Patient profiles, incidence and trends of lung cancer in Ethiopia from 2012 to 2023 using a cancer registry.
1/5 보강
[UNLABELLED] Lung cancer epidemiology varies widely across countries, yet local data remain scarce.
- 95% CI 1.21–2.21
APA
Estifanos N, Egata G, et al. (2026). Patient profiles, incidence and trends of lung cancer in Ethiopia from 2012 to 2023 using a cancer registry.. Scientific reports, 16(1), 6175. https://doi.org/10.1038/s41598-026-36944-x
MLA
Estifanos N, et al.. "Patient profiles, incidence and trends of lung cancer in Ethiopia from 2012 to 2023 using a cancer registry.." Scientific reports, vol. 16, no. 1, 2026, pp. 6175.
PMID
41580502 ↗
Abstract 한글 요약
[UNLABELLED] Lung cancer epidemiology varies widely across countries, yet local data remain scarce. This study analyzed 882 lung cancer cases registered in the Addis Ababa population-based cancer registry from 2012 to 2023 to characterize patient profiles, incidence rates, and trends. We used descriptive statistics, binary and multinomial logistic regression, Poisson regression, and joinpoint regression. We found the crude median age at diagnosis was 56 years (adjusted: 60 years), with one in four patients diagnosed before 45. Adenocarcinoma was the most common histological subtype (34.8%), significantly associated with females (AOR: 1.64, 95% CI: 1.21–2.21), followed by squamous cell carcinoma (8.8%). Carcinoma not otherwise specified (NOS) accounted for 45.2% of cases. Alarmingly, 93% of patients were diagnosed at a late stage. The age-standardized incidence rate (ASIR) was 3.1 per 100,000 (3.3 for males, 2.8 for females), with a male-to-female incidence rate ratio (IRR) of 1.13, showing a nonsignificant decline over time. The incidence rate increased with age and varied significantly across sub-cities (1.5–2.9 per 100,000). While the overall ASIR trend remained stable, sex-specific trends showed a 3% annual increase in female crude incidence rates, whereas male rates remained unchanged significantly. These findings highlight the urgent need for targeted lung cancer prevention, early detection, and treatment strategies tailored to local epidemiological patterns.
[SUPPLEMENTARY INFORMATION] The online version contains supplementary material available at 10.1038/s41598-026-36944-x.
[SUPPLEMENTARY INFORMATION] The online version contains supplementary material available at 10.1038/s41598-026-36944-x.
🏷️ 키워드 / MeSH 📖 같은 키워드 OA만
같은 제1저자의 인용 많은 논문 (2)
📖 전문 본문 읽기 PMC JATS · ~74 KB · 영문
Introduction
Introduction
In sub-Saharan Africa (SSA), the burden of cancer is increasing due to a rapidly growing population, an increased lifespan, and a shift toward unhealthy lifestyle choices1,2. With a population exceeding 120 million, Ethiopia is Africa’s second most populous nation and is projected to become the seventh most populous country by 20503. Lung cancer was the leading cause of cancer morbidity and mortality in 2022, accounting for nearly one in eight cancer diagnoses and one in five cancer-related deaths worldwide4. According to the Global Cancer Observatory (GLOBOCAN) 2022 report, Ethiopia’s incidence and mortality constitute 3.1% and 4.4%, respectively4, and are expected to rise1.
Lung cancer arises from a combination of risk factors, including tobacco smoking, second-hand smoke, indoor and outdoor air pollution, occupational exposures, genetic predisposition, and chronic lung diseases such as tuberculosis4,5. Although smoking is the dominant global driver, in our context, only 25% of patients have a history of smoking6, highlighting the likely importance of other exposures. Despite the relatively low prevalence, smoking rates in Ethiopia are increasing7. Additionally, over 95% of the population relies on polluting fuels and technologies8, and the country’s average particulate matter (PM2.5) concentration is more than five times higher than the WHO’s air quality standard9. Ethiopia is also among the 30 countries with the highest burden of tuberculosis10, which may contribute further to lung cancer risk.
Lung cancer often presents with nonspecific symptoms, such as chronic cough, hemoptysis, chest pain, dyspnea, weight loss, and fatigue11, which may lead to misdiagnosis or late recognition. Early detection is critical, yet most patients in low- and middle-income countries (LMICs) are diagnosed at advanced stages, when treatment options are limited6. In settings with established screening programs, low-dose computed tomography (LDCT) has been shown to detect lung cancer earlier and reduce mortality12. However, such programs are not currently available in Ethiopia. Diagnosis typically relies on imaging, followed by histopathological confirmation through biopsy, but diagnostic capacity remains limited.
Despite the global evidence base, little is known about the epidemiology and clinical characteristics of lung cancer in LMICs like Ethiopia. Previous studies6,13,14 have been small in scale and limited in scope, often conducted in single institutions and with little generalizability. Gaps remain in understanding the demographic and clinical profiles of patients, as well as population-based incidence and trends. Addressing these evidence gaps is essential to inform national cancer control strategies15 and improve patient outcomes. Therefore, this work aims to address these gaps by documenting the distinct clinical profiles of patients, as well as the incidence and trends of lung cancer in Addis Ababa, Ethiopia, from 2012 to 2023.
In sub-Saharan Africa (SSA), the burden of cancer is increasing due to a rapidly growing population, an increased lifespan, and a shift toward unhealthy lifestyle choices1,2. With a population exceeding 120 million, Ethiopia is Africa’s second most populous nation and is projected to become the seventh most populous country by 20503. Lung cancer was the leading cause of cancer morbidity and mortality in 2022, accounting for nearly one in eight cancer diagnoses and one in five cancer-related deaths worldwide4. According to the Global Cancer Observatory (GLOBOCAN) 2022 report, Ethiopia’s incidence and mortality constitute 3.1% and 4.4%, respectively4, and are expected to rise1.
Lung cancer arises from a combination of risk factors, including tobacco smoking, second-hand smoke, indoor and outdoor air pollution, occupational exposures, genetic predisposition, and chronic lung diseases such as tuberculosis4,5. Although smoking is the dominant global driver, in our context, only 25% of patients have a history of smoking6, highlighting the likely importance of other exposures. Despite the relatively low prevalence, smoking rates in Ethiopia are increasing7. Additionally, over 95% of the population relies on polluting fuels and technologies8, and the country’s average particulate matter (PM2.5) concentration is more than five times higher than the WHO’s air quality standard9. Ethiopia is also among the 30 countries with the highest burden of tuberculosis10, which may contribute further to lung cancer risk.
Lung cancer often presents with nonspecific symptoms, such as chronic cough, hemoptysis, chest pain, dyspnea, weight loss, and fatigue11, which may lead to misdiagnosis or late recognition. Early detection is critical, yet most patients in low- and middle-income countries (LMICs) are diagnosed at advanced stages, when treatment options are limited6. In settings with established screening programs, low-dose computed tomography (LDCT) has been shown to detect lung cancer earlier and reduce mortality12. However, such programs are not currently available in Ethiopia. Diagnosis typically relies on imaging, followed by histopathological confirmation through biopsy, but diagnostic capacity remains limited.
Despite the global evidence base, little is known about the epidemiology and clinical characteristics of lung cancer in LMICs like Ethiopia. Previous studies6,13,14 have been small in scale and limited in scope, often conducted in single institutions and with little generalizability. Gaps remain in understanding the demographic and clinical profiles of patients, as well as population-based incidence and trends. Addressing these evidence gaps is essential to inform national cancer control strategies15 and improve patient outcomes. Therefore, this work aims to address these gaps by documenting the distinct clinical profiles of patients, as well as the incidence and trends of lung cancer in Addis Ababa, Ethiopia, from 2012 to 2023.
Methods
Methods
Study setting
This study was part of a multinational lung cancer diagnosis and control project in Ethiopia. Seven implementers collaborated with the Ministry of Health of Ethiopia and executed the project. According to the World Population Review, Addis Ababa’s population in 2025 is estimated to be around 6 million16. The city has twelve state hospitals, forty private hospitals, ninety-six health centers, and over eight hundred clinics, and is administratively organized into eleven sub-cities with 116 districts. Additionally, Addis Ababa is a referral site because of the radiotherapy services offered at Tikur Anbessa Specialized Hospital. Until April 2022, Tikur Anbessa Specialized Hospital was the only facility in the country that provided radiotherapy. The population-based cancer registry is also located at Tikur Anbessa Specialized Hospital. With funding and technical assistance from Martin Luther University, Halle, the registry was founded in 2012 as part of a research project focused on female reproductive system cancers. Since 2014, the registry has received funding from the American Cancer Society. It collects high-quality data published in Cancer Incidence in Five Continents (CI-5) Volume X, a publication of the International Agency for Research on Cancer (IARC).
Study design and data sources
This study is a population-based cancer registry study from 2012 to 2023 aimed at describing patient profiles and estimating the incidence and trends of lung cancer in Addis Ababa, Ethiopia. Since 2012, the registry has been collecting data on newly diagnosed cancer cases from medical facilities in Addis Ababa. Each facility that provides cancer diagnostic and treatment services has a trained contact person responsible for daily data collection. The registry’s permanent staff oversees and coordinates the entire data collection process. Data are gathered via a standardized tool developed by the IARC17. This form records demographic details such as age, sex, residence, and tumor-related information, including cancer topography, morphology, tumor behavior, the basis of diagnosis, and stage. The stage is documented using the TNM classification system specified in the American Joint Committee on Cancer (AJCC) staging manual. Cases diagnosed from 2012 to 2017 were staged according to the AJCC 7th edition, while cases from 2018 onward followed the AJCC 8th edition. The treatment administered at the time of registration was also recorded. Registry staff conduct regular follow-up visits to health facilities to ensure comprehensive case registration and maintain data quality. The registry uses CanReg5 software, developed by the IARC, for data management.
