Trends in mortality rates for oral and oropharyngeal cancer in Spain, 1979–2018

   Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ Esta es la versión aceptada del artículo publicado en: This is a accepted manuscript of a paper published in: Oral Diseases (2022): March 2022 DOI: https://doi.org/10.1111/odi.13754 Copyright: 2020 Wiley Periodicals LLC. El acceso a la versión publicada del artículo puede requerir la suscripción de la revista. Access to the published version may require subscription. “This is the peer reviewed version of the following article: Durán-Romero, Antonio José, Infante Cossío, Pedro Antonio, Pereyra-Rodríguez, José- Juan (2022), Trends in mortality rates for oral and oropharyngeal cancer in Spain, 1979–2018 . Oral Diseases , V. 28, Nº 2, 336 - 344 . which has been published in final form at https://doi.org/10.1111/odi.13754 This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-Archived Versions. This article may not be enhanced, enriched or otherwise transformed into a derivative work, without express permission from Wiley or by statutory rights under applicable legislation. Copyright notices must not be removed, obscured or modified. The article must be linked to Wiley’s version of record on Wiley Online Library and any embedding, framing or otherwise making available the article or pages thereof by third parties from platforms, services and websites other than Wiley Online Library must be prohibited." Figure 1 Figure 2 Figure 3 1 APPENDIX Thisappendixhas been providedby the authorsto give readersadditionalinformationabouttheir work. Supplement to: Duran-Romero A.J., Infante-Cossio P., & Pereyra-Rodriguez J.J. Trends in mortality rates for oral and oropharyngeal cancer in Spain, 1979-2018 APPENDIX LEGENDS Supplementary Table S1. OCOPC mortality in Spain between 1979-2018: deaths by genders and locations. Supplementary Table S2. Goodness-of-fit test for different APC models for OCC and OPC in Spain between 1979-2018. Supplementary Figure S1. Sex ratio (male/female) for all ages: a) OCC; b) OPC. 2 Supplementary Table S1. OCOPC mortalityin Spain between 1979-2018: deaths by genders and locations. a) Male Year LipOral cavity sub-site Tongue Gingiva Floor of mouth Other sites o f the mou th O rophary nx Tot al oral cavity Total o rophary nx 1979 85 263 10 50 102 66 510 66 1980 72 273 13 58 111 87 527 87 1981 66 295 12 62 170 100 605 100 1982 65 311 11 92 184 102 663 102 1983 67 324 883 158 117 640 117 1984 66 372 11 83 103 136 635 136 1985 65 369 19 91 116 155 660 155 1986 67 398 11 118 138 164 732 164 1987 75 423 19 115 136 208 768 208 1988 77 419 17 130 143 228 786 228 1989 77 425 14 137 159 219 812 219 1990 78 419 19 142 145 260 803 260 1991 72 466 19 149 156 297 862 297 1992 73 448 21 138 169 304 849 304 1993 91 415 23 168 185 306 882 306 1994 83 426 21 166 160 329 856 329 1995 59 423 24 147 186 328 839 328 1996 61 398 20 148 169 288 796 288 1997 53 451 23 179 183 337 889 337 1998 53 407 18 161 172 346 811 346 1999 43 395 13 147 182 352 780 352 2000 51 379 15 147 183 386 775 386 2001 50 380 11 161 191 366 793 366 2002 66 433 11 158 186 314 854 314 2003 50 413 19 154 173 366 809 366 2004 46 406 18 143 168 354 781 354 2005 61 371 14 130 203 348 779 348 2006 44 352 14 139 181 332 730 332 2007 47 380 18 156 172 330 773 330 2008 40 350 15 119 149 352 673 352 2009 39 361 14 131 171 319 716 319 2010 41 385 10 122 191 359 749 359 2011 43 404 18 115 186 356 766 356 2012 40 364 22 121 188 357 735 357 2013 46 380 24 117 193 386 760 386 2014 45 388 19 107 198 396 757 396 2015 40 356 25 111 185 398 717 398 2016 42 372 24 126 190 350 754 350 2017 48 356 24 111 219 441 758 441 2018 42 324 22 105 211 420 704 420 TOTAL 2 , 329 15,274 683 5,037 6,765 11,659 30 ,0 88 11,659 3 b) Female Note: OCOPC: Oral cavity and oropharyngeal cancer. Year Oral cavity sub-si te Li p Tongue Gingiva Floor of mouth Other sites o f the mou th O rophary nx Tot al oral c avi ty Total o rophary nx 1979 11 47 5 4 32 699 6 1980 15 57 6 5 32 12 115 12 1981 12 41 3 6 61 4123 4 1982 10 53 3 4 64 10 134 10 1983 14 49 4 5 47 9119 9 1984 12 55 0 5 34 15 106 15 1985 10 58 9 3 17 14 97 14 1986 15 75 7 7 25 17 129 17 1987 11 89 5 3 24 23 132 23 1988 10 75 10 12 22 19 129 19 1989 