or symptom, but there may be a delay in referring to the specialist for definitive diagnosis or to establish the diagnosis. The second factor that may contribute to bias the association is related to errors in diagnosis, but we have no indication of that in our study.
In summary, regarding laryngeal and hypopharyngeal cancers, our results support the logical hypothesis, according to which patient and professional delay are associated with a more advanced clinical stage. In the analysis of lip, mouth, and oropharyngeal cancers, professional delay was a predictor of more advanced disease. However, we found a negative association between patient delay and clinical stage. One of the possible explanations for this finding is the tumor biology hypothesis. Future studies assessing the influence of diagnostic delay in the prognosis of patients with head and neck cancers should take into consideration this hypothesis and would benefit from using biomarkers capable of identifying slow- and fast-growing tumors.
Although collecting information on tumor biology markers would have been useful in the present and other investigations, patient and professional delays are undeniably critical factors influencing clinical stage. Our results support the adoption of continued educational programs for the population and health care professionals regarding the identification of early symptoms of head and neck cancers.
Accepted for publication November 27, 2001.
Mr Schlecht is a recipient of a Doctoral Research Award and Dr Franco is a recipient of a Distinguished Scientist Award, both from the Medical Research Council of Canada, Ottawa, Ontario. This work was supported in part by International Scientific Exchange Award 910035/99-7 from the Medical Research Council of Canada–Conselho Nacional de Pesquisa (Dr Carvalho [visiting fellow]).
This study was presented at the annual meeting of the American Head and Neck Society, Palm Desert, Calif, May 15, 2001.
We are indebted to the following participants in the Ludwig Institute for Cancer Research's Upper Respiratory and Digestive System Cancer Study Group: clinical committee—M. B. Carvalho, MD, A. Rapoport, MD, J. Andrade-Sobrinho, MD, G. Ramos, MD, J. L. Kanda, MD, J. F. Gois, MD, J. S. Chagas, MD, and G. A. Teixeira, MD; pathology committee—H. Torloni, MD (coordinator), W. T. Vieira, MD, L. A. Sampaio, MD, and V. M. Cardoso, MD; data acquisition and management—M. E. Silva (social worker); R. N. Pereira (statistician); N. Campos-Filho (statistician); L. Fanes, RN; V. N. Souza, MD; and M. S. Morais, MD.
Corresponding author and reprints: Luiz Paulo Kowalski, MD, PhD, Rua Antonio Prudente, 211, São Paulo 01509-900, Brazil (e-mail: firstname.lastname@example.org).