Each case was classified on the basis of the tumor’s topography (site of origin) and morphology (type of cancerous cell) via a coding scheme from the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3). The registry identifies lung cancer using the ICD-O-3 codes (C33–C34). In this study, the five histological types of lung cancer categorized on the basis of morphological codes were adenocarcinoma, squamous cell carcinoma, large cell carcinoma with other non-small cell carcinomas, small cell carcinoma, and other carcinomas. The ‘other carcinomas’ category comprised both other specified carcinomas (ICD-O codes 8200, 8240, 8246) and unspecified carcinomas (ICD-O codes 8010, 8011, 8020). There were 48 cases diagnosed clinically or by ultrasound, without histological or cytological confirmation (ICD-O code 8000, n = 48). Non-carcinoma primary lung tumors (n = 11) were excluded. For the calculation of carcinoma NOS proportions (ICD-O codes 8010, 8011, 8020), only histologically or cytologically confirmed cases were counted as carcinoma NOS (Supplementary Table S1). The population data were obtained from the Ethiopian Central Statistical Service Office.
Data processing and analysis
Qualitative variables, including sex, residence, histological subtype, stage, and basis of diagnosis, were summarized as counts and percentages. Age at diagnosis was categorized as < 45 years and ≥ 45 years. We selected 45 years as the cutoff point to reflect the substantial burden of lung cancer at younger ages in our cohort. Furthermore, there is no universally accepted cutoff for defining ‘young’ lung cancer patients; prior studies have used thresholds such as < 40 or < 50 years18. Age at diagnosis was also described as the median and interquartile range (Q1–Q3) and compared across demographic and clinical groups using the Kruskal–Wallis or Wilcoxon–Mann–Whitney tests. Median age was further adjusted using the UN World Population Prospects3 and Segi–Doll World Standard Population4 to enhance comparability19.
Associations between histological subtypes and patient characteristics were assessed using chi-square tests and multivariable multinomial logistic regression, with model fit and multicollinearity checks. Carcinoma NOS was analyzed using chi-square tests, binary logistic regression, and trends by sex (2012–2023) were evaluated using joinpoint regression. The proportion of late-stage diagnoses with 95% CIs was calculated. Inclusion or exclusion of the 48 clinically diagnosed cases can artificially inflate or deflate the proportion of histological subtypes, carcinoma NOS, and associated odds ratios. Therefore, we reported results both with and without these cases. The sensitivity analyses, excluding 48 cases diagnosed by clinical investigation or ultrasound, with full results provided in the Supplementary Material, including a frequency crosstab that shows the difference in proportion after exclusion.
To calculate the age-specific incidence rates from 2012 to 2023, we categorized age into five-year intervals (00–04, 05–09, …, 80–84, 85 + years) and by sex (combined sexes, male and female). We estimated the CIR, age-specific incidence rates, and ASIR by sex. The direct standardization method was used to calculate the ASIR via the Segi–Doll World standard population composition to facilitate comparisons with previous studies4,13,14,20. Rates are presented as cases per 100,000 person-years. The cumulative risk (%) of developing lung cancer before the age of 75 years was estimated, with no competing causes of mortality assumed.
Poisson regression assessed associations of age, sex, and residence with incidence rates, reporting IRRs with 95% CIs and model fit statistics. Trends in the male-to-female incidence ratio were analyzed using joinpoint regression.
Lung cancer incidence trends (2012–2023) were analyzed using Joinpoint Regression Software (version 5.2.0.0, NCI), which identifies statistically significant changes in trends (“joinpoints”)21 using Monte Carlo permutation tests (4,999 permutations)21,22. Annual percent change (APC) and average annual percent change (AAPC) with 95% CIs were calculated, model assumptions were checked, and trends were classified as stable, rising, or falling per National Cancer Institute guidelines (NCI)23. All analyses were two-sided, with p < 0.05 considered significant. All other analyses in the manuscript were conducted using Stata version 16. Reporting followed the STROBE checklist for observational studies (Supplementary STROBE checklist).
Ethical considerations
The Institutional Review Board (IRB) of Addis Ababa University’s College of Health Sciences granted ethical approval under protocol number 073/23/SPH. The IRB conducted all procedures in accordance with the Declaration of Helsinki, International Conferences on Harmonization Good Clinical Practice guidelines, WHO Operating Guidelines for Ethical Review Committees, and the National Guideline for Research Ethics in Ethiopia. Data from Addis Ababa’s population-based cancer registry were used in this study. Because of the nature of the study, patient consent to participate was not applicable. The data used in this research were anonymized and deidentified before access to ensure the confidentiality and privacy of individuals.
Study setting
This study was part of a multinational lung cancer diagnosis and control project in Ethiopia. Seven implementers collaborated with the Ministry of Health of Ethiopia and executed the project. According to the World Population Review, Addis Ababa’s population in 2025 is estimated to be around 6 million16. The city has twelve state hospitals, forty private hospitals, ninety-six health centers, and over eight hundred clinics, and is administratively organized into eleven sub-cities with 116 districts. Additionally, Addis Ababa is a referral site because of the radiotherapy services offered at Tikur Anbessa Specialized Hospital. Until April 2022, Tikur Anbessa Specialized Hospital was the only facility in the country that provided radiotherapy. The population-based cancer registry is also located at Tikur Anbessa Specialized Hospital. With funding and technical assistance from Martin Luther University, Halle, the registry was founded in 2012 as part of a research project focused on female reproductive system cancers. Since 2014, the registry has received funding from the American Cancer Society. It collects high-quality data published in Cancer Incidence in Five Continents (CI-5) Volume X, a publication of the International Agency for Research on Cancer (IARC).
Study design and data sources
This study is a population-based cancer registry study from 2012 to 2023 aimed at describing patient profiles and estimating the incidence and trends of lung cancer in Addis Ababa, Ethiopia. Since 2012, the registry has been collecting data on newly diagnosed cancer cases from medical facilities in Addis Ababa. Each facility that provides cancer diagnostic and treatment services has a trained contact person responsible for daily data collection. The registry’s permanent staff oversees and coordinates the entire data collection process. Data are gathered via a standardized tool developed by the IARC17. This form records demographic details such as age, sex, residence, and tumor-related information, including cancer topography, morphology, tumor behavior, the basis of diagnosis, and stage. The stage is documented using the TNM classification system specified in the American Joint Committee on Cancer (AJCC) staging manual. Cases diagnosed from 2012 to 2017 were staged according to the AJCC 7th edition, while cases from 2018 onward followed the AJCC 8th edition. The treatment administered at the time of registration was also recorded. Registry staff conduct regular follow-up visits to health facilities to ensure comprehensive case registration and maintain data quality. The registry uses CanReg5 software, developed by the IARC, for data management.
Each case was classified on the basis of the tumor’s topography (site of origin) and morphology (type of cancerous cell) via a coding scheme from the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3). The registry identifies lung cancer using the ICD-O-3 codes (C33–C34). In this study, the five histological types of lung cancer categorized on the basis of morphological codes were adenocarcinoma, squamous cell carcinoma, large cell carcinoma with other non-small cell carcinomas, small cell carcinoma, and other carcinomas. The ‘other carcinomas’ category comprised both other specified carcinomas (ICD-O codes 8200, 8240, 8246) and unspecified carcinomas (ICD-O codes 8010, 8011, 8020). There were 48 cases diagnosed clinically or by ultrasound, without histological or cytological confirmation (ICD-O code 8000, n = 48). Non-carcinoma primary lung tumors (n = 11) were excluded. For the calculation of carcinoma NOS proportions (ICD-O codes 8010, 8011, 8020), only histologically or cytologically confirmed cases were counted as carcinoma NOS (Supplementary Table S1). The population data were obtained from the Ethiopian Central Statistical Service Office.
Data processing and analysis
Qualitative variables, including sex, residence, histological subtype, stage, and basis of diagnosis, were summarized as counts and percentages. Age at diagnosis was categorized as < 45 years and ≥ 45 years. We selected 45 years as the cutoff point to reflect the substantial burden of lung cancer at younger ages in our cohort. Furthermore, there is no universally accepted cutoff for defining ‘young’ lung cancer patients; prior studies have used thresholds such as < 40 or < 50 years18. Age at diagnosis was also described as the median and interquartile range (Q1–Q3) and compared across demographic and clinical groups using the Kruskal–Wallis or Wilcoxon–Mann–Whitney tests. Median age was further adjusted using the UN World Population Prospects3 and Segi–Doll World Standard Population4 to enhance comparability19.