874 11 10 43 21 146 21 1990 12 75 721 36 24 151 24 1991 11 87 815 37 26 158 26 1992 13 100 13 17 31 21 174 21 1993 14 101 11 12 42 17 180 17 1994 991 12 13 51 25 176 25 1995 8101 814 49 20 180 20 1996 14 94 17 13 65 29 203 29 1997 14 96 722 61 18 200 18 1998 10 105 10 21 59 23 205 23 1999 13 89 14 19 58 28 193 28 2000 15 92 927 70 28 213 28 2001 7113 20 17 55 35 212 35 2002 16 104 821 76 33 225 33 2003 6129 921 75 40 240 40 2004 12 117 15 20 77 55 241 55 2005 14 119 14 32 74 44 253 44 2006 9127 20 33 57 38 246 38 2007 13 132 17 33 85 51 280 51 2008 13 121 18 27 79 52 258 52 2009 8129 19 24 90 60 270 60 2010 10 150 13 29 110 55 312 55 2011 20 168 17 30 99 62 334 62 2012 13 153 21 28 111 62 326 62 2013 14 193 19 31 118 76 375 76 2014 14 151 22 23 123 70 333 70 2015 11 171 17 23 143 70 365 70 2016 19 173 40 36 151 68 419 68 2017 15 194 25 40 141 74 415 74 2018 20 182 33 37 162 56 434 56 TOTAL 495 4,330 526 763 2,786 1,410 8,900 1,410 4 Supplementary Table S2. Goodness-of-fit test for different APC models for OCC and OPC in Spain between 1979-2018. Model Df Deviance p AIC df Deviance p AIC Males Females Note: Df: degrees of freedom; AIC: Akaike information criteria; OCC: Oral cavity cancer; OPC: Oropharyngeal cancer. OPC A ge-model 98 1391 .32 <0. 001 2003 . 17 98 456 .00 <0. 001 864 .0 4 A ge-drift model 97 1370 .21 <0. 001 1984 . 06 97 325 .50 <0. 001 735 .5 3 A ge-period model 91 683 .46 <0. 001 1309 . 31 91 271 .17 <0. 001 693 .2 1 A ge-period-cohort model 72 77 .33 0.312 741 .17 72 77 .04 0.320 537 .0 7 A ge-cohort model 78 478 .18 <0. 001 1130 . 03 78 135 .03 <0. 001 538 .0 7 OCC Age-model 98 3227 .37 <0. 001 3988 . 34 98 233 .32 <0. 001 855 .3 2 Age-drift model 97 1371 .82 <0. 001 2134 . 78 97 183 .19 <0. 001 807 .2 0 Age-period model 91 724 .26 <0. 001 1499 . 23 91 160 .94 <0. 001 796 .9 3 Age-period-cohort mod el 72 101 .52 0.013 914 .49 72 81 .55 0.207 755 .5 5 Age-cohort model 78 577 .09 <0. 001 1378 . 06 78 100 .67 0.043 762 .6 7 5 Supplementary Figure S1. Sex ratio (male/female) for all ages: a) OCC; b) OPC. 1  TITLE: Trends in mortality rates for oral and oropharyngeal cancer in Spain, 1979-2018 RUNNING HEAD: Mortality trends for oral and oropharyngeal cancer AUTHORS: AJ Duran-Romero 1 P Infante-Cossio 2 JJ Pereyra-Rodriguez 1 1 Department of Dermatology, Virgen del Rocio University Hospital, Seville, Spain 2 Department of Oral and Maxillofacial Surgery, Virgen del Rocio University Hospital, Department of Surgery, School of Medicine, University of Seville, Seville, Spain Correspondence Pedro Infante-Cossio, Department of Surgery, School of Medicine, Seville, Spain. Email: [email protected] 2  Abstract Objective: To analyse mortality rate trends in Spain for oral cavity and oropharyngeal cancer (OCOPC) from 1979 to 2018, evaluating differences between oral cavity cancer (OCC) and oropharyngeal cancer (OPC). Materials and Methods: Death certificates and mid-year population data were collected from the Spanish National Statistics Institute. Age-standardized mortality rates were calculated using the direct method. Joinpoint regressions were used to identify significant changes in mortality trends. Independent effects of age, period, and cohort (APC) were estimated. Results: 52,057 deaths were registered from OCOPC, 38,988 from OCC and 13,069 from OPC between 1979-2018. While OCC mortality rates declined, OCOPC rates increased slightly and OPC significantly. OCC and OPC mortality reached their highest values between 1979-1992, when OCC rates began to decrease in males and OPC levelled off until 2018. Lip cancer suffered the highest drop. APC models showed a mortality increase in males and females from 40-45 and 50-55 years of age, respectively. Conclusions: Favourable OCC mortality trends was plausibly influenced by decreased tobacco/alcohol consumption, while OPC rise was probably associated with increased human papillomavirus infection. The importance of closely monitoring these cancers by age group, sex and location, and continuing with preventive measures against known risk factors, is highlighted. Keywords Mortality rates, oral cavity cancer, oropharyngeal cancer, age-period-cohort analysis, Spain 3  INTRODUCTION Oral cavity cancer (OCC) and oropharyngeal