Incidence trends of different cancers attributable to smoking and alcohol consumption in Liaoning Province, Northeastern China from 2008 to 2019.

Introduction
Cancer is an important chronic disease that imposes a heavy disease burden on the global public health system. In 2020, the number of new cancer cases in China was approximately 4,569,000, accounting for 23.7% of the world’s cases and far exceeding that of other countries or regions [1]. Cancer continues to impose a serious burden on the Chinese population.
Current research has proven that smoking and alcohol consumption have led to a significant cancer burden [2]. Unfortunately, China is the world’s largest consumer and victim of tobacco. In 2018, the smoking prevalence in people aged 15 and older in China was 26.6%, of whom the prevalence was 50.5% in men and 2.1% in women [3]. In addition, alcohol consumption is very common in China. From 2015 to 2017, the prevalence of alcohol consumption in people aged 18 and older in China was 43.7%, with a prevalence of 64.5% in men and 23.1% in women [4].
It is worth noting that there were 3,011,400 new cases of smoking-related cancer and 1,224,900 new cases of alcohol-related cancer in China in 2022 [5]. Therefore, it is necessary to assess the burden of cancer attributable to smoking and alcohol consumption.
In recent years, most studies have utilized the population attributable fraction (PAF) to measure the burden of cancer attributable to risk factors. This metric requires prevalence data from 10 years prior [6, 7] along with corresponding relative risk (RR). Furthermore, given that the RR values differ across regions [8], prioritizing RR values derived from the local population is crucial.
It is worth noting that until now, for smoking or alcohol consumption, many scholars have only used PAF to represent burden [9, 10]; however, one study has explicitly highlighted that interpreting PAFs without age-standardized incidence rates (ASIRs) may not provide a complete picture [11]. On one hand, the PAF can only reflect relative levels. On the other hand, these studies calculate the PAF using a denominator that includes all types of cancer. Although this allows for a better observation of the overall proportion of disease burden attributable to the risk factors, it can lead to an unstable denominator because cancers unrelated to the studied risk factors also affect the calculation.
To address this problem, we use ASIR to more clearly elucidate the disease burden attributable to risk factors. As far as we know, no studies of smoking and alcohol attribution have used both metrics together. Additionally, we have employed a new PAF algorithm that incorporates only cancers associated with risk factors in the denominator and is now used in the field of cancers attributable to infections [12].
The Northeast China region ranks among the highest in the country for both smoking and drinking rates [13–15]. Therefore, assessing the cancer burden attributable to the two risks is urgently required. In this study we calculated the PAF using a new algorithm and evaluated the long-term trends in ASIR for cancers attributed to smoking, alcohol consumption, and these two risk factors combined, among Northeast China residents for the period 2008 to 2019. The aim of the study was to assess cancer burden and provide more information for developing cancer prevention strategies.

Methods

Identification of cancers due to smoking or alcohol consumption
This study was based on research by the International Agency for Research on Cancer (IARC) and the World Cancer Research Fund (WCRF) on the selection of cancer sites associated with smoking and alcohol consumption. Their results are summarized in Table 1. Certain cancer sites classified by IARC as causally related to alcohol, such as female breast and laryngeal cancers, were excluded from this study. Female breast cancer was excluded based on Chinese population data showing no significant association with alcohol consumption [16], while laryngeal cancer was not included due to incomplete data reporting in the cancer registry used.

The collection of cancer incidence and population data
The cancer incidence data used in this study were collected from the Cancer Registry Reporting System of Liaoning Province. To ensure the stability and representativeness of the data, we selected the cancer incidence data from the cancer registries of five cities in Liaoning Province (Shenyang, Dalian, Anshan, Benxi, and Dandong) to represent the cancer incidence levels of urban residents in Liaoning Province. These five cities had been conducting cancer reporting work prior to our study, and were established as national cancer monitoring sites. At present, its cancer registry data have been included in the China Cancer Registry Annual Report and the World Cancer Report and the data quality is considered reliable [1, 17]. Based on attributable characteristics, this study mainly included cancer incidence data for populations aged 30 and older. Population data were sourced from local public security departments.