Associations between histological subtypes and patient characteristics were assessed using chi-square tests and multivariable multinomial logistic regression, with model fit and multicollinearity checks. Carcinoma NOS was analyzed using chi-square tests, binary logistic regression, and trends by sex (2012–2023) were evaluated using joinpoint regression. The proportion of late-stage diagnoses with 95% CIs was calculated. Inclusion or exclusion of the 48 clinically diagnosed cases can artificially inflate or deflate the proportion of histological subtypes, carcinoma NOS, and associated odds ratios. Therefore, we reported results both with and without these cases. The sensitivity analyses, excluding 48 cases diagnosed by clinical investigation or ultrasound, with full results provided in the Supplementary Material, including a frequency crosstab that shows the difference in proportion after exclusion.
To calculate the age-specific incidence rates from 2012 to 2023, we categorized age into five-year intervals (00–04, 05–09, …, 80–84, 85 + years) and by sex (combined sexes, male and female). We estimated the CIR, age-specific incidence rates, and ASIR by sex. The direct standardization method was used to calculate the ASIR via the Segi–Doll World standard population composition to facilitate comparisons with previous studies4,13,14,20. Rates are presented as cases per 100,000 person-years. The cumulative risk (%) of developing lung cancer before the age of 75 years was estimated, with no competing causes of mortality assumed.
Poisson regression assessed associations of age, sex, and residence with incidence rates, reporting IRRs with 95% CIs and model fit statistics. Trends in the male-to-female incidence ratio were analyzed using joinpoint regression.
Lung cancer incidence trends (2012–2023) were analyzed using Joinpoint Regression Software (version 5.2.0.0, NCI), which identifies statistically significant changes in trends (“joinpoints”)21 using Monte Carlo permutation tests (4,999 permutations)21,22. Annual percent change (APC) and average annual percent change (AAPC) with 95% CIs were calculated, model assumptions were checked, and trends were classified as stable, rising, or falling per National Cancer Institute guidelines (NCI)23. All analyses were two-sided, with p < 0.05 considered significant. All other analyses in the manuscript were conducted using Stata version 16. Reporting followed the STROBE checklist for observational studies (Supplementary STROBE checklist).
Ethical considerations
The Institutional Review Board (IRB) of Addis Ababa University’s College of Health Sciences granted ethical approval under protocol number 073/23/SPH. The IRB conducted all procedures in accordance with the Declaration of Helsinki, International Conferences on Harmonization Good Clinical Practice guidelines, WHO Operating Guidelines for Ethical Review Committees, and the National Guideline for Research Ethics in Ethiopia. Data from Addis Ababa’s population-based cancer registry were used in this study. Because of the nature of the study, patient consent to participate was not applicable. The data used in this research were anonymized and deidentified before access to ensure the confidentiality and privacy of individuals.
Results
Results
Sociodemographic and clinical characteristics of the patients
From 2012 to 2023, 882 lung cancer cases were registered in the Addis Ababa population-based cancer registry. The proportion of morphologically verified lung cancer cases (MV%) was 94.5%. The median age (Q1, Q3) at diagnosis was 56 (45, 65) years. There were 207 (23.5%) patients under 45 years of age and 463 (52.5%) males. The highest number of cases was from the Bole sub-city (135, 15.3%), and the lowest was from the Arada sub-city (49, 5.55%). In terms of the year of diagnosis, the highest number of registered cases was 102 (11.6%) in 2019, whereas the lowest was 54 (6.1%) in 2012. Notably, 31.2% of cases lacked histological confirmation; 25.7% were diagnosed using cytology, and 5.5% were diagnosed based on clinical or ultrasound findings. Adenocarcinoma was the predominant specific subtype (34.8%), while carcinoma NOS accounted for 399 cases (45.2%). The stages of disease for 719 (81.3%) patients were not recorded, and among those recorded (163), 138 patients were diagnosed at stage IV (Table 1).
Sociodemographic and clinical characteristics of the patients
From 2012 to 2023, 882 lung cancer cases were registered in the Addis Ababa population-based cancer registry. The proportion of morphologically verified lung cancer cases (MV%) was 94.5%. The median age (Q1, Q3) at diagnosis was 56 (45, 65) years. There were 207 (23.5%) patients under 45 years of age and 463 (52.5%) males. The highest number of cases was from the Bole sub-city (135, 15.3%), and the lowest was from the Arada sub-city (49, 5.55%). In terms of the year of diagnosis, the highest number of registered cases was 102 (11.6%) in 2019, whereas the lowest was 54 (6.1%) in 2012. Notably, 31.2% of cases lacked histological confirmation; 25.7% were diagnosed using cytology, and 5.5% were diagnosed based on clinical or ultrasound findings. Adenocarcinoma was the predominant specific subtype (34.8%), while carcinoma NOS accounted for 399 cases (45.2%). The stages of disease for 719 (81.3%) patients were not recorded, and among those recorded (163), 138 patients were diagnosed at stage IV (Table 1).
Patient profiles: age at diagnosis, histological subtypes, carcinoma NOS, and stage
Patient profiles: age at diagnosis, histological subtypes, carcinoma NOS, and stage
Age at diagnosis
We used the median age along with the range instead of the mean with standard deviation, as the age at diagnosis was negatively skewed. The crude median age (Q2, Q3) at diagnosis was 56 (45, 65). After adjusting for age according to Prospects 2010 and the Segi–Doll World standard population, the median age at diagnosis (IQR) was 60 (53, 65) and 60 (52, 66), respectively.
The crude median age at which women were diagnosed was two years younger than that of men (55 years compared with 57 years). However, the age distribution did not differ significantly between males and females. The age distribution at diagnosis varied among different histological subtypes (p = 0.021, Kruskal–Wallis test). The median age of patients with adenocarcinoma was the highest [median age (Q1, Q3) = 56 (48, 65)], whereas that of patients with small cell carcinoma was the lowest [median age (Q1, Q3) = 41 (30, 57)]. No significant differences in the age distribution at diagnosis were observed based on sex, year of diagnosis, or disease stage (Supplementary Table S2).
The percentage of younger adults under 45 years of age diagnosed with lung cancer was 25% (95% CI: 20.8, 26.4). Additionally, 50% of the patients were under 56 years of age, and 75% were under 72 years.
Histological subtypes
Despite one-third of cases lacking histological confirmation, adenocarcinoma emerged as the most common subtype (34.8%, 95% CI: 31.7–38.0), followed by squamous cell carcinoma (8.8%, 95% CI: 7.1–10.9). A significant association was identified between histological subtypes and factors such as sex, age, year of diagnosis, and basis of diagnosis, as determined by the chi-square association test. Notably, a greater frequency of adenocarcinoma was observed in females (40.8%) than in males (29.4%), whereas males presented a greater frequency of squamous cell carcinoma (10.4%) than females did (7.2%) (p = 0.004, chi-square test). In terms of age categories, individuals under 45 years of age had a lower frequency of adenocarcinoma (29.5%) than did those aged 45 years or older (36.4%) (p = 0.038, chi-square test). When the basis of the diagnosis was examined, patients who were diagnosed histologically had a significantly greater frequency of adenocarcinoma (39.4%) than those who were diagnosed non-histologically (24.7%) (p = < 0.001, chi-square test) (Supplementary Table S3.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations (Supplementary Table S3.2).
Variables that had a significant association with histological subtypes during the chi-square test of association were retained for the multivariable multinomial logistic regression analysis. The histological subtypes were classified into four categories: adenocarcinoma, squamous cell carcinoma, large cell carcinoma with other NSCC, and other carcinomas, and clinically diagnosed, with the latter serving as the reference category. According to multivariable multinomial logistic regression, Females had higher odds of adenocarcinoma compared to males [(AOR = 1.67, 95% CI: (1.24, 2.25)], p = 0.001. Compared with a non-histologic workup, a histologic workup yields more specific subtypes, such as adenocarcinoma (AOR = 2.47, 95% CI: (1.77,3.45), p = < 0.001; squamous cell carcinoma [(AOR = 3.84, 95% CI: (2.02,7.33)], p = < 0.001 and large with other NSCCs [(AOR = 2.80, 95% CI: (1.32,6.10)], p = 0.007, than other carcinomas and clinically diagnosed do (Supplementary Table S4.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations, except for large and other NSCCs, for which the p-value became 0.063, indicating a borderline effect (Supplementary Table S4.2).
Carcinoma NOS
In this study, lung cancers with morphology codes 8010, 8011, and 8020, diagnosed either histologically or cytologically, were classified as carcinoma NOS. The overall percentage of carcinoma NOS was 45.2% (95% CI: 42.97, 48.54). During the chi-square test of association, the year of diagnosis demonstrated a significant association with carcinoma NOS (p = < 0.001). The basis of diagnosis also showed a significant association, where histologically diagnosed patients had a lower frequency of carcinoma NOS (37.4%) compared to those diagnosed via non-histologic means (62.5%; p = < 0.001, chi-square test) (Supplementary Table S5.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations (Supplementary Table S5.2).
According to logistic regression, a lower likelihood of carcinoma NOS histological subtypes was associated with histologic methods of diagnosis than with non-histologic methods [(COR = 0.36, 95% CI: (0.26, 0.48)], p = < 0.001). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of association (Supplementary Table S6).
