Trends in socioeconomic inequalities in preventable lung cancer mortality in three major Spanish cities of the Valencian Community, 2000–2015

Abstract

Lung cancer (LC) is a major global health issue and is considered one of the most preventable diseases. This study aims to analyse socioeconomic inequalities in preventable mortality from lung cancer in the provincial capital cities of the Valencian Community (Valencia, Alicante, and Castellón), stratified by sex, across two time periods: 2000–2007 and 2008–2015. This is an ecological study with trend analysis, where the units of analysis were census tracts. A deprivation index was calculated to classify these tracts based on their level of socioeconomic deprivation. Specific mortality rates by sex, age group (0–75 years), deprivation level (DL), and period were calculated, and Poisson regression models were fitted to estimate relative risks. Lung cancer mortality rates in men were largely stable across periods within deprivation levels and age groups. Among women aged 0–64, rates increased between periods, particularly in more deprived census tracts. Men consistently showed higher mortality risks in more deprived areas, with a positive socioeconomic gradient that strengthened over time, especially at ages 65–74. Comparing periods, mortality risk remained stable in men across most deprivation strata, whereas in women it increased significantly, particularly in more deprived areas and among those aged 65–74, with the inverse socioeconomic gradient observed in 2000–2007 attenuating by 2008–2015. These findings highlight the relevance of monitoring socioeconomic inequalities in preventable lung cancer mortality at the small-area level. Mortality risk was associated with sex, age, and level of socioeconomic deprivation, and tailored public health interventions targeting the most vulnerable groups — particularly women in more deprived areas — are warranted.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-50896-2.

Introduction

Lung cancer (LC) is the second most diagnosed type of cancer worldwide (11.4% of all cases) and the leading cause of cancer-related death (18.0% of all cancer deaths). In 2022, its incidence reached 2.5 million cases, resulting in 1.8 million deaths globally¹. This serious health issue is considered a highly preventable disease, as the majority of lung cancer cases are attributable to smoking².

The age-standardized incidence of LC in Southern Europe is estimated to be approximately three times higher in men than in women¹. However, it is known that in the European Union, LC has remained stable in men, while it has tended to increase in women. Mortality rates for LC are estimated to reach 16 to 18 per-100,000 women in the next decade if timely interventions are not implemented³. These trends are primarily driven by the distribution of tobacco consumption⁴, although some differences may also arise due to other factors. In Spain, recent analyses have shown the beginning of a decline in lung cancer mortality among men, while mortality in women has continued to rise, reflecting different stages of the tobacco epidemic⁵.

These patterns are closely linked to cohort- and sex-specific smoking histories. In Spain, smoking became widespread earlier and more intensely among men, whereas uptake among women occurred later and increased progressively across successive birth cohorts⁶. This temporal lag is important for interpreting current LC mortality patterns, since the effects of smoking on lung cancer often emerge several decades after exposure⁷. Although tobacco remains the main attributable exposure for lung cancer mortality, other risk factors such as environmental pollution, occupational exposures and residential radon may also contribute to the spatial distribution of the disease in specific settings^8,9.

As a disease closely linked to lifestyle and environmental factors, there are significant differences in its incidence across different geographical areas and across different social groups. In Andalusia, a clear association was observed between lung cancer incidence and economically deprived areas, closely correlated with higher tobacco consumption in regions of lower socioeconomic status^10,11. Social inequalities have been linked in various studies to a higher tendency to develop lung cancer and lower survival rates, primarily due to lower treatment acceptance, a higher number of comorbidities, and poorer prevention¹². In general, a correlation has been found between higher mortality from lung cancer and residing in deprived areas^12–14.

In Spain, within the framework of the MEDEAS-II project, a deprivation index was constructed to measure the socio-economic level of small geographical areas¹⁵which has been applied in the study of inequalities in mortality^16–18. The study of mortality in small areas through the deprivation index in three cities on the Mediterranean coast in Spain (Alicante, Valencia, and Castellón), showed significant inequalities according to socio-economic status. Moreover, these inequalities persisted before and after the 2008 economic crisis¹⁹. However, there is no information available on the evolution of LC-specific mortality, which could be essential to understand the distribution of the disease. Given that lung cancer mortality largely reflects past exposure to preventable risk factors such as tobacco use, its analysis may provide an informative indicator of avoidable health inequalities across city populations. The analysis of LC mortality can serve as a vital/key indicator for evaluating the effectiveness of prevention plans and actions targeting the disease. Furthermore, examining different environments as well as the socioeconomic and demographic characteristics of individuals with LC can help to identify the profiles of those at higher risk of developing the disease.

