Head and neck cancers in France: an analysis of the hospital medical information system (PMSI) database

In Europe, head and neck cancer (H&N) accounted for approximately 143 000 new cancers in 2007 and were responsible of more than 68 000 deaths [1]. They have broadly varying rates of incidence and mortality around the world. In North America and Europe, tumours usually arise from the oral cavity, oropharynx, or larynx, whereas nasopharyngeal cancer is more common in the Mediterranean countries and in the Far East. France is particularly concerned by head and neck cancers [2]. With 16 005 new cases in 2005, they represented 3.8% of all incident cancer cases. They are the fourth highest incidence of cancers in men and the eleventh highest incidence in women. In terms of mortality, 5 406 related deaths occurred in 2005, including 82% in men. It represented 2.7% of annual cancer deaths and the tenth more lethal cancer [3, 4].

These cancers are more common in men and in people over age 50. Head and neck cancer refers to a group of cancers originating from the upper aerodigestive tract, including oral cavity, oropharynx, pharynx other than oropharynx (nasopharynx, hypopharynx), larynx, salivary glands, and other sites located in the head and neck area. Head and neck cancers are strongly associated with environmental and lifestyle risk factors. Use of tobacco is the largest risk factor for H&N cancers [5, 6]. During the past two decades, the role of tobacco in the incidence of head and neck cancer has however decreased, thanks to public health efforts at tobacco control and reduction in the prevalence of cigarette smoking. Alcohol consumption is frequently associated with tobacco use as a co-factor in oncogenic risk, especially in oral cavity and hypopharynx cancers.

Moreover, numerous studies have highlighted the increasing role of the sexually transmitted human papillomavirus (HPV) in the incidence of H&N cancers. Contribution of HPV in H&N cancers varies according to tumour site location. Most HPV positive cancers occur in the oropharynx [7], in particular tonsil and base of the tongue. HPV could account for 33% to 72% of oropharynx cancers [8, 9]. Tonsillar cancers have been reported to have the highest prevalence rate of HPV-DNA (more than 50%) [10–12]. HPV positive rates are lower in other localizations, ranging from 5 to 16% for larynx [13, 14], 24% in a report including hypopharynx carcinomas in larynx [8, 15] and 23.5% in oral cavity. The wide variations in HPV prevalence which are reported may depend on the HPV diagnostic methodologies, especially in earlier studies.

Treatment of head and neck cancer depends on the initial localization of the tumor, on patient's comorbidity and on potential side effect of treatment. Surgical resection, radiotherapy, radiochemotherapy, induction chemotherapy, radiobiotherapy (with anti epidermal-growth-factor receptor (EGFR) like cetuximab and other anti EGFR), are the therapeutic methods in locally advanced cases, the most frequent mode of presentation of H&N cancers. The overall prognosis remains poor. In addition, patients frequently suffer from other co-morbidities. A number of advances in therapies for head and neck cancers have been developed in the past several years, including the availability of anti-EGFR agents, induction chemotherapy, new microsurgical and reconstructive surgery techniques or accelerated radiotherapy.

The economic impact of head and neck cancers is poorly documented. The cost of illness has been estimated for some specific countries, like the UK [16], the USA [17], the Netherlands [18], Greece or Germany [19], using various methodologies. Moreover, as these cancers presented great variation in terms of localisation and incidence rate worldwide, the use of economic estimations performed in others countries may be hazardous. In 2004, the French National Institute of Cancer (Institut National du Cancer-INCa) has estimated that the global direct medical costs related to cancer amounted to 11 milliards euros (Md€), including 7 Md€ due to hospitalisation costs, whereas indirect costs due to loss of productivity and premature death were estimated at 17 Md€ [20].

The specific cost related to head and neck cancers in France has however never been documented. The objective of the present study was to assess the annual number of patients hospitalised for head and neck cancers in France and to estimate the economic burden associated with their management.

This survey was designed to estimate the annual number of patients hospitalized for head and neck cancers in France in 2007 and their related management costs. The study showed that the impact of head and neck cancers was important, as, in 2007 there were 36 268 patients hospitalised, corresponding to 60 200 hospital stays and 287 846 sessions of chemo- or radiotherapy. Oropharynx cancer was the most frequent (28% of patients), followed by oral cavity cancer (25% of patients). The study showed that annual hospital cost, (including hospitalisation and expensive drugs) of these patients appeared to be very high, close to 306 millions €. Men are particularly concerned by H&N cancers, as they represented 81% of all patients treated and 83% of all expenses (254 millions €).

Data were extracted from the PMSI database, as it presented a high exhaustiveness and a high level of quality. As discharge reports are the basis of hospitals funding and their reporting is mandatory, the PMSI database contained more than 98% of all public and private hospital stays in 2007, then limiting the sampling errors. Moreover, as all diagnoses are coded using the ICD-10 classification, all discharge reports are standardized, allowing to perform extraction of stays by disease. Coding errors is expected to be rare as this database constitutes the prospective financial system based on hospital activity, and since financial resources are directly related to the coded information. Coding quality in the PMSI is expected to be high and is guaranteed by controls performed annually in each hospital. The PMSI database is then widely used to provide relevant information on the burden of hospitalizations in France [24, 25] and it can be used to evaluate the annual burden of hospitalization of head and neck cancers in France.

