The current analysis included 20,618 new diagnoses of OPC and 9335 death cases. Our data showed that the ASRIW of OPC was 2.22/100,000 and the ASRMW was 0.94/100,000 in China. Additionally, the overall ASRIC of OPC significantly increased by 6.2% annually during 2003–2006 (P < 0.05), but remained stable during 2007–2012. The ASRIC and ASRMC for males and in rural areas significantly increased in the recent decade (P < 0.05); the incidence significantly increased by 30.7% in males (P < 0.01) and by 61.1% in rural areas (P < 0.01). Nevertheless, ASRIC and ASRMC remained stable in females during those periods. The ASRIW of OPC in China was slightly above the average level of Colombia, Costa Rica, and Ecuador , but lower than the United States of America (15.6/100,000 for males and 6.1/100,000 for females) . In contrast, the incidence of OPC in males is high in North France with a rate of 42.3/100,000 in males in the Somme and Bas Rhin regions .
Tobacco smoking, alcohol consumption, and human papillomavirus types 16 and 18 (HPV16/18) infections were identified as the major risk factors for OPC, with tobacco smoking and alcohol consumption having synergistic effects [15,16,17]. Tobacco use is a mutagenic factor that increases the risk of oral cavity and pharyngeal cancers . China constitutes approximately 40% of the world’s total tobacco use, predominantly in males, making China the largest tobacco consumer worldwide . Additionally, a large increase in consumption was observed in urban rather than rural areas over the past three decades . It might partly explain the reason why the incidence of OPC was higher in urban areas than in rural areas. In addition, although the mechanisms for alcohol-related carcinogenesis were not completely clarified, the IARC listed both alcoholic beverages and acetaldehyde, including its major metabolites, as human carcinogens. Alcohol drinking is an established risk factor for head and neck cancer, and this relationship was even stronger among cancers of the oropharynx and hypopharynx . Li et al.  reported that the proportions of current drinkers among Chinese males, females, and the whole population were 55.6%, 15.0%, and 35.7%, respectively. Additionally, among current drinkers, the proportions of excessive, frequent, and binge drinkers were 59.1%, 26.0%, 53.7% in urban areas and 64.7%, 26.4%, 59.3% in rural areas, respectively. Furthermore, a significant association between heavy alcohol drinking and malnutrition was observed, which may have led to reduced intake of vegetables, fruits, and some other foods with cancer-preventive effects . Additionally, the local cytotoxic activity of ethanol might explain the synergistic effect of tobacco and alcohol consumption on the risk of OPC. The cytotoxic effect of ethanol on the cells lining these tissues activated the division of stem cells located in deeper layers to replace dead cells. By activating their division, alcohol leaves the DNA of stem cells highly exposed to the DNA-damaging activity of tobacco carcinogens .
In head and neck squamous cell carcinoma, HPV has been regarded as an important pathogen, in addition to alcohol and tobacco consumption . An increase in the incidence of oropharyngeal squamous cell carcinoma (OPSCC) was observed in some countries [3, 4], and this has been attributed to HPV infection . Currently, the natural history of oral HPV infection has not been fully studied . Nevertheless, some progress has been made in recent decades. It was reported that HPV DNA was detected in approximately 19%–75% of OPSCCs worldwide, 85%–95% of which belong to the HPV-16 type [27, 28]. A recent study performed in China also showed that HPV infection was detected in 16.7% of OPSCC specimens , which was similar to the detection in 18%–36% of OPSCCs reported in studies from developed countries . The development of HPV-associated OPSCC was found to be associated with high-risk sexual behaviours, including an increased number of sexual partners/oral sex partners, and homosexual intercourse . In addition, a significantly higher prevalence of oral HPV infection was found in subjects who had performed oral sex than in subjects who had not, consistent with previous studies that showed that oral sex might be a risk factor for oral HPV infection . Using data from cancer registries and data of HPV prevalence, recent studies in Australia , the United States , and Sweden  have indicated that changes in sexual behaviours among recent birth cohorts have led to increased oral HPV exposure and to an increased incidence of OPC. Nevertheless, the natural history of oral HPV infection is still largely unknown, and ongoing research is underway to explore why males are more susceptible to acquisition and persistent HPV infection than females .
The present study demonstrated temporal trends of OPC incidence and mortality according to the ICD-10 classification. Joinpoint analysis provided a clear picture of temporal patterns in different segments of time. As such, we were able to show the significant changes in trends that have occurred during the study period. Using the APC model, we detected significant differences in incidence and mortality trends of OPC between males and females. Furthermore, OPC incidence significantly increased in rural and urban areas among males. The present study confirmed the increases in incidence and mortality of OPC in the last decade, especially for males in rural areas. Nevertheless, the ASRIC and ASRMC were stable over the study period for females, which may be because of an increased awareness of the need of sunscreen or low smoking rates in females . Our results were consistent with previous studies on trends of OPC in other countries [16, 34, 35].
Our observations of increasing OPC incidence in males and in rural areas help us better understand its epidemiological situation in China and provide more precise scientific information for its control and prevention. The main strength of our study is the population-based design with a large sample size of OPC patients in China. Furthermore, we included 32 cancer registries that had data of sufficient quality over the 10-year period (2003–2012) for inclusion in incidence and mortality trend analyses. Additionally, our epidemiologic study was nationwide and covered a period of 5 years.
Nevertheless, there were limitations that should be addressed. First, the information submitted by the cancer registries was incomplete because the private behaviours such as oral/vaginal sexual behaviours were not known. Second, it did not contain information of other risk factors, such as individual tobacco and alcohol consumption. Therefore, the effect of these factors on temporal trends could only be assessed. Finally, the small number of cases in some subgroups (i.e., small number of death cases in the Western region) may preclude further analysis by age in females.