Impact of referral source on diagnosis and treatment of oral cavity and oropharyngeal cancer in Brazil: a nationwide hospital cancer registry-based study Authors: Matheus de Abreu, Yuri de Lima Medeiros, Carolina Frandsen Pereira da Costa, Fábio Abreu Alves, Luiz Paulo Kowalski, Maria Paula Curado Published: 2026 ABSTRACT AbstractIntroductionIn Brazil, nearly 75% of the population depends on the public health system (PHS), yet disparities with private health insurance (HIS) persist in oncology care. This study aims to evaluate the impact of referral source on the diagnosis, time to treatment and post-treatment outcomes among Brazilian patients with oral cavity (OCC) and oropharyngeal (OPC) cancer.MethodsData from Integrador Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA) (2010–2022) were analysed. Sociodemographic and clinical variables, time from diagnosis to treatment and outcomes were assessed. Logistic regression estimated ORs for cancer care stages. Median diagnosis to treatment intervals were calculated by referral source and Brazilian state.ResultsA total of 62 178 cases were registered, including 33 485 OCC and 28 693 OPC. Referral information was available for 41 249 patients, of whom 87.9% were referred via PHS and 12.1% via HIS. PHS referrals were more frequent among males (79.3%), individuals <50 years (18.4%) and non-White patients (54.8%). Advanced stage (clinical stage, CS IV) was observed in 60.7% of PHS referrals versus 57.4% of HIS. Patients with CS IV had a 22% higher chance of being referred through the PHS. Those who waited more than 60 days for treatment had a 32% higher chance of being referred via PHS. The associated outcomes from PHS referred patients were progressive/relapsed disease (OR 1.28 (95% CI 1.08 to 1.52)) and death post-treatment (OR 1.36 (95% CI 1.15 to 1.61)).ConclusionsReferral through the PHS is associated with advanced clinical staging, delays in oral cancer treatment initiation and worst outcomes in OCC and OPC, underscoring persistent inequities in Brazilian cancer care. WHAT IS ALREADY KNOWN ON THIS TOPIC Oral cavity and oropharyngeal cancers in Brazil are frequently diagnosed at advanced stages, with long delays before treatment initiation. Socioeconomic, geographic and racial disparities affect access to timely oncologic care. However, the role of referral source (public vs private health system) in shaping diagnosis, treatment timelines and outcomes had not been systematically assessed. Oral cavity and oropharyngeal cancers in Brazil are frequently diagnosed at advanced stages, with long delays before treatment initiation. Socioeconomic, geographic and racial disparities affect access to timely oncologic care. However, the role of referral source (public vs private health system) in shaping diagnosis, treatment timelines and outcomes had not been systematically assessed. WHAT THIS STUDY ADDS This study shows that patients referred through the public health system are more likely to be diagnosed at advanced stages, experience delays longer than 60 days before starting treatment and present worse post-treatment outcomes, including higher risk of disease progression and death, compared with those referred via private health insurance. It also identifies regional and racial inequities in access to care. This study shows that patients referred through the public health system are more likely to be diagnosed at advanced stages, experience delays longer than 60 days before starting treatment and present worse post-treatment outcomes, including higher risk of disease progression and death, compared with those referred via private health insurance. It also identifies regional and racial inequities in access to care. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY The findings highlight persistent inequalities in Brazilian cancer care, demonstrating that referral source directly impacts clinical stage, treatment delays and survival outcomes. These results can inform health policy reforms aimed at strengthening referral pathways, reducing delays in treatment initiation and addressing regional and racial disparities to improve equity in oral cancer care. The findings highlight persistent inequalities in Brazilian cancer care, demonstrating that referral source directly impacts clinical stage, treatment delays and survival outcomes. These results can inform health policy reforms aimed at strengthening referral pathways, reducing delays in treatment initiation and addressing regional and racial disparities to improve equity in oral cancer care. Introduction The Global Cancer Observatory estimated 389 496 cases of oral cavity cancer (OCC) and 106 400 cases of oropharyngeal cancer (OPC) in 2022.1 Among Brazilian patients, 86.6% of those with OPC and 75.1% with OCC were diagnosed at advanced stages (III and IV).2 In Latin America, barriers such as inefficient national cancer control strategies, unclear referral systems for users, a shortage of trained professionals and limitations in technologies are barriers to the early diagnosis of oral cancer.3 Substantial health inequities persist globally, driven by factors such as socioeconomic status, gender, ethnicity and geographic location. In cancer treatment, in particular, age plays an important role in these disparities.4 Although Brazil has a complex and dynamic public health system (PHS) that ensures the right to universal health coverage, challenges persist in equitable access, distribution of healthcare professionals and resources and the effectiveness of the services provided.5 The PHS includes a referral and counter-referral system, a mechanism that supports the flow and coordination of users across different levels of care.6 However, due to the fragmentation of the healthcare system, lack of infrastructure and challenges faced by users, such as the lack of ease, knowledge and understanding of the processes—continuity of care remains compromised, leading to delays in the referral and treatment of malignant neoplasms.7 A scoping review that includes studies in countries where most essential medical services are provided free at the point of access identified variations in access to cancer referral and care.4 Individuals from more privileged social and economic groups were found to have greater awareness of their options, to be more assertive and better prepared to negotiate healthcare.8 In addition to regional and socioeconomic disparities, the type of access to the healthcare system—particularly whether referral occurs through the public or private sector—may substantially influence the care trajectory of cancer patients. Individuals with oral cancer initially managed within the private healthcare sector tend to experience shorter time intervals to treatment initiation and more favourable clinical outcomes, reflecting inequalities in access to specialised services, diagnostic procedures and timely therapeutic interventions.9 Factors such as treatment complexity, the need for a multidisciplinary care team and the availability of adequate hospital infrastructure further underscore the role of the referral source in determining the quality of oncologic care.10 To our knowledge, no previous study has specifically investigated how the source of referral influences treatment timelines and outcomes in patients with oral and OPC in Brazil. This study aims to evaluate the impact of referral source on the time from diagnosis to treatment and patients’ outcomes among Brazilian patients with OCC and OPC. Methods IntegradorRHC A secondary data analysis was conducted using records from the IntegradorRHC, a national web-based system that compiles patient data from Hospital Cancer Registries (RHC) across Brazil. These data are submitted from hospitals to the national Cancer Hospital Registry database, managed by Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA) through the IntegradorRHC. On the system, cases are defined based on International Classification of Diseases for Oncology, Third Edition (ICD-O-3). Data access, sample