cancer (OPC) are two closely related malignancies traditionally combined under the term oral cavity and oropharyngeal cancer (OCOPC). They represent a worrying health problem as, in 2018, OCC and OPC were estimated to have caused 177,400 (119,700 males, 7,700 females) and 51,000 deaths (42,100 males, 8,900 females) worldwide, respectively (Ferlay et al., 2019). Although the incidence and mortality from OCOPC had been gradually increasing over the past decades, a decrease in the rate of male deaths from OCC has been reported in Southern European countries over the past 30 years (Bosetti et al., 2020), plausibly associated with a drop in cigarette and alcohol consumption (Chaturvedi et al., 2013). In contrast, OCC rates in males reached high values in recent decades in several Central and Eastern European countries, as well as rates of OPC in the US and most European countries (Chaturvedi et al., 2011; Wang & Palefsky, 2016). These divergent patterns hypothesized different exposure in diverse geographical areas to risk factors such as tobacco and alcohol and, probably, human papillomavirus (HPV) (Prue, Lawler, Bakerm & Warnakulasuriya, 2016; Tachezy et al., 2005; Valls-Ontañón et al., 2019). The association of HPV with OCOPC is variable since it represents a well-established causal factor for OPC, while its role is unclear in OCC (Lai et al., 2017). HPV DNA has been detected in up to 80% of OPC (Gillison, Chaturvedi, Anderson & Fakhry, 2015), particularly genotype 16 (Weatherspoon, Chattopadhyay, Boroumand & Garcia, 2015). A meta-analysis from 2015 estimated that 45.8% of OPC and 24.2% of OCC were attributable to HPV infection based on detection of HPV DNA by PCR (Gillison et al., 2015). In Spain, the HPV positive population is relatively low, with an estimated prevalence at the late 1990s of 2.2% of the population for high-risk genotypes (De Vuyst, Clifford, Li & Franceschi, 2009), but it seems to have increased over time (Rodrigo et al. al., 2014). Information on mortality rates from OCOPC in Spain is scarce (Bonifazi et al., 2011; Garavello et al., 2010). Nieto and Ramos (Nieto & Ramos, 2002) reported trend mortality rates similar to the rest of Southern/Mediterranean European countries, showing a gradually increase in death rates from the 1950s to the mid-1990s, when they began to decline until the beginning of 21st century, particularly in males. With the aim of providing an update on recent trends in mortality from OCOPC in Spain, the present study analysed the death rates between 1979 and 2018, considering gender, age and anatomical cancer sub-site, using data from the official Spanish death records. We compared the general patterns and mortality trends between OCC and OPC combined and separately over a 40-year 4  period and discussed the possible influence on mortality of different risk factors involved in these tumours. MATERIALS AND METHODS Data collection Death records and mid-year population data were collected from the Spanish National Statistics Institute (http://www.ine.es) for the period 1979-2018. Death certificates containing any of the following locations with their respective codes from the 9th or 10th edition of the International Classification of Diseases (ICD-9, ICD-10) as cause of death were considered for the present study. The codes included for OCC were: lip (140, C00), tongue (141, C01-02), gingiva (143, C03), floor of the mouth (144, C04), and other sites of the mouth (145, C05-06); and for OPC: oropharynx (146, C09-10). Malignant neoplasms of the nasopharynx, hypopharynx, and major salivary glands were excluded. Statistical analysis Age-adjusted mortality rates were calculated by the direct method using the 2013 European standard population (European Commission Revision of the European Standard Population- Report of Eurostat's Task Force, 2013) as reference for each year, for combined OCOPC, separately for OCC and OPC, and for each cancer sub-site. These calculations were made for the entire country and stratified by gender and three age groups (<35, 35-64 and >64 years). Mortality rates were expressed as deaths per 100,000, and the software used to obtain them included: Epidat3.1®, Microsoft® Excel and SPSS Statistics 25®. In