Using the joinpoint regression model, we tested the time trend of the percentage of carcinoma NOS cases from 2012 to 2023. Its trend exhibited a significant average annual decrease of 8% from 2012 to 2023 for both sexes combined [(AAPC=-8.03, 95% CI: (-10.11, -5.74)], p = < 0.001 and male [(AAPC=-7.93, 95% CI: (-11.26, -4.26)], p = < 0.0010. For females, there was a nonsignificant decrease of 6% [(AAPC=-5.88, 95% CI: (-12.22, 0.74)], p = 0.070. When the APC was observed, for both sexes combined, there was a nonsignificant decline from 2012 to 2021. However, there was a notable decline in the APC of -29.8% from 2021 to 2023 [(APC=-29.76, 95% CI: (-39.12, 14.59)], p = < 0.001. In males, the analysis indicated no significant change from 2012 to 2021 [(APC=-1.11, 95% CI: (-4.05, 5.80)], P value = 0.739, but a substantial decline was observed from 2021 to 2023 [(APC=-33.26%, 95% CI: (-46.70, -12.88)], p = < 0.001. For females, the trend from 2012 to 2017 showed a slight nonsignificant increase [(APC = 4.04%, 95% CI: (-5.39, 58.24)], p = 0.403. However, a significant negative trend was identified from 2017 to 2023 [(APC=-13.42%, 95% CI: (-43.13, -6.78)], p = 0.012 (Fig. 1, Supplementary Tables S7 and S8).
Stage at diagnosis
We classified stages I and II as early stages, whereas stages III and IV were considered late stages. The disease stage was not recorded for 719 patients (81.5%). Among those with recorded stages (163), 152 patients (93%) (95% CI: 88.2, 96.2) were diagnosed at late stages. No stage I patients were recorded from 2012 to 2023. Only 7% (11/163) of patients were diagnosed with stage II disease, including 7 males and 4 females. Stage III, 8.6% (14/163) and Stage IV, 85% (138/163).
Incidence: overall incidence and incidence by sex, age, and residence
Overall incidence
From 2012 to 2023, the CIR was 2.1 per 100,000 persons, with a rate of 2.4 per 100,000 for males and 1.9 per 100,000 for females. The ASIR was 3.1 per 100,000, 3.3 per 100,000 for males, and 2.8 per 100,000 for females. The cumulative risk of developing lung cancer before the age of 74 years was 0.38%, with 0.4% for males and 3.6% for females (Tables 2).
Incidence by sex
The crude male-to-female IRR was 1.24 (95% CI: 1.10, 1.41; p = 0.001), whereas the age-standardized IRR was 1.13 (95% CI: 1.01, 1.26; p = 0.030). Over the period from 2012 to 2023, the AAPC in the crude IRR was − 3.57 (95% CI: -8.26, 1.23; p = 0.148), and the AAPC in the age-standardized IRR was − 2.79 (95% CI: -8.67, -3.32; p = 0.335).
Incidence by age
Figure 2 presents the estimated age-specific incidence rates by five-year age group and gender for Addis Ababa, Ethiopia, from 2012 to 2023. The data indicate that there are no reported cases in the younger age groups (0–14 years). The first recorded cases appeared in the 15–19 age group, with an age-specific rate of 0.03 per 100,000 for both sexes combined and a slightly higher rate of 0.06 per 100,000 for females. As age increases, the age-specific incidence rates rise. The highest age-specific rate for both sexes combined is observed in the 70–74 age group at 19.59 per 100,000, followed closely by the 65–69 age group at 16.38 per 100,000. Males had the highest incidence rate in the 75–79 years age group (20.94 per 100,000), and females had a peak diagnosis rate slightly earlier, in the 70–74 years age group (19.26 per 100,000) (Fig. 2, Supplementary Table S9).
The Poisson regression model was used to calculate the IRRs with 95% CIs for the age-specific incidence rates for each age category. Age categories 0–4, 5–9, 10–14, 15–19, and 20–24 were excluded since very few cases (14) were recorded in this age category. The 25–29 years age category was used as the reference category. The model likelihood ratio chi-square statistic of 604.969 was highly significant (p < 0.001). This suggests that the model incorporating predictors has a significantly improved fit compared with the intercept-only model. Furthermore, the goodness-of-fit tests revealed favorable results, with the deviance goodness-of-fit statistic at 146.6949 (p = 0.1805) and the Pearson goodness-of-fit statistic at 129.8822 (p = 0.5358). Both p values indicate that the model fits the data well, supporting the validity of the findings. Supplementary Table S10 presents the IRRs for age-specific incidence rates by five-year age group and sex for Addis Ababa, Ethiopia, from 2012 to 2023. As age increases, the IRRs rise. The highest IRR for both sexes was observed at 70–74 years of age (IRR = 37.14 (25.50, 54.10). Males presented the highest IRR in the 75–79 years age group (IRR = 44.62 (27.19, 73.26), whereas females presented a peak diagnosis rate slightly earlier in the 70–74 years age group (IRR = 38.04 (24.45, 59.19) (Supplementary Table S10).
Incidence by residence
The analysis of CIRs per 100,000 population across various sub-cities from 2012 to 2023 revealed significant disparities in incidence rates. The lowest CIR was observed in Arada, with 49 cases registered from 2012 to 2023, resulting in a CIR of 1.526 per 100,000. This served as the reference category for estimating the incidence rate ratio (IRR) via the Poisson regression model, which fit the data well. Kolfe Keranio recorded 115 cases, yielding a CIR of 1.766 per 100,000, whereas Gulele reported 79 cases, resulting in a CIR of 1.944 per 100,000. Compared with those of Arada, the IRRs of Yeka, Lideta, and Addis Ketema were significantly greater, with CIRs of 40.7%, 47.5%, and 50.4%, respectively. Compared with Arada, Bole presented the highest CIR at 2.877 per 100,000, with 135 cases, representing an 88.6% increase. This resulted in an IRR of 1.886 (95% CI: 1.807, 1.968), which was highly significant (p = < 0.001) (Table 3).
Trends of lung cancer in addis Ababa, Ethiopia, from 2012 to 2023
This study examined the annual number of diagnosed cases from 2012 to 2023, providing insights into the CIR and ASIR trends across different sexes. In 2012, 54 cases were reported for both sexes, with a CIR of 1.77 and an ASIR of 2.98. The number of cases fluctuated over the years, peaking at 102 in 2019, which also recorded the highest CIR at 2.85 and the highest ASIR for both sexes at 4.13. With respect to sex-specific trends, male cases peaked in 2019, with a CIR of 3.18 and an ASIR of 4.51. However, these rates declined in 2020, followed by a gradual recovery, with the CIR rising to 2.53 and the ASIR to 3.47 by 2022. The peak rate for females also occurred in 2019, with a CIR of 2.58 and an ASIR of 3.62. The ASIR for females remained relatively stable in the following years, reaching 2.74 in 2023 (Table 4).
Using the joinpoint regression model, we assessed the time trends of both the CIR and the ASIR from 2012 to 2023. For both sexes combined, the AAPC from 2012 to 2023 was 0.74 (AAPC = 0.74, 95% CI: (-1.27, 3.21)], p = 0.425, indicating a nonsignificant increase in incidence rates. In males, the AAPC was − 0.50 [(AAPC= -0.50, (95% CI: -4.20, to 3.4843)], p = 0.8142, suggesting a slight decline in incidence rates that was not statistically significant. Conversely, females presented an increase in incidence rates of 3.0098 [(AAPC = 3.00, CI: (0.16, 6.09)], p = 0.039.
The CIRs for both sexes demonstrated a significant upward trend from 2012 to 2019, with an APC of 4.04% [(APC = 4.04, 95% CI: (-1.12, 16.67)], p = 0.0243. However, from 2019 to 2023, the trend reversed, exhibiting a nonsignificant decline with an APC of -4.78% [(APC = -4.78, 95% CI: (-16.45, 0.57)], p = 0.083. When trends by sex were examined, the analysis revealed that males had a stable incidence rate over the entire study period from 2012 to 2023, with an APC of -0.50% [(APC = -0.50, 95% CI: (-4.20, 3.48)], p = 0.814. In contrast, females experienced a significant increase in the CIR, with an APC of 3.01% [(APC = 3.01, 95% CI: (0.16 to 6.08)], p = 0.039 throughout the same timeframe (Fig. 3A, Supplementary Tables S11 and S12).
The ASIRs were analyzed over the entire range from 2012 to 2023, revealing a negative average annual percent change (AAPC) of -1.87% for both sexes combined. This decline was not statistically significant (p = 0.180). When the data were disaggregated by sex, the analysis for males revealed a more pronounced nonsignificant decrease, with an AAPC of -2.92% (p = 0.301). For females, the AAPC was − 0.29%, with a confidence interval of -2.87% to 2.52%, further reflecting a lack of significant change (p = 0.907). The APC is equivalent to the AAPC since the selected model has zero joinpoints (Fig. 3B, Supplementary Tables S13 and S14).
Age at diagnosis
We used the median age along with the range instead of the mean with standard deviation, as the age at diagnosis was negatively skewed. The crude median age (Q2, Q3) at diagnosis was 56 (45, 65). After adjusting for age according to Prospects 2010 and the Segi–Doll World standard population, the median age at diagnosis (IQR) was 60 (53, 65) and 60 (52, 66), respectively.