preprocessing and selection The data were extracted from the IntegradorRHC website (https://irhc.inca.gov.br/RHCNet/). The databases from 2000 to 2022 were selected and an automated approach was employed to read and merge multiple .dbf files (online supplemental file 1). The final dataset used for the analyses is available in online supplemental file 2. Analytic cases (ie, cases in which the patient received their first treatment at the reporting hospital) with histologically confirmed squamous cell carcinoma (ICD-O-3 morphology code 8070/3) of the oral cavity or oropharynx, diagnosed between 2010 and 2022, were included. Case selection was based on the ICD-O-3 topography codes for oral cavity (C00.3, C00.4, C00.5, C00.8, C00.9, C02.0, C02.1, C02.2, C02.3, C02.4, C02.8, C02.9, C03.0, C03.1, C03.9, C04.0, C04.1, C04.8, C04.9, C05.0, C05.8, C05.9, C06.0, C06.1, C06.2, C06.8, C06.9) and oropharynx (C01.9, C05.1, C05.2, C09.0, C09.1, C09.8, C09.9, C10.0, C10.1, C10.2, C10.3, C10.8, C10.9). Patients with missing referral source information were excluded from the analyses. The sociodemographic and clinical characteristics of the full cohort, as well as the amount of missing data, calculated as the difference between the total number of records and those with available referral source information, are presented in online supplemental table 1. Study variables The variables selected for the study were based on their established associations with cancer care disparities and their availability in the IntegradorRHC database. Sociodemographic variables included: sex (male or female) and age at diagnosis (years), given their known influence on cancer risk profiles and healthcare utilisation patterns; ethnicity (white, Asian, brown, black), self-reported by participants according to the Brazilian Institute of Geography and Statistics classification11; education level (none, incomplete elementary, complete elementary, high school, incomplete higher education, complete higher education), as proxies for socioeconomic vulnerability and inequity in access to healthcare; and place of residence (Brazilian state), to capture regional disparities in oncologic care. Clinical variables included: clinical stage (TNM classification), a key determinant of treatment options and prognosis; referral source (PHS, health insurance (HIS), self-referred), the primary exposure of interest; treatment modality (none, surgery, radiotherapy, chemotherapy and other associations); and dates of first diagnosis and treatment initiation, used to calculate the diagnosis-to-treatment interval, an established indicator of healthcare timeliness. Statistical analysis A descriptive analysis of absolute and relative frequencies stratified by referral source was conducted. The χ2 test was applied with a significance level of 5% to assess the association between categorical sociodemographic and clinical variables and referral source (PHS vs HIS). The median number of days between diagnosis and treatment initiation was calculated for each Brazilian state plus the Federal District, stratified by referral source and clinical stage (initial—I or II and advanced—III or IV). A set of four maps representing Brazil was created based on the median diagnosis-to-treatment interval. A Sankey diagram was created to analyse patient flow for referral source by diagnosis-to-treatment interval and treatment modalities. Simple (online supplemental table 2) and multiple logistic regression analyses were performed to calculate ORs with a 95% CI. In all logistic regression models, referral source (PHS vs HIS) was the dependent (outcome) variable, coded as a binary variable (PHS=1; HIS=0, reference category), while sociodemographic and clinical variables were considered independent variables. The reference group was the HIS referral. The multiple regression model was built by including variables with a significance level of 20%. Three logistic regression models were developed to evaluate outcomes at distinct stages of the cancer journey. Variable