This study aims to examine socioeconomic inequalities in preventable lung cancer mortality across sex and over time in three cities along the Spanish Mediterranean coast—Alicante, Valencia, and Castellón. Specifically, this is a descriptive observational epidemiological study that examines socioeconomic disparities in preventable lung cancer mortality across sex (women and men) and two time periods (2000–2007 and 2008–2015). Additionally, it seeks to assess how these dynamics evolved following the onset of the 2008 economic crisis, through comparisons between two time periods (2000–2007 and 2008–2015).

Methods

Design, study population and analysis units

This ecological study analyzes mortality trends by comparing two distinct time periods: 2000–2007 and 2008–2015. The units of analysis were the census tracts (CTs) of three provincial capital cities in southeastern Spain, located along the Mediterranean coast in the Valencian Community: Alicante (178 CTs), Castellón (58 CTs), and Valencia (531 CTs).

During the first period (2000–2007), the average total population across these cities was 1,240,744 inhabitants, which increased to 1,310,123 inhabitants in the second period (2008–2015). Population data disaggregated by CT, year, age, and sex were obtained from the Valencian Institute of Statistics, the official body responsible for producing population statistics for the region. Detailed population distributions by sex, age group, and deprivation level are presented in Supplementary Table S1.

Mortality data

All deaths occurring among residents of Alicante, Castellón, and Valencia were included in the analysis. Mortality data were obtained from the Mortality Register of the Valencian Community and were disaggregated by year of death, age, sex, city of residence, and cause of death.

The study focused exclusively on deaths attributed to lung cancer, identified using the International Classification of Diseases, 10th Revision (ICD-10) codes C33 and C34. In accordance with the conceptual framework used in the MEDEA project¹³, and following the standard definition of preventable mortality proposed by Nolte and McKee²⁰, the analysis focused on deaths from lung cancer occurring before age 75. Therefore, the mortality database includes observations only for individuals aged 0–74. All deaths were georeferenced to their corresponding census tract of residence.

Socioeconomic privation level of the census tract

For each period and city, a deprivation index (DI) was calculated for each census tract (CT) using the following socioeconomic indicators: unemployment, manual work, temporary employment, insufficient education among young people (16 to 29 years old), and insufficient education in the general population. These indicators were derived from the Population and Housing Censuses conducted in 2001 (for the period 2000–2007) and 2011 (for the period 2008–2015).

The DI used was developed within the framework of the MEDEA-III project, in which both the socioeconomic and mortality data of this study are included¹⁵ For each period and city, the 10th (P10), 25th (P25), 75th (P75) and 90th (P90) percentiles for DI were calculated, classifying census tracts into five deprivation levels (DL) according to their value: DL1, DI values below P10 (lower deprivation); DL2, DI values between P10 and P25; DL3 DI values between P25 and P75; DL4 DI values between P75 and P90 and DL5, DI values above P90 (higher deprivation). This classification was designed to better quantify the risks associated with socioeconomic disparities, comparing the most advantaged areas (DL1) with the most deprived areas (DL5). Detailed information on the average values of the five socioeconomic indicators across the DLs, as well as the geographic distribution of deprivation levels within the cities, can be found in a previous publication¹⁹.

Data analysis

To examine the evolution of mortality risk over time, data were classified into two time periods: 2000–2007 (P1) and 2008–2015 (P2). Deaths were grouped into the following age intervals: 0–64 and 65–74. Additional analyses using finer age categories (0–44, 45–64 and 65–74) are presented in the supplementary material. Specific mortality rates were calculated by sex, age group, deprivation level (DL), and period.

To estimate the relative risks (RR) between the categories of the variables under study, Poisson regression models were applied, and analyses were conducted separately for men and women. Models included deprivation level (DL), period, and age group as covariates. In the joint analysis combining the three cities, city was included as an adjustment variable, with no evidence of effect modification by city. Robust estimation was used to account for any potential overdispersion in the data.