Estimation of smoking and alcohol consumption rates from 1998 to 2009
We selected a latency period of 10 years for exposures and cancer occurrence as a compromise; therefore, to calculate the attributable burden for 2008 to 2019, it was necessary to calculate the exposure levels of risk factors from 1998 to 2009.
We first obtained data on the prevalence of smoking and drinking according to gender and age in Liaoning Province in 1999 and 2009 from the Chronic Disease and Risk Factor Surveillance for Liaoning Province for 1999 and 2009. Both surveillance studies were carried out province-wide by the Liaoning Provincial Center for Disease Control and Prevention. In these surveys, a multistage cluster sampling approach was used: the 14 cities and 44 rural counties across the province were stratified into three tiers based on their economic development levels. One surveillance site was randomly selected from each tier, and three neighborhoods were randomly chosen within each site. Finally, two residential communities were randomly selected within each neighborhood for the survey. The survey framework used in this study was consistent with the national survey framework. All interviewers received standardized training at the national level. During the survey, on-site quality control personnel carried out quality control throughout the entire questionnaire process in accordance with standardized quality control procedures and protocols. Then, by assuming a linear change in smoking rates between 1998 and 2009, we were able to estimate the smoking rates for different genders and age groups from 1998 to 2009 [18]. The smoking rate and alcohol consumption rate from 1998 to 2009 can be calculated using the following formulas: 
where x represents the year of estimation, y represents the rate for the year of estimation, x0 and x1 represent a certain year with known rates, and y0 and y1 represent the rates for the years x0 and x1, respectively.

Selection of RR of each cancer site attributed to the risk factors
RR data were obtained from studies identified from different sources, including PubMed, other websites, the China National Knowledge Infrastructure (CKNI) database, and other databases. The search words used were “meta-analysis,” “case-control study,” “cohort study,” and the names of specific risk factors and cancers. The studies were selected to contain information on RRs or odds ratios and corresponding 95% confidence intervals and to have been published in the last 20 years. Studies had to be in English or Chinese, and the highest priority was given to meta-analyses and large-scale surveys of representative samples in China, followed by nonrepresentative samples of Chinese populations, meta-analysis from other Asian countries and, finally, meta-analyses from non-Asian countries. The results are shown in Table 2. This selection approach was intended to improve the contextual relevance and applicability of the RR estimates to the Chinese population.

Calculation of the attributable burden
Rather than use average rates to estimate the cancer burden attributable to risk factors, we preferred to calculate the cancer burden in detail, based on gender and age. It was necessary, therefore, to first determine the cancer incidence PAF for each risk factor by gender, by age group, and by type of cancer. PAF estimates the proportion of disease cases in a population caused by a specific exposure. By accounting for other influencing factors through statistical adjustment, PAF reflects the independent effect of the exposure and helps assess its impact on public health. PAF was calculated based on the RR and frequency of population exposure to the risk factor, using the following formula:
where g represents different genders, i represents different age groups, r represents different types of risk factor, c represents different types of cancer, P represents the exposure rate to risk factors, and RR represents the relative risk of cancer occurrence due to the risk factors.
Because some cancers are caused by two risk factors, in our study, we considered the degree of overlap of PAF, assuming independence between exposure and risk. Combined PAF for exposure to multiple risk factors could be calculated using the following formula:
where g represents different genders, i represents different age groups, r represents different types of risk factor, and c represents different types of cancer.
The number of cancer cases attributable to specific risk factors was calculated using the following formula:
where AI represents the number of attributable cancer cases for the respective group, and I represents the incidence of cancer cases for the respective group.
Subsequently, the number of cancer cases in each group attributable to risk factors were summed by gender, age group, or cancer type. This total was then divided by the incidence of the corresponding cancers associated with those risk factors, resulting in the aggregated PAF by gender, age group, and cancer type. The PAFm for g, gr, gc, grc, gi, and gir were calculated using the following formula:

where g represents different genders, i represents different age groups, r represents different types of risk factors, and c represents different types of cancer.
The incidence rates attributable to risk factors were calculated using a similar method, with the formula as follows:

To estimate the ASIR, standardization was performed using data from the 2010 China Population Census. The methodological flowchart is shown in Fig. 1.

Statistical analysis
We used R (version 4.2.2) for statistical analysis. Trends in changes were represented using the annual percentage change (APC), calculated as APC = 100 × (eβ − 1). We denoted the rate as γ, with γ as the dependent variable and the year as the independent variable, fitting a log-linear model: ln(γ) = α + βx + ε, where α is the constant term, β is the regression coefficient, and ε is the random error term. The APC was tested using a t-test with a significance level of α = 0.05. The 95% CIs for the PAFm and ASIR were estimated using a bootstrap simulation method with 1000 simulations.