addition, sex ratio was calculated to assess mortality differences between genders, defined as the ratio of male to female mortality rate. Joinpoint regression models were used to identify significant trend changes in mortality rates during the studied period and to estimate the average annual percent change (AAPC). For this purpose, the Joinpoint software developed by the Surveillance Research Program of the US National Cancer Institute was employed (National Cancer Institute. Joinpoint Regression Program, 2020). The maximum Joinpoints allowed in the analysis were 5. Age-period-cohort (APC) models were fitted to assess independent effects of those items on 5  mortality. To address the problem of "non-identifiability", penalty functions proposed by Decarli et al. (Decarli, Vecchia, Malvezzi & Micciolo, 2014), based on the Osmond and Garner model (Osmond & Gardner, 1982), were used. APC models were estimated using Poisson regression and generalized linear interactive modelling macros for R® software provided by the authors to complete the calculations. We divided data into 5-year periods and age quinquenniums to carry out these estimates. Additionally, goodness-of-fit of APC models was compared. RESULTS Between 1979 and 2018, 52,057 deaths from OCOPC were registered in Spain, of which 38,988 were separately attributed to OCC (30,088 males, 8,900 females) and 13,069 to OPC (11,659 males, 1,410 females). In 2018, there were 1,614 deaths from OCOPC (OCC=1,138, OPC=476), with an OCC/OPC ratio=2.39 (Supporting Information Table S1). Regarding age-standardized mortality rates from 1979-83 to 2014-18 (Table 1), the main findings to be highlighted were: 1) The overall mortality rates for OCOPC grew slightly from 3.24 to 3.45/100,000 between 1979-1983 and 2014-2018; 2) Interestingly, while mortality rates for OPC rose significantly from 0.37 to 1.03/100,000, OCC rates showed a reduction from 2.86 to 2.42 /100,000 in the last 4 decades; 3) Joinpoint analysis showed two trend changes in OCOPC mortality rates during three periods: a) from 1979 to 1992, mortality rates grew with its highest AAPC=3.3 (95% CI 2.7-3.9); b) from 1992 to 2009, they experienced a sustained decrease; and c) from 2009 to 2018, they levelled off; 4) Joinpoint analysis for OOC and OPC revealed that, after increasing from 1979 to 1991, mortality from OCC showed a dropping phase while mortality from OPC levelled off until 2018. Relating to sex-stratified analysis (Table 2), male mortality rates for OCOPC and OCC declined from 1979-1983 to 2014-2018, while mortality for OPC increased. By contrast, female mortality rates rose for OCOPC, OCC and OPC, with OPC showing the most important increase in relative terms, having rates 5 times higher at the end of the period. The Joinpoint analysis for OCOPC in males defined two different trend periods: from 1979 to 1991, in which mortality rates grew to their peak and, from 1991 to 2018, when they decreased significantly. Similar trends were found for OCC. However, for OPC mortality rates, after an initial ascent from 1979 to 1991 and a subsequent fall in the period 1991-2009, a progressive increase was observed until 2018. In females, mortality rates rose steadily from 1979 to 2018, more sharply in OPC. 6  Age group analysis (Table 3) showed a widespread decline of mortality rates in the group <35 year of age, while rates rose in those over 35 years for OPC, more markedly in the group over 64 years. Trend analysis reflected: 1) A sustained reduction for the <35 year group,; 2) A high OPC mortality rate value for the 35-64 year group in the period 1979-1991; 3) An increase for OCOPC and OPC for the >64 year group, particularly the last one that consistently grew since the 1990s. Joinpoints and mortality trends in relation to the general population by gender and cancer location (previously described with Table 1) are depicted in Figure 1. Interestingly, lip cancer declined from 1979 to 2018, with two marked drops in the early 1980s and mid-1990s. In males, lip, tongue, and other locations showed decreasing mortality rate trends, while gingiva and floor of the mouth rose slightly. In females, mortality trends had a sustained increase from 1979 to 2018, except for