The crude median age at which women were diagnosed was two years younger than that of men (55 years compared with 57 years). However, the age distribution did not differ significantly between males and females. The age distribution at diagnosis varied among different histological subtypes (p = 0.021, Kruskal–Wallis test). The median age of patients with adenocarcinoma was the highest [median age (Q1, Q3) = 56 (48, 65)], whereas that of patients with small cell carcinoma was the lowest [median age (Q1, Q3) = 41 (30, 57)]. No significant differences in the age distribution at diagnosis were observed based on sex, year of diagnosis, or disease stage (Supplementary Table S2).
The percentage of younger adults under 45 years of age diagnosed with lung cancer was 25% (95% CI: 20.8, 26.4). Additionally, 50% of the patients were under 56 years of age, and 75% were under 72 years.
Histological subtypes
Despite one-third of cases lacking histological confirmation, adenocarcinoma emerged as the most common subtype (34.8%, 95% CI: 31.7–38.0), followed by squamous cell carcinoma (8.8%, 95% CI: 7.1–10.9). A significant association was identified between histological subtypes and factors such as sex, age, year of diagnosis, and basis of diagnosis, as determined by the chi-square association test. Notably, a greater frequency of adenocarcinoma was observed in females (40.8%) than in males (29.4%), whereas males presented a greater frequency of squamous cell carcinoma (10.4%) than females did (7.2%) (p = 0.004, chi-square test). In terms of age categories, individuals under 45 years of age had a lower frequency of adenocarcinoma (29.5%) than did those aged 45 years or older (36.4%) (p = 0.038, chi-square test). When the basis of the diagnosis was examined, patients who were diagnosed histologically had a significantly greater frequency of adenocarcinoma (39.4%) than those who were diagnosed non-histologically (24.7%) (p = < 0.001, chi-square test) (Supplementary Table S3.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations (Supplementary Table S3.2).
Variables that had a significant association with histological subtypes during the chi-square test of association were retained for the multivariable multinomial logistic regression analysis. The histological subtypes were classified into four categories: adenocarcinoma, squamous cell carcinoma, large cell carcinoma with other NSCC, and other carcinomas, and clinically diagnosed, with the latter serving as the reference category. According to multivariable multinomial logistic regression, Females had higher odds of adenocarcinoma compared to males [(AOR = 1.67, 95% CI: (1.24, 2.25)], p = 0.001. Compared with a non-histologic workup, a histologic workup yields more specific subtypes, such as adenocarcinoma (AOR = 2.47, 95% CI: (1.77,3.45), p = < 0.001; squamous cell carcinoma [(AOR = 3.84, 95% CI: (2.02,7.33)], p = < 0.001 and large with other NSCCs [(AOR = 2.80, 95% CI: (1.32,6.10)], p = 0.007, than other carcinomas and clinically diagnosed do (Supplementary Table S4.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations, except for large and other NSCCs, for which the p-value became 0.063, indicating a borderline effect (Supplementary Table S4.2).
Carcinoma NOS
In this study, lung cancers with morphology codes 8010, 8011, and 8020, diagnosed either histologically or cytologically, were classified as carcinoma NOS. The overall percentage of carcinoma NOS was 45.2% (95% CI: 42.97, 48.54). During the chi-square test of association, the year of diagnosis demonstrated a significant association with carcinoma NOS (p = < 0.001). The basis of diagnosis also showed a significant association, where histologically diagnosed patients had a lower frequency of carcinoma NOS (37.4%) compared to those diagnosed via non-histologic means (62.5%; p = < 0.001, chi-square test) (Supplementary Table S5.1). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of associations (Supplementary Table S5.2).
According to logistic regression, a lower likelihood of carcinoma NOS histological subtypes was associated with histologic methods of diagnosis than with non-histologic methods [(COR = 0.36, 95% CI: (0.26, 0.48)], p = < 0.001). Sensitivity analyses excluding the 48 cases diagnosed via clinical investigation or ultrasound showed no change in the direction or significance of association (Supplementary Table S6).
Using the joinpoint regression model, we tested the time trend of the percentage of carcinoma NOS cases from 2012 to 2023. Its trend exhibited a significant average annual decrease of 8% from 2012 to 2023 for both sexes combined [(AAPC=-8.03, 95% CI: (-10.11, -5.74)], p = < 0.001 and male [(AAPC=-7.93, 95% CI: (-11.26, -4.26)], p = < 0.0010. For females, there was a nonsignificant decrease of 6% [(AAPC=-5.88, 95% CI: (-12.22, 0.74)], p = 0.070. When the APC was observed, for both sexes combined, there was a nonsignificant decline from 2012 to 2021. However, there was a notable decline in the APC of -29.8% from 2021 to 2023 [(APC=-29.76, 95% CI: (-39.12, 14.59)], p = < 0.001. In males, the analysis indicated no significant change from 2012 to 2021 [(APC=-1.11, 95% CI: (-4.05, 5.80)], P value = 0.739, but a substantial decline was observed from 2021 to 2023 [(APC=-33.26%, 95% CI: (-46.70, -12.88)], p = < 0.001. For females, the trend from 2012 to 2017 showed a slight nonsignificant increase [(APC = 4.04%, 95% CI: (-5.39, 58.24)], p = 0.403. However, a significant negative trend was identified from 2017 to 2023 [(APC=-13.42%, 95% CI: (-43.13, -6.78)], p = 0.012 (Fig. 1, Supplementary Tables S7 and S8).
Stage at diagnosis
We classified stages I and II as early stages, whereas stages III and IV were considered late stages. The disease stage was not recorded for 719 patients (81.5%). Among those with recorded stages (163), 152 patients (93%) (95% CI: 88.2, 96.2) were diagnosed at late stages. No stage I patients were recorded from 2012 to 2023. Only 7% (11/163) of patients were diagnosed with stage II disease, including 7 males and 4 females. Stage III, 8.6% (14/163) and Stage IV, 85% (138/163).
Incidence: overall incidence and incidence by sex, age, and residence
Overall incidence
From 2012 to 2023, the CIR was 2.1 per 100,000 persons, with a rate of 2.4 per 100,000 for males and 1.9 per 100,000 for females. The ASIR was 3.1 per 100,000, 3.3 per 100,000 for males, and 2.8 per 100,000 for females. The cumulative risk of developing lung cancer before the age of 74 years was 0.38%, with 0.4% for males and 3.6% for females (Tables 2).
Incidence by sex
The crude male-to-female IRR was 1.24 (95% CI: 1.10, 1.41; p = 0.001), whereas the age-standardized IRR was 1.13 (95% CI: 1.01, 1.26; p = 0.030). Over the period from 2012 to 2023, the AAPC in the crude IRR was − 3.57 (95% CI: -8.26, 1.23; p = 0.148), and the AAPC in the age-standardized IRR was − 2.79 (95% CI: -8.67, -3.32; p = 0.335).
Incidence by age
Figure 2 presents the estimated age-specific incidence rates by five-year age group and gender for Addis Ababa, Ethiopia, from 2012 to 2023. The data indicate that there are no reported cases in the younger age groups (0–14 years). The first recorded cases appeared in the 15–19 age group, with an age-specific rate of 0.03 per 100,000 for both sexes combined and a slightly higher rate of 0.06 per 100,000 for females. As age increases, the age-specific incidence rates rise. The highest age-specific rate for both sexes combined is observed in the 70–74 age group at 19.59 per 100,000, followed closely by the 65–69 age group at 16.38 per 100,000. Males had the highest incidence rate in the 75–79 years age group (20.94 per 100,000), and females had a peak diagnosis rate slightly earlier, in the 70–74 years age group (19.26 per 100,000) (Fig. 2, Supplementary Table S9).
The Poisson regression model was used to calculate the IRRs with 95% CIs for the age-specific incidence rates for each age category. Age categories 0–4, 5–9, 10–14, 15–19, and 20–24 were excluded since very few cases (14) were recorded in this age category. The 25–29 years age category was used as the reference category. The model likelihood ratio chi-square statistic of 604.969 was highly significant (p < 0.001). This suggests that the model incorporating predictors has a significantly improved fit compared with the intercept-only model. Furthermore, the goodness-of-fit tests revealed favorable results, with the deviance goodness-of-fit statistic at 146.6949 (p = 0.1805) and the Pearson goodness-of-fit statistic at 129.8822 (p = 0.5358). Both p values indicate that the model fits the data well, supporting the validity of the findings. Supplementary Table S10 presents the IRRs for age-specific incidence rates by five-year age group and sex for Addis Ababa, Ethiopia, from 2012 to 2023. As age increases, the IRRs rise. The highest IRR for both sexes was observed at 70–74 years of age (IRR = 37.14 (25.50, 54.10). Males presented the highest IRR in the 75–79 years age group (IRR = 44.62 (27.19, 73.26), whereas females presented a peak diagnosis rate slightly earlier in the 70–74 years age group (IRR = 38.04 (24.45, 59.19) (Supplementary Table S10).
Incidence by residence
The analysis of CIRs per 100,000 population across various sub-cities from 2012 to 2023 revealed significant disparities in incidence rates. The lowest CIR was observed in Arada, with 49 cases registered from 2012 to 2023, resulting in a CIR of 1.526 per 100,000. This served as the reference category for estimating the incidence rate ratio (IRR) via the Poisson regression model, which fit the data well. Kolfe Keranio recorded 115 cases, yielding a CIR of 1.766 per 100,000, whereas Gulele reported 79 cases, resulting in a CIR of 1.944 per 100,000. Compared with those of Arada, the IRRs of Yeka, Lideta, and Addis Ketema were significantly greater, with CIRs of 40.7%, 47.5%, and 50.4%, respectively. Compared with Arada, Bole presented the highest CIR at 2.877 per 100,000, with 135 cases, representing an 88.6% increase. This resulted in an IRR of 1.886 (95% CI: 1.807, 1.968), which was highly significant (p = < 0.001) (Table 3).