Results

Estimated smoking and alcohol consumption rates during the period 1998 to 2009
We selected a latency time of 10 years for exposures and cancer occurrence, so we calculated the smoking and alcohol consumption rates in Liaoning urban population from 1998 to 2009. The smoking prevalence results showed an increase among the male population in all age groups except the 20–29 years age group. Smoking prevalence was much higher among men than women. In the female population, smoking prevalence increased in the 20–49 age group and decreased in other age groups (Table S2). Estimates of alcohol consumption prevalence showed a decline in the male population in all age groups except the 50–59 age group. Alcohol consumption rates were much higher among men than women. In the female population, alcohol consumption prevalence declined in all age groups (Table S4).

Number of cancer cases in Liaoning Province from 2008 to 2019
The sex distribution of cancers related to smoking and alcohol consumption is shown in Table S5. Among males, in 2019 there were 14,464 cases of cancer associated with smoking or alcohol consumption, a substantial increase from 11,790 cases in 2008. Among females, in 2019 there were 8,424 cases of cancer associated with smoking or alcohol consumption, a substantial increase from 6,565 cases in 2008.

Number of cancer cases attributable to smoking, alcohol consumption, and their combination in Liaoning Province from 2008 to 2019
The sex distribution of cancers attributable to smoking, alcohol consumption, and their combination is shown in Table S6–S8. Among males, the number of cancer incidence cases attributable to smoking increased from 3,319 in 2008 to 4,292 in 2019. During the same period, the number attributable to alcohol consumption increased from 583 to 644, and the number attributable to their combination increased from 3,814 to 4,847. Among females, the number of cancer incidence cases attributable to smoking increased from 716 in 2008 to 730 in 2019. Over the same period, the number attributable to alcohol consumption decreased from 32 cases to 20 cases, and the number attributable to their combination remained relatively stable, changing slightly from 744 cases to 749 cases.