In addition to the main analyses, where deprivation level was modelled as a categorical variable (with DL1 as the reference category), supplementary Poisson regression models were fitted with deprivation entered as an ordinal variable to assess linear trends across deprivation levels. These models were stratified by sex and period and adjusted for age group. Interaction terms between deprivation level, period, and age group were tested to assess potential effect modification.

Mortality risks were modelled using Poisson regression, assuming independence between census tracts. This assumption is considered reasonable in this context, as comparisons are made across deprivation strata rather than between geographically adjacent areas. Census tracts within each deprivation group do not form spatially contiguous clusters, and areas classified in the same deprivation level may be located in different parts of the city. Therefore, the structure of the data corresponds to socioeconomic strata rather than spatial neighbourhood units, supporting the validity of the independence assumption in the Poisson framework.

The statistical software SPSS© (v.25) was used for the computations.

Results

Between 2000 and 2015, there were 6,661 preventable deaths from lung cancer across the three cities under study (1,577 in Alicante, 795 in Castellón, and 4,289 in Valencia). Of these, 81 (1.2%) could not be georeferenced and assigned to the corresponding census section due to missing residential addresses or addresses that did not correspond to the cities under study. Of the 6,580 available deaths, 3,162 occurred between 2000–2007, and 3,418 occurred between 2008–2015.

Overall, higher risks of lung cancer mortality were observed among men living in more deprived areas, whereas patterns in women varied across age groups and deprivation levels.

Table 1 shows that the number of lung cancer deaths among men was similar in 2000–07 (n = 2,739) and 2008–15 (n = 2,704), with comparable distributions across the 0–64 and 65–74 age groups. In contrast, lung cancer deaths among women increased from 423 to 714, mainly due to a rise in deaths in the 0–64 age group. The proportion of lung cancer within all-cause mortality remained substantially higher in men, although it also increased among women over time. Detailed distributions of lung cancer deaths by sex and period using finer age groups (0–44, 45–64, and 65–74 years) are provided in Supplementary Table S2.

Table 1.

Number and proportion of lung cancer deaths by sex, age group, and period (2000–2007 and 2008–2015) in Alicante, Castellón, and Valencia.

Period	Sex	Age	Lung cancer		% of total deaths in this group	Total deaths
			n %			n %
2000–07	Men	0–64	1479	54.0	13.4	11,036	24.8
		65–74	1260	46.0	3.78	33,376	75.2
		Total	2739	100.0	6.17	44,412	100.0
	Women	0–64	297	70.2	6.0	4910	11.7
		65–74	126	29.8	0.34	37,157	88.3
		Total	423	100.0	1.01	42,067	100.0
2008–15	Men	0–64	1398	51.7	14.5	9628	22.0
		65–74	1306	48.3	3.82	34,214	78.0
		Total	2704	100.0	6.17	43,842	100.0
	Women	0–64	491	68.8	10.2	4834	10.8
		65–74	223	31.2	0.56	39,794	89.2
		Total	714	100.0	1.60	44,628	100.0

Table 2 shows socioeconomic differences in lung cancer mortality rates by sex, deprivation level, and period. Among men, rates were higher in the most deprived areas, particularly at ages 65–74, where DL5 reached 421.3 per 100,000 in 2000–07 and 398.0 in 2008–15, compared with 306.3 and 235.7 in DL1, respectively. In men aged 0–64, rates were lower overall but remained higher in DL5 than in DL1 in both periods (42.1 vs 37.9 in 2000–07; 39.0 vs 32.3 in 2008–15). Within the same deprivation levels, lung cancer mortality rates in men showed no significant differences between periods across all deprivation levels and age groups, as indicated by overlapping confidence intervals, reflecting marked stability over time. Among women, mortality rates were substantially lower than in men but increased over time in both age groups. In women aged 0–64, rates increased across most deprivation levels, with more pronounced increases observed in the more deprived strata (DL3–DL5); for example, rates in DL5 rose from 5.5 to 12.7 per 100,000, and in DL3 from 6.4 to 10.1. In women aged 65–74, rates increased across all deprivation levels, notably in DL1 (from 36.6 to 69.1 per 100,000) and DL5 (from 27.3 to 55.9), though confidence intervals overlapped in most strata. An exception was DL3, where the increase in women aged 65–74 (from 18.1 to 37.0) was more clearly differentiated. These patterns were consistent when using finer age group stratification (Supplementary Table S3).