Trends of lung cancer in addis Ababa, Ethiopia, from 2012 to 2023
This study examined the annual number of diagnosed cases from 2012 to 2023, providing insights into the CIR and ASIR trends across different sexes. In 2012, 54 cases were reported for both sexes, with a CIR of 1.77 and an ASIR of 2.98. The number of cases fluctuated over the years, peaking at 102 in 2019, which also recorded the highest CIR at 2.85 and the highest ASIR for both sexes at 4.13. With respect to sex-specific trends, male cases peaked in 2019, with a CIR of 3.18 and an ASIR of 4.51. However, these rates declined in 2020, followed by a gradual recovery, with the CIR rising to 2.53 and the ASIR to 3.47 by 2022. The peak rate for females also occurred in 2019, with a CIR of 2.58 and an ASIR of 3.62. The ASIR for females remained relatively stable in the following years, reaching 2.74 in 2023 (Table 4).
Using the joinpoint regression model, we assessed the time trends of both the CIR and the ASIR from 2012 to 2023. For both sexes combined, the AAPC from 2012 to 2023 was 0.74 (AAPC = 0.74, 95% CI: (-1.27, 3.21)], p = 0.425, indicating a nonsignificant increase in incidence rates. In males, the AAPC was − 0.50 [(AAPC= -0.50, (95% CI: -4.20, to 3.4843)], p = 0.8142, suggesting a slight decline in incidence rates that was not statistically significant. Conversely, females presented an increase in incidence rates of 3.0098 [(AAPC = 3.00, CI: (0.16, 6.09)], p = 0.039.
The CIRs for both sexes demonstrated a significant upward trend from 2012 to 2019, with an APC of 4.04% [(APC = 4.04, 95% CI: (-1.12, 16.67)], p = 0.0243. However, from 2019 to 2023, the trend reversed, exhibiting a nonsignificant decline with an APC of -4.78% [(APC = -4.78, 95% CI: (-16.45, 0.57)], p = 0.083. When trends by sex were examined, the analysis revealed that males had a stable incidence rate over the entire study period from 2012 to 2023, with an APC of -0.50% [(APC = -0.50, 95% CI: (-4.20, 3.48)], p = 0.814. In contrast, females experienced a significant increase in the CIR, with an APC of 3.01% [(APC = 3.01, 95% CI: (0.16 to 6.08)], p = 0.039 throughout the same timeframe (Fig. 3A, Supplementary Tables S11 and S12).
The ASIRs were analyzed over the entire range from 2012 to 2023, revealing a negative average annual percent change (AAPC) of -1.87% for both sexes combined. This decline was not statistically significant (p = 0.180). When the data were disaggregated by sex, the analysis for males revealed a more pronounced nonsignificant decrease, with an AAPC of -2.92% (p = 0.301). For females, the AAPC was − 0.29%, with a confidence interval of -2.87% to 2.52%, further reflecting a lack of significant change (p = 0.907). The APC is equivalent to the AAPC since the selected model has zero joinpoints (Fig. 3B, Supplementary Tables S13 and S14).
Discussion
Discussion
The ultimate goal of understanding lung cancer epidemiology is to prevent the disease and tailor lung cancer management to the specific needs of each community, ultimately improving patient outcomes. Using a population-based cancer registry, this paper aims to assess the profiles of lung cancer patients, as well as the incidence and trends of lung cancer in Ethiopia from 2012 to 2023. The main results revealed that young patients and patients with advanced-stage disease at the time of diagnosis were common. Notably, 31.2% of cases in the registry lacked histological confirmation. Adenocarcinoma was the most prevalent histological subtype, particularly among women. Notably, nearly half of the lung cancer cases were classified as carcinoma, which was not otherwise specified. The overall incidence of lung cancer remains relatively low, with minimal differences in risk between males and females. However, the incidence significantly increases with age and varies across residential areas. From 2012 to 2023, the trends in the ASIR remained stable, whereas females presented a significant increase in the CIR.
Lung cancer is common in younger individuals in our setting, as our crude median age at diagnosis was 56 years, with one-fourth of patients diagnosed before the age of 45. This finding aligns with a previous local study report6. After adjustments, the median age increased to 60; however, this remains at the lower end compared with findings from a study conducted across 65 countries, including Kenya, Algeria, South Africa, Zimbabwe, and Seychelles, which reported a crude median age ranging from 60 to 74 years, with an adjusted median age of 61–73 years19. Our proportion of patients aged under 45 years was also significantly higher than those reported from Uganda (16%) 24, South Africa (5%)25, Lebanon (< 11%) 26, China (5%)27, and Ireland (8%) 28. Lung cancer presents with different clinical features in younger individuals25,28–32. The relatively young age at presentation suggests that the disease may exhibit a more aggressive nature in our context or that unique risk factors may be involved. Therefore, further research in this area is essential. This also underscores the need for heightened awareness and potentially earlier screening strategies for younger populations. Additionally, the conventional view of lung cancer as affecting primarily elderly individuals may need re-evaluation in populations with similar risk profiles. However, this finding should be interpreted cautiously, as Ethiopia does not yet have a well-developed vital registration system, and age information is often self-reported or abstracted from medical records, which may introduce inaccuracies.
Despite one-third of cases lacking histological confirmation, which reflects ongoing diagnostic challenges, adenocarcinoma is the most commonly diagnosed histological subtype, followed by squamous cell carcinoma. Demographically, adenocarcinoma is more prevalent among females. This finding aligns with those of the studies by Gebremariam et al.. 6 and Global Trends33. The global rise of adenocarcinoma is linked to factors such as changes in smoking habits, including alterations in cigarette design and composition, as well as increased exposure to environmental pollutants33. However, unlike the recent increasing trend of adenocarcinoma globally33, adenocarcinoma remains a persistently dominant subtype in our context. The observed predominance and its association with females may be attributed to the low prevalence of smoking (3.4%), particularly among women (1.2%)7. Lung adenocarcinoma is more prevalent in women and nonsmokers34. Consistent with this, in Ethiopia, a previous study reported that only 25% of patients have a history of smoking6. Additionally, several studies35–37 have established a link between air pollution and adenocarcinoma, placing women at heightened risk due to indoor air pollution. Approximately 95% of our population relies on polluting fuels for cooking, a task predominantly performed by women8. This finding suggests that our prevention strategies should extend beyond smoking and pay equal attention to women. Furthermore, the notable prevalence of squamous cell carcinoma in our context may reflect the impact of tobacco, with its magnitude likely to increase owing to the increasing trend in smoking7.
Prognostic outcomes and treatment responses can differ significantly among histological subtypes34,38,39. Carcinoma NOS accounts for nearly half of our cases. Although there are few updated studies for comparison, this figure is higher than those reported in the Latin American cohort (12.8%) 40 and India (18.1%) 41. However, there was a significant annual decline of 8% from 2012 to 2023. A more pronounced reduction was observed in recent years, beginning in 2021, potentially driven by advancements in diagnostic technologies and more rigorous histopathological practices. The higher proportion of carcinoma NOS cases underscores the challenges faced by oncologists in treatment planning. Addressing these limitations is imperative to enable personalized treatment plans, optimize therapeutic decisions, and improve prognostic outcomes.
Late-stage diagnosis is common in our setting, with 93.2% of patients at an advanced stage. This highlights a critical gap in early detection efforts within our population. This finding aligns with those of a previous local study6 and studies from other developing countries25,42–44. These results point to common challenges in the early detection of the diseases. The high rate of late-stage diagnosis in our setting may be attributed to several factors. First, the asymptomatic nature of early disease stages complicates early detection, particularly given the low public awareness of early symptoms. Second, there is currently no organized screening program. Third, poor referral systems and the presence of concomitant diseases, such as tuberculosis, can obscure the final diagnosis10. Additionally, most of our patients are nonsmokers6, and a younger age at presentation is also common, potentially lowering the level of clinical suspicion among physicians. Consequently, further research should prioritize identifying systemic, social, and individual factors that lead to delayed diagnoses. This finding underscores the urgent need for targeted screening initiatives, increased public awareness of early symptoms, and improved diagnostic capacity. Implementing these measures could shift the diagnostic timeline toward earlier stages.
Our ASIR is relatively low, at 3.3 and 2.8 per 100,000 for males and females, respectively. This rate is comparable to those of previous local studies14,45 and the GLOBOCAN report from 20224. However, the ASIR varies significantly across regions, ranging from 51.4 in Eastern Asia to 2.5 in Western Africa for males and from 30.4 in North America to 1.5 in Western Africa for females. In African regions, the ASIRs are southern Africa (25.6 for males and 10.4 for females) and western Africa (2.5 for males and 1.5 for females)4. The lower smoking prevalence (3.4%) 7 and potentially distinct environmental or genetic influences, as well as lower life expectancy, may explain this lower incidence. On the other hand, this may not accurately reflect the true incidence. Developing countries with limited healthcare resources often report lower incidence rates due to underdiagnosis or misdiagnosis46. Additionally, the absence of lung cancer screening services may have contributed to this. In China, following the introduction of lung cancer screening, the incidence significantly increased in females47.