PAFm of cancers attributable to smoking, alcohol consumption, and their combination
The sex distribution of PAFm attributable to smoking, alcohol consumption, and their combination is shown in Figure S1-S2 and Fig. 2. Among males, the PAFm of cancers attributable to smoking rose from 33.21% (95% CI 21.79–43.46) in 2008 to 38.68% (95% CI 25.37–50.22) in 2019. The PAFm of cancers attributable to alcohol consumption declined from 12.86% (95% CI 5.58–20.48) in 2008 to 10.91% (95% CI 3.87–19.31) in 2019. The PAFm of cancers attributable to their combination exhibited an overall rise, increasing from 31.70% (95% CI 20.22–41.81) in 2008 to 32.81% (95% CI 20.73–43.37) in 2019. Among females, the PAFm of cancers attributable to smoking decreased from 11.13% (95% CI 5.18–19.22) in 2008 to 8.73% (95% CI 3.03–17.99) in 2019. The PAFm of cancers attributable to alcohol consumption in females declined from 1.28% (95% CI 0.37–2.65) to 0.68% (95% CI 0.10–2.02) over the same period, and the PAFm of cancers attributable to their combination exhibited an overall decline, decreasing from 11.34% (95% CI 5.22–19.66) in 2008 to 8.87% (95% CI 3.03–18.42) in 2019.
The age distribution of PAFm attributable to smoking, alcohol consumption, and their combination is shown in Fig. 2. Among males, the PAFm of cancers attributable to smoking exhibited overall increasing variation across all age groups except for 80+. In 2019, the highest PAFm for smoking among males was observed in the 60–69 age group (41.80%, 95% CI 28.31–53.03). The PAFm of cancers attributable to alcohol consumption declined across all age groups In 2019, the highest PAFm for alcohol among males was observed in the 50–59 age group (13.39%, 95% CI 5.19–22.77). The PAFm of cancers attributable to their combination in males exhibited an overall decline across most age groups, except for increases observed in the 60–69 and 70–79 age groups. Additionally, the PAFm of cancers attributable to smoking and alcohol in combination in males was highest in the 60–69 age groups with a 2019 value of 35.31% (95% CI 23.02–45.63). Among females, the PAFm of cancers attributable to smoking exhibited distinct trends across different age groups. In female age groups < 60 years, the PAFm showed increasing variation, whereas in age groups > 60 years, it displayed decreasing variation. However, in 2019, the PAFm in the older female age groups remained higher than in the younger groups, with the highest PAFm observed in the 80 + age group (13.29%, 95% CI 5.26–24.93).The PAFm of cancers attributable to alcohol consumption declined across all female age groups. In 2019, the highest PAFm was observed in the 50–59 age group (0.94%, 95% CI 0.16–2.69). Throughout the same period, PAFm for cancer attributable to alcohol consumption was higher in males than in females across all age groups, while remaining considerably lower than PAFm for cancers attributable to smoking. The variation in the PAFm of cancers attributable to their combination was consistent with the PAFm of cancers attributable to smoking: in 2019, the highest PAFm was observed in the 80 + age group (13.22%, 95% CI 5.20-24.94). Nonetheless, the PAFm across all age groups for alcohol/smoking combination in women was considerably lower than observed in males during the same period.
The cancer distribution of PAFm attributable to smoking, alcohol consumption, and their combination is shown in Figure S1-S2 and Fig. 2. Among males, the PAFm of cancers attributable to smoking increased across all cancers. In 2019, the three cancers with the highest PAFm were lung cancer (59.39%, 95% CI 43.24-72.00), oropharyngeal cancer (36.95%, 95% CI 22.50-49.88), and oral cancer (36.52%, 95% CI 22.18–49.43). In contrast, the PAFm of cancers attributable to alcohol consumption in men declined across all cancers: in 2019, the three cancers with the highest PAFm were oropharyngeal cancer (34.20%, 95% CI 23.18–45.47), oral cancer (33.81%, 95% CI 22.85–45.03), and esophageal cancer (24.06%, 95% CI 12.03–37.15). The PAFm of cancers attributable to their combination increased across all cancers except for colorectal cancer: in 2019, the three cancers with the highest PAFm were lung cancer (59.39%, 95% CI 43.24-72.00), oropharyngeal cancer (58.51%, 95% CI 40.47–72.67), and oral cancer (57.98%, 95% CI 39.96–72.20). Among females, the PAFm of cancers attributable to smoking decreased for all cancers except for oral cancer: in 2019, the highest three PAFm were observed for lung cancer (15.00%, 95% CI 6.11–27.87), oral cancer (7.76%, 95% CI 0.64–22.29), and nasopharyngeal cancer (7.11%, 95% CI 0.57–20.8). The PAFm of cancers attributable to alcohol consumption in females declined across all cancers: in 2019, the highest three PAFm were observed for oropharyngeal cancer (3.66%, 95% CI 1.29–7.88), oral cancer (2.86%, 95% CI 0.83–6.61), and esophageal cancer (1.80%, 95% CI 0.41–4.84). In contrast, the PAFm of cancers attributable to their combination in women decreased for all cancers: in 2019, the highest three PAFm were observed for lung cancer (15.00%, 95% CI 6.11–27.87), oral cancer (10.39%, 95% CI 1.46–27.43), and oropharyngeal cancer (9.55%, 95% CI 1.73–25.24).