Table 2.

Age-specific lung cancer mortality rates (per 100,000; 95% CI) by sex, deprivation level, and period (2000–2007 and 2008–2015) in Alicante, Castellón, and Valencia.

			2000–2007			2008–2015
Sex	Age	(DL)a	n	Rate	95%CI	n	Rate	95%CI
Men	0–64	DL1	119	37.86	31.06–44.66	89	32.25	25.55–38.95
		DL2	203	29.95	25.83–34.07	180	22.77	19.44–26.10
		DL3	746	34.51	32.04–36.99	760	33.45	31.07–35.83
		DL4	257	42.44	37.25–47.63	217	36.89	31.98–41.80
		DL5	154	42.10	35.45–48.75	152	38.95	32.76–45.14
	65–74	DL1	95	306.26	244.68–367.85	78	235.67	183.37–287.97
		DL2	152	264.00	222.03–305.97	218	311.19	269.88–352.50
		DL3	624	317.08	292.20–341.96	658	310.53	286.81–334.26
		DL4	214	343.74	297.68–389.79	208	362.84	313.53–412.15
		DL5	175	421.32	358.90–483.75	144	398.03	333.02–463.04
Women	0–64	DL1	42	12.43	8.67–16.20	50	16.96	12.26–21.66
		DL2	58	8.31	6.17–10.44	96	11.75	9.40–14.10
		DL3	140	6.42	5.36–7.48	232	10.14	8.83–11.44
		DL4	38	6.47	4.41–8.53	68	11.95	9.11–14.79
		DL5	19	5.54	3.05–8.03	45	12.73	9.01–16.44
	65–74	DL1	15	36.55	18.05–55.05	30	69.06	44.34–93.77
		DL2	25	33.00	20.07–45.94	40	46.33	31.97–60.68
		DL3	46	18.11	12.87–23.34	98	36.99	29.66–44.31
		DL4	25	32.83	19.96–45.70	29	39.78	25.30–54.25
		DL5	15	27.30	13.49–41.12	26	55.87	34.39–77.34

(a) DL: Deprivation level of the residential census section based on the deprivation index (DI): DL1: DI < P10; DL2: P10 ≤ DI < P25; DL3: P25 ≤ DI < P75; DL4: P75 ≤ DI < P90; DL5: DI ≥ P90; Pq = Percentile q.

To assess whether the effects of DL were significantly different across cities, multivariate models with city, DL, period, and age group effects were adjusted, checking for the absence of statistical significance in the interaction terms between city and DL. The interactions were non-significant for both men (p = 0.14) and women (p = 0.27), leading to the estimation of effects jointly for all three cities under study.

In the joint analysis of the three cities, Poisson regression models suggested a second-level interaction between DL, period, and age group, both for men and women.

Figure 1 shows age-specific lung cancer mortality rates (per 100,000) by sex, deprivation level and period, with 95% confidence intervals. Among men, a clear deprivation gradient is observed in both age groups, particularly at ages 65–74, with persistently higher rates in DL4–DL5 and only modest declines over time. Among women, overall rates are lower but tend to increase in the second period, especially in DL5 and in the 65–74 age group, suggesting a widening gap between the least and most deprived areas.

Fig. 1.

Age-specific lung cancer mortality rates (per 100,000; 95% CI) by sex, deprivation level, and period (2000–2007 and 2008–2015) in Alicante, Castellón, and Valencia. (a) DL: Deprivation level of the residential census section based on the deprivation index (DI): DL1: DI < P10; DL2: P10 ≤ DI < P25; DL3: P25 ≤ DI < P75; DL4: P75 ≤ DI < P90; DL5: DI ≥ P90; Pq = Percentile q.

Due to the presence of interaction, relative risks between DL categories (a measure of inequality by DL) were estimated, specific to sex, age, and period, as well as relative risks between periods (a measure of the increase or decrease in mortality risks between periods), specific to sex, age, and DL.