selection was guided by both theoretical relevance and statistical considerations. Only variables available either before or at the time of each analytical stage were considered to maintain temporal validity and prevent model overfitting. Model selection was performed using a stepwise approach, testing p values from smallest to largest. The final model was established based on the following criteria: (1) No change in OR >10%; (2) Improved precision at 95% CI; (3) Total df allowed for the outcome. The Hosmer-Lemeshow test was applied to assess model fit. Statistical analyses were conducted using the R programming language within the RStudio environment. Patient and public involvement This study was based on anonymised secondary data from the IntegradorRHC database. No patients or members of the public were directly involved in the design, conduct or reporting of this research. Results cannot be disseminated directly to study participants, given the anonymous nature of the data. Results In Brazil, 41 249 cases of OCC and OPC with information on referral source were recorded from 2010 to 2022 across 260 hospital cancer registries: 130 (50.0%) were located in the Southeast region, 51 (19.6%) in the South, 47 (18.1%) in the Northeast, 18 (6.9%) in the Mid-west and 14 (5.4%) in the North (online supplemental table 3). Of these, 36 242 (87.9%) were referred through the PHS and 5007 (12.1%) through the HIS. Most males were referred through the PHS (n=28 736; 79.3%). Among females, n=1278 (14.5%) were referred through the HIS, compared with n=7506 (85.5%) via the PHS (p<0.001). Individuals under 50 years old were more frequently referred via the PHS (n=6678; 18.4%) compared with the HIS (n=800; 16.0%), whereas those over 69 years old had more referrals through HIS (n=1282; 25.6%) compared with the PHS (n=7251; 20.0%) (p<0.001). Among white patients, n=2314 (50.7%) were referred through the HIS, compared with n=14 213 (39.1%) via the PHS, while brown and black individuals had a higher proportion of referrals through the PHS (p<0.001). Patients who were married or in a consensual union more frequently received referrals via the HIS (n=2751; 57.4%) compared with singles or separated individuals (n=2041; 42.6%) (p<0.001). The Mid-west region had the highest proportion of referrals via the PHS (n=1335; 92.7%), while the North region had the highest proportion of HIS referrals (n=281; 14.6%) (p<0.001) (table 1). Table 1Sociodemographic and clinical characteristics of 41 249 cases of oral cavity and oropharyngeal cancer diagnosed in Brazil from 2010 to 2022 (IntegradorRHC — Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA))Public (n=36 242)Health insurance (n=5007)P valueSex<0.001 Male28 736 (79.3)3729 (74.4) Female7506 (20.7)1278 (25.6)Age at diagnosis (years)<0.001 <506678 (18.4)800 (16.0) 50–6922 313 (61.6)2925 (58.4) >697251 (20.0)1282 (25.6)Ethnicity<0.001 White14 213 (41,4)2314 (50.7) Asian211 (0.6)63 (1.4) Brown17 341 (50,5)1953 (42.8) Black2577 (7.5)239 (5.2)Education (years)<0.001 0–925 949 (86.4)2664 (65.6) 10–123361 (11.2)892 (22.0) >12731 (2.4)506 (12.5)Marital status<0.001 Single/widowed/separated17 963 (52.7)2041 (42.6) Married/consensual union16 143 (47.3)2751 (57.4)Region of residence<0.001 South8703 (24.1)1171 (23.6) Southeast13 598 (37.7)1861 (37.5) Mid-West1335 (3.7)105 (2.1) Northeast10 826 (30.0)1543 (31.1) North1639 (4.5)281 (5.7)Alcohol consumption<0.001 Non drinker6686 (23.1)1058 (29.0) Former drinker9460 (32.8)1147 (31.4) Current drinker12 744 (44.1)1443 (39.6)Smoking<0.001 Non smoker4301 (14.1)839 (21.3) Former smoker8272 (27.2)1311 (33.2) Current smoker17 842 (58.7)1794 (45.5)Family history of cancer<0.001 Yes9946 (46.1)1352 (50.9) No11 618 (53.9)1302 (49.1)Clinical stage<0.001 I1711 (6.7)301 (8.5) II2919 (11.4)436 (12.4) III5417 (21.2)762 (21.6) IV15 543 (60.7)2024 (57.4)Treatment<0.001 Surgery6861 (19.5)1137 (23.4) Surgery plus7060 (20.1)976 (20.1) Radiotherapy5016 (14.3)876 (18.1) Chemotherapy3512 (10.0)264 (5.4) Radiotherapy+chemotherapy10 