ASIR of cancers attributable to smoking, alcohol consumption, and their combination
The sex distribution of ASIR attributable to smoking, alcohol consumption, and their combination is shown in Figure S3-S4 and Fig. 3. Among males, the ASIR of cancer attributable to smoking showed no significant change: the rate in 2019 (99.14 per 100,000, 95% CI 64.87-128.86) was almost equivalent to that in 2008 (100.67 per 100,000, 95% CI 66.02-131.78). The ASIR of cancer attributable to alcohol consumption in males showed an overall decreasing trend, from 17.70 per 100,000 (95% CI 7.67–28.23) in 2008 to 14.95 per 100,000 (95% CI 5.32–26.48) in 2019, although this change was not significant (APC = − 0.66%, P = 0.15). Although the ASIR of cancer attributable to their combination in males exhibited fluctuations, the rate in 2019 (110.45 per 100,000, 95% CI 69.33-147.05) were nearly equivalent to that in 2008 (114.06 per 100,000, 95% CI 72.44-151.09). Among females, the ASIR of cancer attributable to smoking declined from 19.01 per 100,000 (95% CI 8.80-33.01) in 2008 to 14.06 per 100,000 (95% CI 4.74–29.56) in 2019 (APC = − 1.73%, P < 0.01). The ASIR of cancer attributable to alcohol consumption in females declined from 0.87 per 100,000 (95% CI 0.25–1.81) in 2008 to 0.42 per 100,000 (95% CI 0.06–1.24) in 2019 (APC = − 5.42%, P < 0.001). During this period, the ASIR in males remained substantially higher than that in females. The ASIR of cancer attributable to their combination in females saw a decline from 19.82 per 100,000 (95% CI 9.05–34.60) in 2008 to 14.46 per 100,000 (95% CI 4.80–30.70) in 2019 (APC = − 1.85%, P < 0.01).
The age distribution of ASIRs attributable to smoking, alcohol consumption, and their combination is shown in Figure S3-S4 and Fig. 3. Among males, the ASIR of cancer attributable to smoking showed a decreasing trend in the 40–49 age group (P < 0.001) and an increasing trend in the 60–69 age group (P < 0.001). In 2019, the male age group with the highest ASIR was the 80 + group, with rates of 343.15 per 100,000 (95% CI 197.72–485.70). For the ASIR of cancer attributable to alcohol consumption in males, a decreasing trend was observed in the 30–39, 40–49, and 70–79 age groups (P < 0.001), while it exhibited an increasing trend in the 60–69 age group (P < 0.001). In 2019, the male age group with the highest ASIR was the 60–69 group, with a rate of 41.50 per 100,000 (95% CI 14.88-72.00), The trend in the ASIR of cancer attributable to their combination in males showed a pattern consistent with the trend observed for cancers attributable to smoking. In 2019, the male age group with the highest ASIR was the 80 + group, with rates of 372.15 per 100,000 (95% CI 207.11-536.94). Among females, the ASIR of cancer attributable to smoking in age groups < 60 years exhibited an upward trend (P < 0.01), while the 60–69 and 70–79 age groups showed a decreasing trend (P < 0.01). In 2019, the female age group with the highest ASIR was the 80 + group, at 103.74 per 100,000 (95% CI 41.09-194.59). The ASIR of cancer attributable to alcohol consumption in females showed a decreasing trend in the 40–49, 50–59, 70–79, and 80 + age groups (P < 0.01). In 2019, the highest ASIR in females was in the 70–79 age group, at 1.22 per 100,000 (95% CI 0.13–4.03). The trend in the ASIR of cancer attributable to their combination in females showed a pattern consistent with the trend observed for cancers attributable to smoking. In 2019, the female age group with the highest ASIR was the 80 + group, with rates of 104.75 per 100,000 (95% CI 41.16-197.84).