lip cancer, which showed a steady decrease. With regard to the sex-ratio (Supporting Information Figure S1), the gap between males and females showed an increase in OCC rates until the late 1980s, when a progressive reduction of this gap began, more pronounced for the >64 year group. A similar reduction in the sex ratio was also observed in OPC rates throughout the period, although a growing trend in the gap was detected from 2013. In the analysis of the independent effects of APC on mortality, a constant rise in male mortality was found from 40-45 years of age, and a strong increase in female mortality from 50-55 years of age (Figure 2). Estimates of the age-specific mortality rate reached values close to 35 and 80/100,000 in males, and 15 and 55/100,000 in females, respectively, for OPC and OCC. Focusing on birth cohort effect, death rates rose gradually in females from those born in the 1940s, particularly for OPC. By contrast, males born from the 1940s showed lower mortality rates for OCC, which could also be interpreted as a cohort effect. However, cohort analysis showed two peaks in the 1950s and 1990s for OPC rates in males. Moreover, a dropping mortality trend is seen in males in the 1990s for OCC and OPC, which could be interpreted as a period effect. The degree of goodness for each possible regression model has been calculated for the evaluation of the APC independent effects. The age-cohort model is the best fitting two-factor model for males and females, as it has a lower Akaike information criterion (AIC) and a narrower deviation for OCC and OPC (Supporting Information Table S2). The contribution of OCC and OPC to the total mortality by year and gender may be seen in Figure 3. 7  DISCUSSION This study represents the longest mortality series of OCOPC reported to date in Spain. The most relevant finding confirmed a significant reduction mortality from OCC during the last 40 years in contrast to mortality from OPC which exhibited a clear increase. OCOPC showed a slight rise in the global mortality rates, more markedly until the 1990s, when they began a downward trend that has levelled off to the present day. While OCC mortality rates decreased in males, OCOPC mortality increased in females, and OPC in both genders. To interpret and compare our findings with other studies, it is necessary to consider that the classification of OCC and OPC is not the same worldwide. They are two different sites which can lead to misclassification. While OCC arises from the structures comprising the lips, the anterior two-third of mobile tongue, the buccal mucosa, the gingiva, the hard palate, the retromolar trigone, and the floor of the mouth (Montero & Patel, 2015); OPC develops from the posterior area to the mouth including the base of tongue (posterior one‑third), the soft palate, the tonsillar complex (tonsil, tonsillar fossa and pillars), and the posterior and lateral pharyngeal walls (Cohan et al., 2009). However, other studies have considered only a few locations in the oral cavity or the oropharynx; others have included the soft palate as part of the mouth, the entire pharynx, or the salivary glands; and others have excluded the lip, and the hypo, naso and oropharynx (Seoane- Mato et al., 2014; van Dijk, Brands, Geurts, Merkx & Roodenburg, 2016; Zheng, Kirita, Kurumatani, Sugimura & Yonemasu, 1999). This controversy can be explained by the close topographical continuity between the oral cavity and the oropharynx mucosal lining tissue in a highly focused area. Therefore, an agreement is needed to clarify and standardize the terms of OCC and OPC to identify the rate of each neoplasm in epidemiological reports, given the growing evidence of changing trends in cancer registries and risk factors between both cancers (Conway, Purkayastha & Chestnutt, 2018). Incidence and mortality of OCOPC have experienced a paradigm shift in recent years. Currently, two groups of malignancies are generally accepted, those influenced by tobacco and alcohol exposure, and those associated with HPV infection (Chaturvedi et al., 2011; Gillison et al., 2015; Lai et al., 2017). In Spain, legislative and cultural changes in the late 1980s favoured a decrease in the population's exposure to