There is a minimal difference in the risk of lung cancer between males and females. The male-to-female crude and age-standardized IRRs were 1.241 and 1.128, respectively, with a nonsignificant declining trend from 2012 to 2023. This suggests that while males still generally have a higher incidence rate of lung cancer, the IRRs indicate that the difference is narrow. Globally, in regions such as North America and Northern Europe, the IRRs approach unity, whereas in North Africa and Eastern Europe, they can be as high as four- to fivefold4. Historically, the elevated male-to-female IRRs in many Western countries have been attributed to significantly higher smoking prevalence among males. However, this disparity has recently narrowed due to increasing smoking rates among females4,33. In contrast, countries with traditionally low smoking prevalence for both sexes, such as Ethiopia, consistently have relatively lower IRRs7. This makes it difficult to attribute the observed narrow IRRs in our setting to changes in smoking patterns alone. These variations imply that, as previously mentioned, factors other than smoking might be important in determining the incidence of lung cancer, especially in females in our setting.
In agreement with other studies26,48,49, IRRs significantly increase with advancing age, indicating that aging remains a critical determinant of lung cancer risk. However, the peak age groups for these rates vary across countries. Comparisons are limited due to the lack of local studies documenting age-specific incidence rates. A study by Tamási et al.. conducted in Hungary reported peak age groups of 70–79 for males and 60–69 for females49, whereas a study from Lebanon identified peak age groups of 70–74 for males and 75 + for females26. However, a study conducted in China by Zheng et al.. revealed that the 80–84-year-old age group had the highest incidence rates for both males and females48. In our study, the peak age groups for males and females were 75–79 and 70–74, respectively. The observed differences may be attributed to countries with higher life expectancies often reporting later peaks. As more individuals survive into older age groups, they increase their cumulative exposure to risk factors. In addition, the risk factors themselves can influence the timing of peak incidence. This emphasizes how crucial it is to consider age-specific factors when developing prevention and early detection plans.
There is considerable variation in crude lung cancer incidence rates across sub-cities, ranging from 1.526 in Arada to 2.877 in Bole per 100,000 persons. Further analysis revealed significant increases in the IRRs for Yeka, Lideta, Kirkos, Addis Ketema, and Bole. Although we could not find local studies for comparison, the observed variation across sub-cities aligns with findings from similar studies in other regions50–56, where incidence rates can vary significantly on the basis of geographical, socioeconomic, and environmental factors. According to a study by Sung et al.., higher exposure to risk factors such as air pollution, smoking, and occupational hazards causes some districts or neighborhoods to have disproportionately high incidence rates55. These findings suggest that specific sub-cities may have unique risk profiles or environmental exposures that warrant further investigation.
While no significant change was detected in the trend of the ASIR, the analysis of sex-specific CIR trends revealed notable differences. The incidence rate among males has remained relatively stable over the years. In contrast, females experienced a consistent 3% annual increase in the CIR, indicating a concerning rise in the incidence of lung cancer among females. This finding aligns with a local study suggesting that tobacco-related cancers have increased among female13, as well as with global trends57–59. However, while global trends are often attributed to increasing smoking prevalence among females57–59, it is unlikely that smoking alone accounts for the trend in our context, given the extremely low smoking prevalence among females, as previously noted7. Furthermore, although the period from 2012 to 2023 represents the most extended study duration since the registry’s establishment in 2012, this timeframe may still be insufficient for identifying long-term trends. Therefore, caution is warranted when interpreting these results, and future research should aim to update the trend analysis.
This study has several limitations. As a cancer registry-based study, vital variables—such as detailed patient information, risk factors, symptoms, metastasis sites, performance scores, cancer stage, survival data, and smoking status—were incomplete or unavailable, limiting the scope of some analyses. Eleven rare non-carcinoma lung tumors were also excluded, which may affect the comprehensiveness of the morphology data. Diagnostic challenges, including limited specialized services, reliance on ultrasound or clinical investigation for some cases, and the high burden of tuberculosis and other respiratory diseases, may have contributed to underdiagnosis or misclassification. Although the registry used the standardized African Cancer Registry Network notification form developed by IARC, variations in diagnostic bases may affect comparability. Furthermore, as the only cancer registry in the country, covering a population of around 6 million, the findings may have limited generalizability to the entire Ethiopian population of over 120 million. Finally, although the study period from 2012 to 2023 represents the most extended duration since the registry’s inception, it may still be considered relatively short for assessing long-term trends.
This study contributes to research, policy, and practice by addressing a critical evidence gap on lung cancer in LMICs. It can serve as a baseline for future studies that incorporate detailed patient characteristics and investigate systemic, social, and individual factors that delay diagnosis. To achieve the goals of SDG 3, target 3.4 and the Ethiopian National Cancer Control15, which emphasize prevention, early detection, and treatment, cancer registries must be expanded to other regions to ensure comprehensive national representation. Improving diagnostic capacity—through strengthened pathology services, expanded access to biopsy and imaging, and standardized protocols—is essential to reduce late-stage presentation, increase histological confirmation, and enable effective treatment planning. Policies should also promote cost-effective, context-specific lung cancer awareness and screening, highlighting risk factors beyond smoking and prioritizing younger individuals, females, and nonsmokers. Clinicians should consider age- and sex-specific risk factors when assessing lung cancer risk.
The ultimate goal of understanding lung cancer epidemiology is to prevent the disease and tailor lung cancer management to the specific needs of each community, ultimately improving patient outcomes. Using a population-based cancer registry, this paper aims to assess the profiles of lung cancer patients, as well as the incidence and trends of lung cancer in Ethiopia from 2012 to 2023. The main results revealed that young patients and patients with advanced-stage disease at the time of diagnosis were common. Notably, 31.2% of cases in the registry lacked histological confirmation. Adenocarcinoma was the most prevalent histological subtype, particularly among women. Notably, nearly half of the lung cancer cases were classified as carcinoma, which was not otherwise specified. The overall incidence of lung cancer remains relatively low, with minimal differences in risk between males and females. However, the incidence significantly increases with age and varies across residential areas. From 2012 to 2023, the trends in the ASIR remained stable, whereas females presented a significant increase in the CIR.
Lung cancer is common in younger individuals in our setting, as our crude median age at diagnosis was 56 years, with one-fourth of patients diagnosed before the age of 45. This finding aligns with a previous local study report6. After adjustments, the median age increased to 60; however, this remains at the lower end compared with findings from a study conducted across 65 countries, including Kenya, Algeria, South Africa, Zimbabwe, and Seychelles, which reported a crude median age ranging from 60 to 74 years, with an adjusted median age of 61–73 years19. Our proportion of patients aged under 45 years was also significantly higher than those reported from Uganda (16%) 24, South Africa (5%)25, Lebanon (< 11%) 26, China (5%)27, and Ireland (8%) 28. Lung cancer presents with different clinical features in younger individuals25,28–32. The relatively young age at presentation suggests that the disease may exhibit a more aggressive nature in our context or that unique risk factors may be involved. Therefore, further research in this area is essential. This also underscores the need for heightened awareness and potentially earlier screening strategies for younger populations. Additionally, the conventional view of lung cancer as affecting primarily elderly individuals may need re-evaluation in populations with similar risk profiles. However, this finding should be interpreted cautiously, as Ethiopia does not yet have a well-developed vital registration system, and age information is often self-reported or abstracted from medical records, which may introduce inaccuracies.
Despite one-third of cases lacking histological confirmation, which reflects ongoing diagnostic challenges, adenocarcinoma is the most commonly diagnosed histological subtype, followed by squamous cell carcinoma. Demographically, adenocarcinoma is more prevalent among females. This finding aligns with those of the studies by Gebremariam et al.. 6 and Global Trends33. The global rise of adenocarcinoma is linked to factors such as changes in smoking habits, including alterations in cigarette design and composition, as well as increased exposure to environmental pollutants33. However, unlike the recent increasing trend of adenocarcinoma globally33, adenocarcinoma remains a persistently dominant subtype in our context. The observed predominance and its association with females may be attributed to the low prevalence of smoking (3.4%), particularly among women (1.2%)7. Lung adenocarcinoma is more prevalent in women and nonsmokers34. Consistent with this, in Ethiopia, a previous study reported that only 25% of patients have a history of smoking6. Additionally, several studies35–37 have established a link between air pollution and adenocarcinoma, placing women at heightened risk due to indoor air pollution. Approximately 95% of our population relies on polluting fuels for cooking, a task predominantly performed by women8. This finding suggests that our prevention strategies should extend beyond smoking and pay equal attention to women. Furthermore, the notable prevalence of squamous cell carcinoma in our context may reflect the impact of tobacco, with its magnitude likely to increase owing to the increasing trend in smoking7.
Prognostic outcomes and treatment responses can differ significantly among histological subtypes34,38,39. Carcinoma NOS accounts for nearly half of our cases. Although there are few updated studies for comparison, this figure is higher than those reported in the Latin American cohort (12.8%) 40 and India (18.1%) 41. However, there was a significant annual decline of 8% from 2012 to 2023. A more pronounced reduction was observed in recent years, beginning in 2021, potentially driven by advancements in diagnostic technologies and more rigorous histopathological practices. The higher proportion of carcinoma NOS cases underscores the challenges faced by oncologists in treatment planning. Addressing these limitations is imperative to enable personalized treatment plans, optimize therapeutic decisions, and improve prognostic outcomes.