The cancer distribution of ASIR attributable to smoking, alcohol consumption, and their combination is shown in Figure S3-S4 and Fig. 3. Among males, the ASIR of cancers attributable to smoking showed an increasing trend for oral cancer, oropharyngeal cancer, renal cancer, and bladder cancer, while it showed a decreasing trend for nasopharyngeal cancer and esophageal cancer (P < 0.05). In 2019, the three cancers in males with the highest ASIR attributable to smoking were lung cancer (68.47 per 100,000, 95% CI 49.83-83.00), gastric cancer (8.31 per 100,000, 95% CI 4.72–11.93), and bladder cancer (7.51 per 100,000, 95% CI 2.52–11.95). The ASIR of cancers in males attributable to alcohol consumption showed an increasing trend for oropharyngeal and colorectal cancer, while it exhibited a decreasing trend for nasopharyngeal cancer, esophageal cancer, and liver cancer (P < 0.05). The three cancers in males with the highest ASIR attributable to alcohol consumption were colorectal cancer (5.80 per 100,000, 95% CI 0.78–12.22), liver cancer (3.92 per 100,000, 95% CI 1.70–6.45), and esophageal cancer (3.42 per 100,000, 95% CI 1.71–5.29). The ASIR of cancers in males attributable to their combination showed an increasing trend for oral cancer, oropharyngeal cancer, colorectal cancer, and bladder cancer (P < 0.05), and a decreasing trend for nasopharyngeal cancer, esophageal cancer, and liver cancer (P < 0.05). In 2019, the three cancers in males with the highest ASIR attributable to their combination were lung cancer (68.47 per 100,000, 95% CI 49.83-83.00), liver cancer (10.06 per 100,000, 95% CI 5.27–14.84), and gastric cancer (8.31 per 100,000, 95% CI 4.72–11.93). Among females, the ASIR of nasopharyngeal cancer, gastric cancer, colorectal cancer, liver cancer, pancreatic cancer, lung cancer, and renal cancer attributable to smoking all exhibited a decreasing trend (P < 0.05), while the ASIR of other cancers attributable to smoking showed no significant changes. In 2019, the three cancers with the highest ASIR attributable to smoking in women were lung cancer (11.15 per 100,000, 95% CI 4.38–21.28), colorectal cancer (1.31 per 100,000, 95% CI 0.18–3.46), and liver cancer (0.68 per 100,000, 95% CI 0.03–1.96). The ASIR of oral cancer, nasopharyngeal cancer, esophageal cancer, colorectal cancer, and liver cancer attributable to alcohol consumption all showed a decreasing trend (P < 0.05), while the ASIR of other cancers attributable to alcohol consumption showed no significant changes. The highest ASIRs attributable to alcohol consumption in females were for colorectal cancer (0.24 per 100,000, 95% CI 0.02–0.80), liver cancer (0.09 per 100,000, 95% CI 0.02–0.22), and oral cancer (0.05 per 100,000, 95% CI 0.01–0.11). For cancers attributable to the combination of alcohol consumption and smoking in females, the ASIRs of nasopharyngeal cancer, esophageal cancer, colorectal can cer, liver cancer, pancreatic cancer, lung cancer, renal cancer, and gastric cancer all exhibited a decreasing trend (P < 0.05). In 2019, the highest ASIRs for combination in females were for lung cancer (11.15 per 100,000, 95% CI 4.38–21.28), colorectal cancer (1.54 per 100,000, 95% CI 0.19–4.20), and liver cancer (0.76 per 100,000, 95% CI 0.05–2.15).