Table 3 presents relative risks of lung cancer mortality by deprivation level, estimated within each sex, age group, and period separately, using DL1 as the reference category. Among men aged 0–64, in 2000–2007, only DL2 showed a significantly lower risk compared with DL1 (RR 0.79, 95% CI 0.63–0.99), while no significant differences were observed across other deprivation levels; the same pattern was observed in 2008–2015, where DL2 again showed a significantly lower risk (RR 0.71, 95% CI 0.55–0.91). Among men aged 65–74, in 2000–2007, only DL5 showed a significantly higher risk than DL1 (RR 1.38, 95% CI 1.08–1.77); in 2008–2015, the gradient became more pronounced, with significantly higher risks observed across DL2 through DL5 (RRs ranging from 1.32 to 1.69), reflecting a widening socioeconomic gradient in the second period. Among women aged 0–64, in 2000–2007, all deprivation levels DL2–DL5 showed significantly lower risks than DL1; in 2008–2015, this inverse gradient persisted but was attenuated, with significantly lower risks observed only in DL2 and DL3. Among women aged 65–74, in 2000–2007, only DL3 showed a significantly lower risk than DL1 (RR 0.50, 95% CI 0.28–0.92); in 2008–2015, DL3 and DL4 showed significantly lower risks (RRs 0.54 and 0.58, respectively). The observed socioeconomic gradients were consistent across finer age strata (Supplementary Table S4).

Table 3.

Relative risks (RR) of lung cancer mortality by deprivation level, stratified by sex, age group, and period (2000–2007 and 2008–2015).

			2000–2007				2008–2015
		Deprivation level (DL)a	n	RR	95% CI		n	RR	95% CI
Sex	Age				Lower	Upper			Lower	Upper
Men	0–64	DL5	119	1.112	0.875	1.413	89	1.208	0.930	1.569
		DL4	203	1.121	0.902	1.393	180	1.144	0.894	1.464
		DL3	746	0.912	0.751	1.106	760	1.037	0.833	1.292
		DL2	257	0.791	0.631	0.992	217	0.706	0.548	0.910
		DL1	154	1			152	1
	65–74	DL5	95	1.376	1.075	1.772	78	1.689	1.286	2.234
		DL4	152	1.122	0.884	1.435	218	1.540	1.193	2.008
		DL3	624	1.035	0.839	1.292	658	1.318	1.049	1.679
		DL2	214	0.862	0.669	1.117	208	1.320	1.025	1.720
		DL1	175	1			144	1
Women	0–64	DL5	42	0.445	0.259	0.766	50	0.750	0.502	1.122
		DL4	58	0.520	0.336	0.807	96	0.705	0.489	1.015
		DL3	140	0.516	0.366	0.729	232	0.598	0.440	0.811
		DL2	38	0.668	0.449	0.994	68	0.693	0.492	0.975
		DL1	19	1			45	1
	65–74	DL5	15	0.747	0.362	1.540	30	0.809	0.475	1.367
		DL4	25	0.898	0.479	1.743	40	0.576	0.345	0.962
		DL3	46	0.495	0.283	0.918	98	0.536	0.361	0.820
		DL2	25	0.903	0.482	1.752	29	0.671	0.419	1.085
		DL1	15	1			26	1

DL: Deprivation level of the residential census section based on the deprivation index (DI): DL1: DI < P10; DL2: P10 ≤ DI < P25; DL3: P25 ≤ DI < P75; DL4: P75 ≤ DI < P90; DL5: DI ≥ P90; Pq = Percentile q.

Models adjusted for city. Analyses were conducted separately by sex, age group, and period. DL1 was used as the reference category.

In supplementary analyses modelling deprivation as an ordinal variable, a significant positive linear trend in lung cancer mortality was observed among men in both periods. The relative risk associated with each one-level increase in deprivation was 1.07 (95% CI: 1.04–1.09) in 2000–2007 and 1.05 (95% CI: 1.03–1.08) in 2008–2015. Among women, the corresponding relative risk was 0.92 (95% CI: 0.86–0.99) in 2000–2007, indicating an inverse gradient, whereas no significant linear trend was observed in 2008–2015 (RR = 0.96; 95% CI: 0.91–1.01).