Late-stage diagnosis is common in our setting, with 93.2% of patients at an advanced stage. This highlights a critical gap in early detection efforts within our population. This finding aligns with those of a previous local study6 and studies from other developing countries25,42–44. These results point to common challenges in the early detection of the diseases. The high rate of late-stage diagnosis in our setting may be attributed to several factors. First, the asymptomatic nature of early disease stages complicates early detection, particularly given the low public awareness of early symptoms. Second, there is currently no organized screening program. Third, poor referral systems and the presence of concomitant diseases, such as tuberculosis, can obscure the final diagnosis10. Additionally, most of our patients are nonsmokers6, and a younger age at presentation is also common, potentially lowering the level of clinical suspicion among physicians. Consequently, further research should prioritize identifying systemic, social, and individual factors that lead to delayed diagnoses. This finding underscores the urgent need for targeted screening initiatives, increased public awareness of early symptoms, and improved diagnostic capacity. Implementing these measures could shift the diagnostic timeline toward earlier stages.
Our ASIR is relatively low, at 3.3 and 2.8 per 100,000 for males and females, respectively. This rate is comparable to those of previous local studies14,45 and the GLOBOCAN report from 20224. However, the ASIR varies significantly across regions, ranging from 51.4 in Eastern Asia to 2.5 in Western Africa for males and from 30.4 in North America to 1.5 in Western Africa for females. In African regions, the ASIRs are southern Africa (25.6 for males and 10.4 for females) and western Africa (2.5 for males and 1.5 for females)4. The lower smoking prevalence (3.4%) 7 and potentially distinct environmental or genetic influences, as well as lower life expectancy, may explain this lower incidence. On the other hand, this may not accurately reflect the true incidence. Developing countries with limited healthcare resources often report lower incidence rates due to underdiagnosis or misdiagnosis46. Additionally, the absence of lung cancer screening services may have contributed to this. In China, following the introduction of lung cancer screening, the incidence significantly increased in females47.
There is a minimal difference in the risk of lung cancer between males and females. The male-to-female crude and age-standardized IRRs were 1.241 and 1.128, respectively, with a nonsignificant declining trend from 2012 to 2023. This suggests that while males still generally have a higher incidence rate of lung cancer, the IRRs indicate that the difference is narrow. Globally, in regions such as North America and Northern Europe, the IRRs approach unity, whereas in North Africa and Eastern Europe, they can be as high as four- to fivefold4. Historically, the elevated male-to-female IRRs in many Western countries have been attributed to significantly higher smoking prevalence among males. However, this disparity has recently narrowed due to increasing smoking rates among females4,33. In contrast, countries with traditionally low smoking prevalence for both sexes, such as Ethiopia, consistently have relatively lower IRRs7. This makes it difficult to attribute the observed narrow IRRs in our setting to changes in smoking patterns alone. These variations imply that, as previously mentioned, factors other than smoking might be important in determining the incidence of lung cancer, especially in females in our setting.
In agreement with other studies26,48,49, IRRs significantly increase with advancing age, indicating that aging remains a critical determinant of lung cancer risk. However, the peak age groups for these rates vary across countries. Comparisons are limited due to the lack of local studies documenting age-specific incidence rates. A study by Tamási et al.. conducted in Hungary reported peak age groups of 70–79 for males and 60–69 for females49, whereas a study from Lebanon identified peak age groups of 70–74 for males and 75 + for females26. However, a study conducted in China by Zheng et al.. revealed that the 80–84-year-old age group had the highest incidence rates for both males and females48. In our study, the peak age groups for males and females were 75–79 and 70–74, respectively. The observed differences may be attributed to countries with higher life expectancies often reporting later peaks. As more individuals survive into older age groups, they increase their cumulative exposure to risk factors. In addition, the risk factors themselves can influence the timing of peak incidence. This emphasizes how crucial it is to consider age-specific factors when developing prevention and early detection plans.
There is considerable variation in crude lung cancer incidence rates across sub-cities, ranging from 1.526 in Arada to 2.877 in Bole per 100,000 persons. Further analysis revealed significant increases in the IRRs for Yeka, Lideta, Kirkos, Addis Ketema, and Bole. Although we could not find local studies for comparison, the observed variation across sub-cities aligns with findings from similar studies in other regions50–56, where incidence rates can vary significantly on the basis of geographical, socioeconomic, and environmental factors. According to a study by Sung et al.., higher exposure to risk factors such as air pollution, smoking, and occupational hazards causes some districts or neighborhoods to have disproportionately high incidence rates55. These findings suggest that specific sub-cities may have unique risk profiles or environmental exposures that warrant further investigation.
While no significant change was detected in the trend of the ASIR, the analysis of sex-specific CIR trends revealed notable differences. The incidence rate among males has remained relatively stable over the years. In contrast, females experienced a consistent 3% annual increase in the CIR, indicating a concerning rise in the incidence of lung cancer among females. This finding aligns with a local study suggesting that tobacco-related cancers have increased among female13, as well as with global trends57–59. However, while global trends are often attributed to increasing smoking prevalence among females57–59, it is unlikely that smoking alone accounts for the trend in our context, given the extremely low smoking prevalence among females, as previously noted7. Furthermore, although the period from 2012 to 2023 represents the most extended study duration since the registry’s establishment in 2012, this timeframe may still be insufficient for identifying long-term trends. Therefore, caution is warranted when interpreting these results, and future research should aim to update the trend analysis.
This study has several limitations. As a cancer registry-based study, vital variables—such as detailed patient information, risk factors, symptoms, metastasis sites, performance scores, cancer stage, survival data, and smoking status—were incomplete or unavailable, limiting the scope of some analyses. Eleven rare non-carcinoma lung tumors were also excluded, which may affect the comprehensiveness of the morphology data. Diagnostic challenges, including limited specialized services, reliance on ultrasound or clinical investigation for some cases, and the high burden of tuberculosis and other respiratory diseases, may have contributed to underdiagnosis or misclassification. Although the registry used the standardized African Cancer Registry Network notification form developed by IARC, variations in diagnostic bases may affect comparability. Furthermore, as the only cancer registry in the country, covering a population of around 6 million, the findings may have limited generalizability to the entire Ethiopian population of over 120 million. Finally, although the study period from 2012 to 2023 represents the most extended duration since the registry’s inception, it may still be considered relatively short for assessing long-term trends.
This study contributes to research, policy, and practice by addressing a critical evidence gap on lung cancer in LMICs. It can serve as a baseline for future studies that incorporate detailed patient characteristics and investigate systemic, social, and individual factors that delay diagnosis. To achieve the goals of SDG 3, target 3.4 and the Ethiopian National Cancer Control15, which emphasize prevention, early detection, and treatment, cancer registries must be expanded to other regions to ensure comprehensive national representation. Improving diagnostic capacity—through strengthened pathology services, expanded access to biopsy and imaging, and standardized protocols—is essential to reduce late-stage presentation, increase histological confirmation, and enable effective treatment planning. Policies should also promote cost-effective, context-specific lung cancer awareness and screening, highlighting risk factors beyond smoking and prioritizing younger individuals, females, and nonsmokers. Clinicians should consider age- and sex-specific risk factors when assessing lung cancer risk.
Conclusion
Conclusion
The incidence of lung cancer remains relatively low, with a narrow difference between males and females. The ASIR trend is stable, whereas the CIR in females increased from 2012 to 2023. This study emphasized the importance of developing context-specific strategies and guidelines for the prevention, control, detection, diagnosis, and treatment of lung cancer, while appropriately giving attention to younger patients, females, and nonsmokers.
The incidence of lung cancer remains relatively low, with a narrow difference between males and females. The ASIR trend is stable, whereas the CIR in females increased from 2012 to 2023. This study emphasized the importance of developing context-specific strategies and guidelines for the prevention, control, detection, diagnosis, and treatment of lung cancer, while appropriately giving attention to younger patients, females, and nonsmokers.
Supplementary Information
Supplementary Information
Below is the link to the electronic supplementary material.
Below is the link to the electronic supplementary material.
출처: PubMed Central (JATS). 라이선스는 원 publisher 정책을 따릅니다 — 인용 시 원문을 표기해 주세요.
🏷️ 같은 키워드 · 무료전문 — 이 논문 MeSH/keyword 기반
- Prostate Cancer Care for Men with an Intellectual Disability: A Population-based Cohort Study of Symptoms, Diagnosis, Treatment, and Survival.
- Clinical and Liquid Biomarkers of 20-Year Prostate Cancer Risk in Men Aged 45 to 70 Years.
- Association between polygenic risk scores and cardiovascular events in prostate cancer patients receiving androgen deprivation therapy in Han Chinese.
- Diagnostic accuracy of Ga-PSMA PET/CT versus multiparametric MRI for preoperative pelvic invasion in the patients with prostate cancer.
- Comprehensive analysis of androgen receptor splice variant target gene expression in prostate cancer.
- Clinical Presentation and Outcomes of Patients Undergoing Surgery for Thyroid Cancer.