454 (29.8)1391 (28.7) No treatment2235 (6.4)205 (4.2)Diagnosis-to-treatment interval (days)<0.001 0–306784 (21.1)1263 (29.1) 31–607321 (22.8)1103 (25.4) >6017 995 (56.1)1975 (45.5)Disease status at the end of treatment<0.001 Complete remission5783 (24.0)929 (30.8) Partial remission/stable disease8414 (34.9)1012 (33.6) Progressive disease/relapsed disease4558 (18.9)512 (17.0) Death5359 (22.2)561 (18.6) Sociodemographic and clinical characteristics of 41 249 cases of oral cavity and oropharyngeal cancer diagnosed in Brazil from 2010 to 2022 (IntegradorRHC — Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA)) Individuals referred via HIS were more frequently non-drinkers (n=1058; 29.0%) and non-smokers (n=839; 21.3%), whereas those referred via PHS were more often drinkers (n=12 744; 44.1%) and smokers (n=17 842; 58.7%) (p<0.001). Among PHS referrals, 60.7% were diagnosed at clinical stage (CS) IV, while 18.1% were at early CSs (I/II). In contrast, among HIS referrals, 57.4% were diagnosed at CS IV, and 20.9% at early stages (table 1). Treatments involving chemotherapy had the highest proportion of referrals via PHS (88.3%–93.0%), whereas surgery (n=1137; 14.2%) and radiotherapy (n=876; 14.9%) had higher frequencies of referrals via HIS table 1. The diagnosis-to-treatment interval was longer for patients referred via PHS, with n=17 995 (56.1%) experiencing delays of more than 60 days, compared with n=1975 (45.5%) among those referred through HIS (table 1). Figure 1 presents the multiple logistic regression models for each phase of the patients’ oncological care journey. For diagnosis, patients with CS IV were associated with a 22% higher likelihood of being referred via PHS (OR 1.22; 95% CI 1.01 to 1.47) compared with those at CS I (reference). Additionally, individuals with the lowest educational level (0–9 years) had an almost fourfold higher likelihood of PHS referral (OR 4.12; 95% CI 3.38 to 5.02) compared with those with more than 12 years of education. Married individuals were less likely to be referred via PHS (OR 0.68; 95% CI 0.62 to 0.76) compared with single individuals. Current smokers had a 1.4 times higher likelihood of PHS referral (OR 1.38; 95% CI 1.20 to 1.60) compared with non-smokers. Younger individuals (aged <50 years) were associated with a 33% higher likelihood of being referred via PHS (OR 1.33; 95% CI 1.12 to 1.57) compared with those aged over 69 years (figure 1; table 2). Impact of referral source from diagnosis to treatment outcome in 41 249 cases of oral cavity and oropharyngeal cancer diagnosed in Brazil from 2010 to 2022. (IntegradorRHC—Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA)). *A separate multiple logistic regression model was created for each phase (diagnosis, treatment and post-treatment). The Hosmer-Lemeshow test was used to assess model fit: diagnosis (p=0.875), treatment (p=0.882), post-treatment (p=0.891). Table 2Multiple logistic regression models for referral source (public health system vs health insurance) across three phases of the cancer care journeyDiagnosisTreatmentPost-treatmentOR (95% CI)Sex Female1.001.001.00 Male1.10 (0.97 to 1.25)1.24 (1.09 to 1.40)1.12 (0.96 to 1.30)Age at diagnosis (years) >691.001.001.00 <501.33 (1.12 to 1.57)1.23 (1.04 to 1.45)1.36 (1.13 to 1.63) 50–691.17 (1.02 to 1.33)1.12 (0.99 to 1.28)1.25 (1.07 to 1.45)Education (years) >121.00—1.00 0–94.12 (3.38 to 5.02)—4.65 (3.75 to 5.75) 10–122.12 (1.70 to 2.65)—2.14 (1.69 to 2.71)Marital status Single*1.001.001.00 Married*0.68 (0.62 to 0.76)0.67 (0.60 to 0.74)0.71 (0.63 to 0.79)Smoking status Non-smoker1.00—1.00 Former smoker0.92 (0.79 to 1.07)—0.90 (0.74 to 1.10) Current smoker1.38 (1.20 to 1.60)—1.40 (1.17 to 1.67)Family history of cancer No1.001.00— Yes0.86 (0.78 to 0.95)0.80 (0.73 to 0.89)—Ethnicity White—1.00— Asian—0.60 (0.38 to 0.99)— Brown—1.34 (1.21 to 1.49)— Black—1.27 (1.05 to 1.55)—Alcohol consumption Non-drinker——1.00 Former drinker——0.95 (0.79 to 1.14) Current drinker——0.90 (0.76 to 1.06)Clinical stage I1.001.001.00 