Table 4 presents period relative risks (2008–15 vs 2000–07) by age, sex, and deprivation level. Among men, no statistically significant differences between periods were observed across most age groups and deprivation strata, indicating overall stability in lung cancer mortality risk over time. A possible exception was DL2 in ages 0–64, where a modest reduction was suggested, though this finding appears isolated and should be interpreted cautiously. Among women, mortality risk increased significantly between periods across all deprivation levels, though the magnitude and significance of these increases varied by age group. In the 65–74 age group, risk increases were consistently significant across all deprivation levels. In the 0–64 age group, significant increases were observed predominantly in DL2–DL4. In DL5, period RRs were approximately 2.3 for ages 0–64 and 2.0 for ages 65–74, with the latter clearly reaching statistical significance. Similar results were observed in analyses using finer age groups (Supplementary Table S5).

Table 4.

Relative risks (RR) of lung cancer mortality in 2008–2015 (P2) compared with 2000–2007 (P1), stratified by sex, age group, and deprivation level in Alicante, Castellón, and Valencia.

		Men				Women
Deprivation level		n (P1/P2)	RR	95% CI		n (P1/P2)	RR	95% CI
(DL)	Age			Lower	Upper			Lower	Upper
DL1	0–64	119/89	0.852	0.647	1.121	42/50	1.364	0.905	2.056
	65–74	95/78	0.770	0.569	1.037	15/30	1.889	1.017	3.512
	0–64	203/180	0.760	0.622	0.929	58/96	1.415	1.021	1.960
DL2	65–74	152/218	1.179	0.959	1.452	25/40	1.404	0.852	2.313
	0–64	746/760	0.969	0.876	1.072	140/232	1.579	1.280	1.948
DL3	65–74	624/658	0.979	0.878	1.093	46/98	2.043	1.439	2.899
DL4	0–64	257/217	0.869	0.726	1.041	38/68	1.847	1.242	2.746
	65–74	214/208	1.056	0.872	1.278	25/29	1.211	0.710	2.068
DL5	0–64	154/152	0.925	0.739	1.158	19/45	2.298	1.344	3.928
	65–74	175/144	0.945	0.757	1.177	15/26	2.046	1.084	3.863

DL: Deprivation level of the residential census section based on the deprivation index (DI): DL1: DI < P10; DL2: P10 ≤ DI < P25; DL3: P25 ≤ DI < P75; DL4: P75 ≤ DI < P90; DL5: DI ≥ P90; Pq = Percentile q.

Models adjusted for city. Analyses were stratified by sex, age group, and deprivation level.

Discussion

In the three cities studied (Alicante, Castellón and Valencia), lung cancer mortality risk among men remained largely stable between periods across most deprivation levels and age groups, with no statistically significant differences observed in most strata. Among women, however, the patterns were more heterogeneous across age groups and deprivation levels. Mortality risk increased significantly in the second period, particularly among women aged 65–74 across all deprivation strata, and among those aged 0–64 in intermediate-to-high deprivation levels (DL2–DL4). It is important to note that the present study examines preventable lung cancer mortality (0–74 years), which reflects premature deaths that are largely attributable to modifiable exposures such as tobacco use, occupational hazards, and environmental risks. This focus helps interpret the sex-specific trends observed, which mirror historical differences in smoking initiation and intensity.

This aligns with national trends, which suggest that the observed changes could be attributed to a shift in tobacco habits, with a decrease in smoking among men and an increase among women⁵. Indeed, a reconstruction of smoking prevalence in Spain indicates that it would have risen from 42.4% in men and 0.9% in women in 1945, to 57.8% in men and 5.8% in women in 1970, and then to 48.9% in men and 22.5% in women in 1995⁶. Moreover, it is known that the prevalence of smokers in the Valencian Community is among the highest in Spain²¹. Given the well-established latency between tobacco exposure and lung cancer development, typically ranging from 20 to 50 years, current mortality patterns largely reflect smoking behaviours adopted decades earlier⁷. In this context, the observed trends are also consistent with cohort-driven smoking dynamics among Spanish women, whereby smoking uptake increased markedly among cohorts born from the 1950s onwards. These cohort effects help explain contemporary mortality patterns, as they mirror behavioural changes that occurred several decades before the onset of the disease.