The PMSI database also presented few limitations. First, it was not possible to evaluate with precision the number of sessions of radiotherapy performed in the private setting, as they are not included in the PMSI database. This number was calculated using the SAE database, after the application of the ratio between public and private radiotherapy sessions. The calculation of the ratio was however not specific to head and neck cancers, but global for all cancers. Moreover, the global number of patients hospitalized for head and neck cancers in 2007 may be underestimated, as it was not possible to link the radiotherapy sessions performed in the private sectors with the stays from the PMSI database.

This administrative database only concerned patient who are hospitalized at least one time per year. Patients with head and neck cancers managed exclusively in an outpatient setting are not included in the present analysis, leading therefore to a potential underestimation of the total annual number of cases of H&N cancers. Moreover, despite presence of an identification number, it is not possible to distinguish incident cases (newly diagnosed patients, in active cancer treatment, with expensive cancer treatment cost) from the other cases (patients in maintenance treatment, with less expensive treatment cost). Thus, the estimated mean cost per hospitalized patient reflects the average cost of management of all cases admitted to hospital (prevalent cases). This may be lower than the cost per patient of management of new cases only, as shown by van Agthoven et al. (17), who demonstrated that the cost of H&N cancers was € 21 858 for the two years following diagnosis and it decreased to 423 € during the 10 years of follow-up. Our estimations are similar with those for European countries: the total cost to society from head and neck cancers ranged from € 5.5 millions in Greece (for 650 cases per year) to € 893 millions in Germany. It is however difficult to make a comparison, given the differences in terms of incidence and structure of the Sickness Fund.

The use of PMSI database leads to underestimation of the burden of disease, as it is restricted to hospitalization costs. Other categories of cost, like expensive drug, indirect or outpatient costs were estimated by using other sources of data. Cost of expensive and innovative drugs, like cetuximab or docetaxel, was estimated using the FICHCOMP database from the year 2008 (as this database was not yet available in 2007). It was also not possible to directly evaluate the indirect costs related to loss of productivity. In France, the valuation of indirect costs used the budgetary approach which relates to the budget impact for the National Sickness Fund. The monetary compensation served to patients to offset partially their loss of income. This amounted to €46.2 per day of sick leave (calculated by dividing the total amount of compensation paid in Year 2007 by the total number of sick leave days (data provided by the National Health Accounts)). In the present study, an estimation of indirect costs can be calculated using results from the INCa report 2007 [20], which showed that the proportion of patients who stop working because of head and neck cancers was 91% and the mean length of sick leave was 120 days. According to these assumptions, indirect costs amounted to about €76 millions. These indirect costs only represent the part supported by Sickness Fund. From a societal perspective, other indirect costs, due to premature death, would also have to be considered. Moreover, except for patients requiring a hospitalization, cost of management of side-effects of treatment for H&N cancers, like dysphagia or mucositis, which are usually managed in ambulatory care, were not evaluated in the present study. For example, the cost of treatment of dysphagia was evaluated at $153 to $710 [26], whereas it reached € 3 480 to 4 764 per episode of mucositis [27]. Moreover, Penel et al [28] estimated that 31% of patients presented nosocomial infections or pneumonia following H&N surgery, inducing an extra-cost of management estimated at respectively 16 000 and 17 000 €, as compared to patients without nosocomial infections or pneumonia. This extra-cost was not included in the present analysis, because of the risk of double-counting process (as many nosocomial infections are managed in hospital for H&N patients). Concerning the outpatient costs, as no specific data on outpatient management for head and neck cancers were available to our knowledge, we used estimates of the management costs for H&N tumours, published by Weill et al [29], reporting the frequency and cost of 30 long-term disorders in France based on an analysis of national health insurance reimbursement databases. We retrieved from this study that the outpatient costs (including visits, drugs, exams, transportation) represented 31.6% of global cost in patients suffering from head and neck cancers. In the present study, the annual outpatient cost can then be estimated at €131 million. The total cost of management of head and neck cancers, from a Payer perspective, then amounted to €531 million per year, including hospitalisation, expensive drugs, outpatient and indirect costs. Some costs, like rehabilitation cost, were not estimated as no data are available. For head and neck cancers patients who had devastating physical and functional changes, rehabilitation care plays an important role on restoring function and assisting the patient to achieve an acceptable quality of life. It then appeared important to evaluate their cost.

Given the considerable burden of disease induced by head and neck cancers, it appears important to identify strategies allowing to reduce costs and to save money. Nevertheless, at this time, there is no data indicating if prevention (reduction of alcohol and smoking, HPV incidence), screening, targeted treatment will allow to reduce the costs in the future. The cost-benefit of these strategies may be evaluated in the specific context of head and neck cancers.

There is a growing evidence for a causal role for HPV (mainly HPV 16) in a subset of H&N cancers [11]. The proportion of HPV-positive oral cancers has significantly increased over the past decades, in parallel with a decline in HPV-negative cancers [10, 30]. HPV vaccination has demonstrated a high prophylactic efficacy against HPV 16 and 18 related pre-cancerous lesions of the cervix, vulvar and vaginal. One can expect that HPV vaccination might extend protection against other cancers associated to HPV 16 and 18. These study's data could provide useful information for decision-makers in order to assess the potential additional effect of the HPV vaccination on H&N cancers.