Discussion
This study analyzed the trends in cancer burden attributable to smoking and alcohol consumption in Liaoning Province from 2008 to 2019 by calculating the PAFm and ASIR. In 2019, lung cancer had the highest PAFm and ASIR among smoking-related cancers in both sexes. For alcohol-related cancers, oropharyngeal cancer had the highest PAFm, while colorectal cancer had the highest ASIR in both sexes.
Our findings indicate that the changes in PAFm for smoking and alcohol consumption are consistent with changes in smoking and alcohol consumption rates, respectively. Due to higher smoking and alcohol consumption rates among males, the PAFm in males was higher than in females. This observation is consistent with the findings of Whiteman et al. [28]. Furthermore, peak ages of PAFm for smoking or alcohol consumption were consistent in our results with previous research in males, but occurred later in females compared with earlier studies [29, 30]. This could possibly be due to differences in the age at which peak smoking and alcohol consumption rates occur across different regions [31, 32]. The combined effect of the two risk factors in both sexes is primarily driven by smoking, as its changes and peak ages are consistent with those of smoking. These results suggest that smoking causes a severe disease burden, which is consistent with research from South Korea [33].
The differences in PAFm among various cancers primarily stem from the inconsistent RRs posed by risk factors leading to cancer incidence. For both males and females, lung cancer, oral cancer, and oropharyngeal cancer demonstrate a relatively high PAFm attributable to smoking, consistent with research from China [34]. However, our results showed a slight difference to results of a study in the United States (US) in which the PAFm for esophageal cancer ranked higher [35]. This may be due to differences in the RR used in the studies. In the US studies, the RR for smoking leading to esophageal cancer was approximately two times higher than the RR used in our study.
For the PAFm of alcohol-attributable cancers, oral cancer, oropharyngeal cancer, and esophageal cancer were relatively high in both males and females in our study. This was similar to the results of a study conducted in China [34], but slightly different to males in a US study [35], in which the PAFm for liver cancer ranked higher. This difference may stem from variations in the methods of calculating the PAFm.
The PAFm of cancers attributable to the combined effect of the two risk factors were influenced by a combination of smoking rates, drinking rates, and the selected RR values. Our study showed that oral and liver cancer were more affected by smoking [35], while colorectal cancer and esophageal cancer were more affected by alcohol consumption. The conclusions for Liver cancer, colorectal cancer, and esophageal cancer were consistent with the findings of a Chinese study conducted in 2013 [34], but the conclusions for oral cancer differed from that study. Our conclusions for oral cancer, Liver cancer, and colorectal cancer were consistent with a 2014 US study [35], whereas the conclusions for esophageal cancer contradicted the findings of that study.
In addition to PAFm, the ASIR is an indicator of the absolute level of attributable burden. Therefore, we calculated the ASIR of cancers related to smoking, drinking, and their combination. ASIRs attributable to smoking or alcohol consumption are decreasing in females; however, the ASIR attributable to smoking and alcohol consumption in males has not shown any significant change in recent years, and remains substantially higher than that of females. This highlights the urgent need for greater attention to smoking and alcohol consumption among males. Among males, particular attention should be given to the 60–69 age group, in which the ASIR of cancer attributable to the aforementioned factors all showed an increasing trend. This may be explained by males in this age group or their spouses approaching retirement or being recently retired, as studies show that spousal retirement increases alcohol consumption [36]. Although an individual’s retirement has no significant effect on their own smoking, the smoking of current smokers increases when their spouse retires [36]. As most smokers are male, the rise in ASIR among men aged 60–69 is understandable. Among females, the ASIR of cancer attributable to the aforementioned factors all demonstrated a decreasing trend; however, the ASIR of cancer attributable to smoking increased in younger age groups, and this increasing trend in younger female smokers needs more attention.
According to our knowledge, this is the first attempt to study ASIRs of cancers attributable to smoking and alcohol consumption. Our findings indicate that among males, the ASIRs of cancers attributable to smoking has decreased only for nasopharyngeal, esophageal, and liver cancers, while for other cancers, the ASIRs attributable to smoking have not shown significant changes. Similarly, for cancers attributable to alcohol consumption, the ASIR among males has decreased only for nasopharyngeal and esophageal cancers, while remaining stable or increasing for others. In contrast, the ASIRs of most cancers attributable to smoking and alcohol consumption among females has declined. These findings highlight the urgent need to further strengthen tobacco and alcohol control efforts, particularly among males. In addition, the ASIRs of cancers attributable to the combined effect of the two risk factors are strongly influenced by smoking in both sexes, indicating that smoking is a more serious risk factor than alcohol consumption for the overall population.
The formulation of health policies necessitates a comprehensive consideration of both the PAFm and ASIR. The PAFm, serving as an RR assessment metric, quantifies the contribution of risk factors to diseases from an etiological standpoint. In contrast, the attributable ASIR, acting as an absolute disease burden indicator, reflects the actual magnitude of the impact of risk factors on population health. From a theoretical perspective, focusing on preventing cancers with a high PAFm can be very effective. However, some cancers may have a high PAFm but a low attributable ASIR, such as oropharyngeal cancer. In 2019, the PAFm for oropharyngeal cancer attributable to smoking in males was as high as 36.94%, but its attributable ASIR was only 0.67 per 100,000. Even if we completely eliminate the risk factor, the absolute reduction in the total number of cases in the population would still be Limited. Conversely, for cancers such as gastric cancer, the situation is different. In 2019, the PAFm attributable to smoking in males was relatively low at 19.98%, yet the ASIR remained relatively high at 8.31 per 100,000. Even if the PAFm is less than 20%, due to its relatively high ASIR, intervening in the etiology can reduce more cases. This highlights the importance of considering both PAFm and ASIR in public health interventions, as even a modest reduction in risk factors can lead to significant absolute benefits in high-burden diseases.
Overall, this study pioneers the comprehensive assessment of the cancer burden attributable to smoking and alcohol consumption in Liaoning Province from 2008 to 2019 by integrating PAFm and ASIR. The results indicate that the cancer burden caused by smoking or alcohol consumption remains high. Notably, lung cancer attributable to smoking exhibits the “dual-high” characteristic of both a high PAFm and ASIR, which urgently requires attention. For alcohol-related cancers, oropharyngeal cancer had the highest PAFm, while colorectal cancer had the highest ASIR in both sexes. In addition, cancers such as gastric cancer and renal cancer, which have a low PAFm but high ASIR, also require active etiological interventions. Given these findings, it is evident that the formulation of future health policies should consider these indicators comprehensively and develop targeted interventions to achieve rational allocation of health resources and better protect the health of the population.
There are several limitations to this study. First, we used RR estimates from Chinese or East Asian populations to improve contextual relevance. While this enhances applicability to our target population, it may have led to lower RR values for some cancers—such as lung cancer (RR = 3.56)—compared to global meta-analyses, potentially resulting in underestimation. Second, some cancer sites causally linked to alcohol consumption by IARC, including female breast and laryngeal cancers, were excluded. Breast cancer was excluded based on Chinese population data showing no significant association, and laryngeal cancer was omitted due to incomplete reporting in the cancer registry used in this study.

Supplementary Information