II1.39 (1.10 to 1.76)1.36 (1.06 to 1.73)1.22 (0.94 to 1.59) III1.28 (1.03 to 1.57)1.36 (1.07 to 1.71)1.16 (0.89 to 1.49) IV1.22 (1.01 to 1.47)1.34 (1.07 to 1.67)1.04 (0.81 to 1.32)Diagnosis-to-treatment interval (days) 0–30—1.001.00 31–60—1.03 (0.89 to 1.19)1.19 (1.01 to 1.39) >60—1.32 (1.16 to 1.50)1.34 (1.16 to 1.54)Treatment modality Surgery—1.001.00 Surgery plus—0.86 (0.71 to 1.04)1.07 (0.87 to 1.30) Radiotherapy—0.72 (0.58 to 0.88)0.75 (0.60 to 0.93) Chemotherapy—1.40 (1.08 to 1.82)1.53 (1.15 to 2.05) Radio+chemotherapy—0.75 (0.62 to 0.90)0.84 (0.69 to 1.02)Disease status at end of treatment Complete remission——1.00 Partial remission/stable disease——1.57 (1.36 to 1.82) Progressive/relapsed disease——1.28 (1.08 to 1.52) Death——1.36 (1.15 to 1.61)*Single includes widowed and separated; Married includes consensual union. — =variable not included in this model. OR=1.00 indicates the reference category for each predictor variable.Reference category for the outcome: health insurance referral. Hosmer-Lemeshow test: diagnosis (p=0.875), treatment (p=0.882), post-treatment (p=0.891). Multiple logistic regression models for referral source (public health system vs health insurance) across three phases of the cancer care journey Single includes widowed and separated; Married includes consensual union. — =variable not included in this model. OR=1.00 indicates the reference category for each predictor variable.Reference category for the outcome: health insurance referral. Hosmer-Lemeshow test: diagnosis (p=0.875), treatment (p=0.882), post-treatment (p=0.891). Regarding treatment, patients who waited more than 60 days to start treatment had a 32% higher likelihood of being referred via PHS (OR 1.32; 95% CI 1.16 to 1.50), compared with those who received treatment within 30 days of diagnosis. Individuals who received radiotherapy (OR 0.72; 95% CI 0.58 to 0.88) or radiotherapy plus chemotherapy (OR 0.75; 95% CI 0.62 to 0.90) were 28% and 25% less likely, respectively, to be referred via PHS. In contrast, those who received chemotherapy alone had a 1.4 times higher likelihood of PHS referral (OR 1.40; 95% CI 1.08 to 1.82). Advanced CS (IV) remained associated with PHS referral (OR 1.34; 95% CI 1.07 to 1.67). Brown and black patients were associated with a 30% higher likelihood of being referred via PHS (figure 1; table 2). At post-treatment, patients referred via the PHS were more likely to present partial remission or stable disease (OR 1.57; 95% CI 1.36 to 1.82), as well as progressive or relapsed disease (OR 1.28; 95% CI 1.08 to 1.52), compared with those referred via HIS. Patients who died after treatment had a 36% higher likelihood of being referred via PHS (OR 1.36; 95% CI 1.15 to 1.61). Current smokers were also associated with PHS referral (OR 1.40; 95% CI 1.17 to 1.67). All three models (diagnosis; treatment; post-treatment) demonstrated adequate fit, as indicated by the Hosmer-Lemeshow goodness-of-fit test (diagnosis: p=0.875; treatment: p=0.882; post-treatment: p=0.891) (figure 1; table 2). Figure 2 illustrates the patient flow by referral source, CS, diagnosis-to-treatment interval and treatment modality. Among early-stage cases, a higher proportion of HIS-referred patients initiated treatment within 30 days compared with PHS-referred patients, particularly for surgery and surgery plus modalities. Among advanced-stage cases, the majority of patients in both referral groups waited more than 60 days to begin treatment, with little difference observed between referral sources. Overall, more patients referred through HIS received treatment within 30 days for surgery and surgery plus, while radio+chemotherapy was the most frequent modality among those waiting more than 60 days, especially among PHS referrals (figure 2). Sankey diagram. Referral source for oral cavity and oropharyngeal cancer by diagnosis-to-treatment interval and treatment modality in Brazil from 2010 to 2022. (IntegradorRHC—Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA)). Figure 3 