smoking and alcohol consumption which resulted in a decline in the incidence and mortality rates of those cancers associated with these etiological factors, in 8  particular OCC. This change has been more pronounced in males since the 1990s, as mortality has not declined in females throughout the period analysed, probably linked to their incorporation into the tobacco habit that occurred in Spain during the last decades of the 20th century (Conway et al., 2018; Nieto & Ramos, 2002). In addition, the prevalence of alcohol drinking in males has decreased considerably since then, leading to a more favourable incidence and mortality trend (Bosetti et al., 2020). With regard to oral HPV infection, a prevalence of 4.5% has been reported for the general adult population worldwide, higher in developing countries (7.3%) than in developed ones (3.6%) (Kreimer et al., 2010). In Spain, the prevalence of genital HPV infection is estimated to be among the lowest in Europe (de Sanjose et al., 2003; Rodrigo et al., 2014). However, changes in sexual behaviour in our country are inducing a greater expansion of HPV infection in the population (Seoane-Mato et al., 2014), rising from 1.3% in 1990 to 6.1% in 2009 (Rodrigo et al., 2014). The first coordinated vaccination campaigns against HPV began in 2007 in 14-year-old girls and are currently being expanded to men who report having sex with men. It seems reasonable to consider that the mortality rate increase observed in OPC from 2009 to 2018 could coincide with a rise in the HPV incidence. Nevertheless, the potential benefit effect of HPV vaccination on OCOPC incidence and mortality are still unknown in Spain, since barely a decade has elapsed from the start of institutional vaccination campaigns to detect its influence as a protective factor. Recently, a study in the US has reported a significant lower prevalence of OPC in patients vaccinated against HPV (Katz, 2020). Our findings support the theory of the end of OCC epidemic behaviour described some decades ago in some European countries due to the reduction of tobacco and alcohol consumption in males (Bonifazi et al., 2011). In Spain, OCC mortality rates increased worryingly from the 1970s to the early 1990s substantially in males (Nieto & Ramos, 2002), when they tended to level off. Seoane- Mato et al. (Seoane-Mato et al., 2014) examined oral cavity, pharyngeal, oesophageal and gastric cancer mortality rates up to 2006, finding a decrease in mortality since the 1990s. Our data, updated to 2018, confirm these trends. This same pattern has been perceived in the first decade of the 21st century in Southern and most Central countries of Europe (Bonifazi et al., 2011; Garavello et al., 2010). Moreover, mortality in some Northern and Eastern Europe countries has shown an increase between 1990 and 2004 and even, in one country, this trend has been reported until 2012 (van Dijk et al., 2016). The US, China, Japan and Australia also showed a decline in mortality (Bosetti et al., 2020). 9  In Spain, early cancer diagnosis campaigns, lifestyle changes, and high fruit and vegetable consumption may also have influenced the favourable patterns from the 1990s (Nieto & Ramos, 2002; Seoane-Mato et al., 2014). Nonetheless, it should be noted that only moderate improvements in survival have been reported in the past three decades despite progress in the management of these neoplasms (Infante-Cossio, Torres-Carranza, Cayuela, Gutierrez-Perez & Gili-Miner, 2009). Consequently, it is unlikely that advances in therapeutic modalities and care for patient quality of life have played a key role in mortality rate trends. On the other hand, the proportion of advanced stages at diagnosis has remained stable in recent decades (McGurk.,Chan, Jones, O'regan & Sherriff, 2005; Warnakulasuriya, 2009), so that changes in incidence exhibit a fairly faithful reflection of mortality. In addition, the Spanish public health system offers universal and free assistance and has not undergone substantial changes in recent decades. Therefore, the influence of access to health care on mortality trends may be small compared to other countries. We have found significant differences in the mortality trends analysis by