On the other hand, according to the level of deprivation, men showed a consistent positive socioeconomic gradient in preventable lung cancer mortality, with significantly higher risks in more deprived areas, and this gradient became more pronounced in the second period, particularly at ages 65–74. In women, an inverse pattern was observed in 2000–2007, with higher risks in less deprived areas; however, this pattern attenuated by 2008–2015, with risk differences across deprivation levels becoming largely non-significant, and an emerging increase in more deprived strata. This inverse relationship between deprivation level and lung cancer mortality in men and women had already been observed in various regions of Spain^11,16,22,23, and in other countries such as France¹³, and Italy²⁴. However, this pattern appears to differ depending on the geographical areas and the period studied^25–27.

Evidence from studies of smoking initiation in Spain suggests that this pattern may reflect the historical diffusion of smoking among women across socioeconomic groups. Smoking initiation among women was initially concentrated among those with higher levels of education but later spread to women with lower educational attainment as smoking became more socially widespread²⁸. This social diffusion process has been described in several European countries and is consistent with the stages of the tobacco epidemic.

In this context, the higher lung cancer mortality observed in more affluent areas among women may reflect earlier smoking uptake among higher socioeconomic groups. However, this study has observed an increase in lung cancer mortality among women in more deprived areas in the second period, suggesting the possible emergence of a socioeconomic gradient similar to that historically observed among men. If risks and rates continue to trend toward stabilization across deprivation levels, as observed in men, this temporal pattern would support the continuation of broad tobacco control public health action across all social strata. Given the increase in smoking among women between the 1980s and 1990s^6,28, this hypothesis suggests that in the coming decades, a significant rise in lung cancer cases will be seen in women of lower socioeconomic status.

Another possible cause of the observed trends is the effect of the 2008 economic crisis on lung cancer mortality. Throughout this period, two contrasting mortality trends were noted. The first, referred to as the countercyclical trend, results from increased poverty, which subsequently leads to higher rates of illness and death within the population²⁹. On the other hand, the procyclical trend is seen during economic downturns when the encouragement of certain health-promoting behaviours enhances the living conditions of certain population groups³⁰. However, despite a potential general procyclical trend following a recession, mortality may still increase in particular social groups or regions. In certain countries, an increase in smoking prevalence was observed after the crisis, particularly among individuals of lower socioeconomic status^31,32, with the worsening of mental health being attributed to this pattern³³. Furthermore, it is crucial to highlight that disparities in mortality continued after the economic crisis, with no overall rise in death rates, except in specific groups when stratified by sex and age¹⁹. Specifically, in lung cancer, the stability observed in men is consistent with a procyclical pattern, while the significant increases observed in women — particularly in more deprived areas and older age groups — are consistent with a countercyclical effect, underscoring the persistent inequalities in lung cancer mortality after the economic crisis. In addition, these trends should be interpreted in the context of tobacco control policies implemented in Spain during the study period, particularly the comprehensive smoke-free legislation introduced in 2005 and strengthened in 2010. These policies substantially reduced exposure to second-hand smoke and contributed to changes in social norms around smoking^34,35. However, given the long latency between tobacco exposure and lung cancer, their impact on lung cancer mortality is likely to become apparent only in the longer term. Consequently, the mortality patterns observed in this study largely reflect smoking behaviours established decades earlier.

A key contribution of this study is its explicit focus on preventable lung cancer mortality as an analytical lens to understand social inequalities in a highly avoidable cause of death. By isolating premature deaths closely linked to modifiable exposures, the study advances a conceptual distinction often overlooked in the literature, where total mortality may obscure the extent to which inequalities reflect avoidable risks. Our findings demonstrate that socioeconomic gradients remain strong—and in some cases emerging—when examining only preventable mortality, underscoring the central role of structural determinants in shaping premature loss of life. By integrating a harmonized deprivation measure at small-area level, the study further clarifies how social context influences avoidable cancer mortality, offering a framework that can be applied beyond the specific cities analysed. Altogether, this work contributes conceptually and empirically to the understanding of preventable cancer mortality as a sensitive indicator of social inequity.