sub-sites. Lip cancer has shown a dramatic decrease since the beginning of the period, more marked in males. Lip cancer mortality trends has recently been analysed in 185 countries, finding a reduction in most of them (Miranda-Filho & Bray, 2020), probably due to sun exposure prevention campaigns (Greinert et al., 2015), since ultraviolet radiation is an important risk factor for lip cancer. We have also found reductions in mortality from cancer of the tongue, floor of the mouth and other sites of the mouth. Gingiva cancer has shown a variable pattern, although the low number of annual deaths produces large variations in its rates. Joinpoint analysis showed similarities between OCC and OPC, with a coincidence of change points. The similarity of cohort effects suggested exposure to shared risk factors. Analysing the effects of age, period and cohort, an interval of several decades is observed between the first exposure to risk factors and death. In males, OCC mortality rates have decreased in those generations born after the 1950s, indicating that mortality is likely to continue descending in the near future, as the same generations advance in age. In contrast, OCC mortality rates in females have remained high throughout the period. For OPC, mortality rates grew in cohorts born between the 1940s and 1960s for both genders and, therefore high death rates are expected to continue to be recorded in coming years. Regarding the APC models, although the age cohort model is the most suitable two-factor model 10  for males and females, we believe that the period effect also plays an important role in males, with a mortality drop in the 1990s for all age groups simultaneously. The best fit for the male age cohort model with a lower AIC and a narrower deviation could be explained by the method used to solve the problem of "non-identifiability", since penalty function models tend to overestimate the effect cohort (González, Llorca & Moreno, 2002). This study poses some limitations. One limitation is that data collected are subject to omissions at the time of completing death certificates. However, there were no major changes to the classification and coding in subsequent revisions of the ICD. Hence, it is unlikely that mortality trends were influenced by changes in disease diagnosis and certification since these cancers are relatively easy to detect. It would have been desirable to carry out the analysis considering histopathological diagnosis, tobacco/alcohol consumption, HPV exposure, tumour staging, and treatment performed, but death certificates do not include such information in Spain. Another limitation is correlated to the intrinsic problem of APC analysis models, which may overvalue the cohort or period effect depending on the model. This study monitored the total Spanish population over a 40-year period, comprising generations born approximately between 1900 and 2000 and, consequently, it constitutes an important time series with a large data package that allowed the characterization of epidemiological trends. The standardization of rates has been carried out with the standard European population of 2013, so our data are sufficiently reliable and comparable to permit an inference of mortality trends in most of the population of Europe, at least in the Southern/Mediterranean countries. Mortality data are the only ones that allow assessing such a long series of population, and in the case of cancer, previous studies have shown the quality of these data (Seoane-Mato et al., 2014). CONCLUSIONS In conclusion, our study provided a comprehensive analysis of OCOPC mortality trends in the last 40 years in Spain. The similarities and differences shown by OCC and OPC mortality trends in this period could explain the variable influence on exposure to risk factors and the efficacy of available therapeutic tools. Some risk factors involved can be prevented through anti-smoking and alcohol consumption campaigns. As OCC death rates have decreased in males in the past three decades, there is still room for improvement in females and OPC mortality, which have become an increasing health problem. Vaccination campaigns for the prevention of HPV- associated malignancies from 2007 onwards could have a preventive effect that should be 17  TABLE LEGENDS Table 1. Age standardized mortality rates and AAPC for all ages. Table 2. Age standardized mortality rates and AAPC by sex. Table 3. Age standardized mortality rates and AAPC by age group. FIGURE LEGENDS Figure 1. Joinpoints and mortality trends by genders and location: a) OCOPC; b) OCC; c) OPC; d) Lip; e) Tongue; f) Gingiva; g) Floor of the mouth; h) Other sites of the mouth. Figure 2. APC models for males and females. Age estimates are expressed as age-specific mortality rates per 100,000. Cohort and period estimates are expressed as multiplicative effects in relation to age estimates: a) OCC; b) OPC. Figure 3. Mortality rates in males and females per year (deaths per 100,000). ADDITIONAL SUPPORTING INFORMATION LEGENDS Supplementary Table S1. OCOPC mortality in Spain between 1979-2018: deaths by genders and locations. Supplementary Table S2. Goodness-of-fit test for different APC models for OCC and OPC in Spain between 1979-2018. Supplementary Figure S1. Sex ratio (male/female) for all ages: a) OCC; b) OPC. 18  Table 1. Age standardized mortality rates and AAPC for all ages. 1979-1983 ASMR (95% CI) 2014-2018 ASMR (95% CI) Period of years: AAPC (95% CI) Join point OCOPC (all ages) 3.24 (2.97;3.49) 3.45 (3.30;3.64) 1979-1992: 3.3 (2.7;3.9)* 2 1992-2009: -1.5 (-1.9;-1.2)* 2009-2018: -0.2 (-0.9;0.5) OCC (all ages) 2.86 (2.62;3.11) 2.42 (2.28;2.59) 1979-1991: 2 (0.7;3.3)* 1 1991-2018: -1.4 (-1.6;-1.2)* OPC (all ages) 0.37 (0.29;0.49) 1.03 (0.92;1.16) 1979-1991: 10.8 (9.2;12.5)* 1 1991-2018: -0.1 (-0.4;0.2) Note. ASMR: age standardized mortality rate per 100,000; AAPC: annual average percentage change; 95% CI: 95% confidence interval. * p<0.05. 19  Table 2. Age standardized mortality rates and AAPC by sex. 1979-1983 ASMR (95% CI) 2014-2018 ASMR (95% CI) Period of years: AAPC (95% CI) Join point 1979-1983 ASMR (95% CI) 2014-2018 ASMR (95% CI) Period of years: AAPC (95% CI) Join point Men Women OCOPC (all ages) 6.18 (5.67;6.77) 5.47 (5.15;5.82) 1979-1991: 3.3 (2.7;4)* 3 0.99 (0.81;1.21) 1.70 (1.54;1.86) 1979-2018: 1.7 (1.5;1.9)* 0 1991-2003: -1.5 (-2;-0.9)* 2003-2008: -3.9 (-6.4;-1.3)* 2008-2018: -0.7 (-1.3;0)* OCC (all ages) 5.45 (4.95;6.01) 3.58 (3.30;3.87) 1979-1991: 1.9 (0.9;2.9)* 1 0.93 (0.75;1.17) 1.42 (1.28;1.57) 1979-1981: 16.9 (-5.9;45.3) 4 1991-2018: -2.4 (-2.6;-2.2)* 1981-1985: -9.9 (- 19.6;1) 1985-1988: 11.7 (-11.1;40.4) 1988-2008: 0.8 (0.2;1.4)* 2008-2018: 2.6 (1.6;3.7)* OPC (all ages) 0.74 (0.59;0.98) 1.89 (1.70;2.10) 1979-1991: 10.6 (9;12.2)* 3 0.06 (0;0.16) 0.27 (0.21; 0.36) 1979-2018: 3.3 (2.7;3.9)* 0 1991-2000: 1 (- 0.7;2.8) 2000-2009: -2.6 (-4.1;-1)* 2009-2018: 1.1 (- 0.2;2.4) Note. ASMR: age standardized mortality rate per 100,000; AAPC: annual average percentage change; 95% CI: 95% confidence interval. * p<0.05. 20  Table 3. Age standardized mortality rates and AAPC by age group. 1979-1983 ASMR (95% CI) 2014-2018 ASMR (95% CI) Period of years: AAPC (95% CI) Join point <35 years OCOPC 0.09 (0.02;0.17) 0.04 (0;0.11) 1979-1994: -10.1 (-14.1;-5.9)* 1 1994-2018: 0.7 (-1.9;3.3) OCC 0.08 (0.01;0.17) 0.04 (0;0.11) 1979-1982: 32 (-24;129.3) 2 1982-1992: -16.5 (-24.5;-7.6)* 1992-2018: 2.1 (0.1;4.3)* OPC 0.01 (0;0.07) 0.00 (0;0.02) ** - 35-64 years OCOPC 3.44 (3.12;3.80) 3.21 (2.94;3.48) 1979-1981: 18.3 (-0.9;41) 2 1981-1993: 4.1 (3.1;5)* 1993-2018: -2.6 (-2.9;-2.4)* OCC 2.87 (2.57;3.19) 2.02 (1.82;2.22) 1979-1992: 3.4 (2.3;4.4)* 1 1992-2018: -3.1 (-3.4;-2.7)* OPC 0.57 (0.43;0.73) 1.20 (1.06;1.39) 1979-1991: 11.6 (9.7; 13.5)* 2 1991-2000: 0.3 (-1.5; 2.1) 2000-2018: -2.3 (-2.8; -1.8)* >64 years OCOPC 9.28 (8.28;10.38) 10.95 (10.27;11.67) 1979-1992: 1.7 (1.1;2.4)* 2 1992-2008: -0.6 (-1;-0.2)* 2008-2018: 1.4 (0.7;2)* OCC 8.55 (7.60;9.64) 8.16 (7.57;8.77) 1979-1985: -1.7 (-4.3;0.9) 3 1985-1991: 2.6 (-0.1;5.5) 1991-2008: -1 (-1.4;-0.6)* 2008-2018: 0.7 (0.2;1.3)* OPC 0.72 (0.49;1.11) 2.80 (2.44;3.18) 1979-1987: 13.1 (7.1;19.3)* 1 1987-2018: 1.9 (1.5;2.3)* Note. ASMR: age standardized mortality rate per 100,000; AAPC: annual average percentage change; 95% CI: 95% confidence interval. * p<0.05. **Because in several years the age-adjusted rate is “0”, Joinpoint regression analysis cannot be performed.