This study has several limitations that should be acknowledged. First, as an ecological analysis based on census tracts, the results cannot be interpreted at the individual level, and the deprivation categories used inevitably mask within-area heterogeneity; not all residents of deprived areas experience deprivation, and conversely, deprived individuals may live in less deprived tracts. Second, although the study benefits from a large overall population, some combinations of sex, age group, period and deprivation level involve small numbers of deaths, which may lead to instability in specific estimates and wide confidence intervals. This is particularly relevant in younger age groups and in the least or most deprived strata. For this reason, the analyses were conducted using broader age categories and two time periods in order to improve the stability of the estimates.

Third, the use and interpretation of the MEDEA deprivation index require some clarification. This index was originally constructed using data from large Spanish metropolitan areas — specifically Barcelona, Bilbao, Madrid, Seville, and Valencia — and its applicability to smaller provincial capitals such as Castellón may be limited, as the socioeconomic and spatial structure of these cities may not be fully captured by an index developed in larger city contexts. Furthermore, the MEDEA index is primarily constructed from census indicators related to employment conditions and educational attainment, which largely reflect socioeconomic position through labour-market participation. This may be particularly problematic when assessing women’s socioeconomic status, as area-based deprivation measures may underestimate women’s true social position when household or partner socioeconomic status is not considered³⁶. Consequently, lung cancer risks in women in the cities studied may have been partially overestimated in some strata. The index may also be less sensitive to socioeconomic circumstances among retired populations or to gender-specific socioeconomic dynamics and may not fully capture other relevant dimensions such as household income or accumulated wealth. Future research could complement area-based deprivation measures with individual or household-level socioeconomic information from survey-based sources, such as the Spanish National Living Conditions Survey. Additionally, the MEDEA stratification showed some inconsistencies in intermediate deprivation levels, with a lack of alignment across neighbouring social strata in certain age groups and sexes, which should be interpreted with caution.

Fourth, mortality risks were modelled assuming independence between census tracts and without explicitly including spatially structured random effects; therefore, some residual spatial correlation between neighbouring areas cannot be completely ruled out, which might lead to a slight underestimation of uncertainty but is unlikely to change the main conclusions. Finally, a small proportion of deaths (1.2%) could not be georeferenced, although this percentage is considerably lower than in similar studies and is unlikely to materially affect the results. Despite these limitations, the study’s use of harmonized small-area indicators across three urban settings provides robust and comparable evidence on social inequalities in preventable lung cancer mortality.

Conclusion

Lung cancer mortality among men was stable across periods within deprivation levels, while a positive socioeconomic gradient — with higher risks in more deprived areas — was consistently observed and strengthened over time. Among women, rates increased between periods particularly in more deprived strata, and the inverse socioeconomic gradient observed in the first period attenuated by the second, suggesting a gradual convergence towards the pattern historically seen in men. These findings highlight the importance of monitoring small-area inequalities in preventable lung cancer mortality and support the continued implementation of targeted tobacco control strategies across all social strata, with particular attention to women in more deprived areas.

Addressing the structural determinants that shape these inequalities remains essential. Public health strategies should prioritize vulnerable populations, particularly those in more deprived areas, and should be sensitive to sex- and age-specific patterns of risk. The observed trends underscore the value of studying preventable mortality as a sensitive indicator of social inequity and of maintaining long-term surveillance to detect emerging gradients before they consolidate.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Pamela Pereyra-Zamora and Andreu Nolasco. The first draft of the manuscript was written by Javier Casillas-Clot and all authors commented on previous versions of the manuscript. All authors read and approved of the final manuscript.

Funding

This study is part of and has been partially funded by the research project "Socioeconomic changes and evolution of inequalities in mortality in small areas of large cities in the Valencian Community" (PI16/00670), financed by the Instituto de Salud Carlos III (co-financed by the European Regional Development Fund, ERDF).

Data availability

All population, and mortality data can be downloaded from the website of the Spanish National Institute of Statistics: [https://www.ine.es/] (https://www.ine.es). The georeferenced data are part of the MEDEA-3 project database: [https://www.uv.es/medea/medeapp.html] (https://www.uv.es/medea/medeapp.html).

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The data proceeds from secondary sources with anonymized databases, so approval from an ethics committee is not required for this study.