presents the median diagnosis-to-treatment interval (in days) for OCC and OPC across Brazilian states, stratified by referral source and CS. Overall, the diagnosis-to-treatment interval was shorter among HIS referrals compared with PHS referrals for both early-stage (PHS: 101 days; HIS: 90 days) and advanced-stage cases (PHS: 90 days; HIS: 84 days). For early-stage OCC and OPC, the shortest diagnosis-to-treatment intervals were observed among HIS referrals in the states of Rio Grande do Sul (33 days), Ceará (36 days) and Tocantins/Paraná (37 days). In contrast, the longest delays were found in Northern Brazil, particularly in Roraima (~200 days), Acre (172 days) and Pará (141 days), all referred through public healthcare (figure 3) (online supplemental table 4). Median time of diagnosis-to-treatment interval by referral source in 41 249 cases of oral cavity and oropharyngeal cancer diagnosed at early and advanced clinical stage in Brazil from 2010 to 2022. (IntegradorRHC—Brazilian National Cancer Institute (Instituto Nacional de Câncer, INCA)). *Brazil is divided into five geographic regions: North (Acre, Amapá, Amazonas, Pará, Rondônia, Roraima and Tocantins), Northeast (Alagoas, Bahia, Ceará, Maranhão, Paraíba, Pernambuco, Piauí, Rio Grande do Norte and Sergipe), Mid-west (Distrito Federal, Goiás, Mato Grosso and Mato Grosso do Sul), Southeast (Espírito Santo, Minas Gerais, Rio de Janeiro and São Paulo) and South (Paraná, Rio Grande do Sul and Santa Catarina). A map illustrating the geographic location of each state is provided in online supplemental figure 1. SCC, squamous cell carcinoma. For advanced CS, Amapá had the longest median times to treatment in both HIS (156 days) and public health (126 days), Goiás had the second longest delay, with a median of about 120 days to treatment for OCC and OPC among patients referred through the public health system. The shortest diagnosis-to-treatment intervals were also observed among HIS referrals in advanced stage, Ceará had the shortest time (32 days), followed by Mato Grosso do Sul (34.5 days) and Paraná (45 days) (figure 3) (online supplemental table 4). Discussion Substantial inequalities in access to and quality of cancer care between patients referred via public and private institutions are observed in Brazil. We evaluated the impact of referral source on the time from diagnosis to treatment initiation, as well as on post-treatment outcomes among Brazilian patients with OCC and OPC. Our findings revealed that patients referred through the PHS are associated with advanced staging, delays in accessing treatment and worse post-treatment outcomes such as death and relapsed disease. Among the critical factors associated with delays in diagnosis and treatment are the shortage of public resources, the fragility of public–private collaboration and limited access to specialised care.12 In this Brazilian cohort, referral through the public system for treatment of OCC and OPC was identified as an aggravating factor, further limiting timely access to oncological care. Martínez-Ramírez et al2 demonstrated structural and institutional barriers to the early diagnosis and management of oral cancer in Latin America, highlighting the absence of specific oral cancer control plans, the low prioritisation of the disease in public health policies and the lack of clearly defined referral pathways from primary care to specialised centres. These disparities may also reflect barriers occurring before hospital admission, particularly between first symptom recognition and specialist referral. In Brazil, primary oral healthcare teams are expected to contribute to the detection and monitoring of suspicious oral mucosal lesions. Previous studies have suggested that delayed oral cancer diagnosis may be influenced by limited professional training, low knowledge with potentially malignant disorders and oral cancer, and difficulties in referral processes and timely access to specialised services.291314 Although these factors were not directly assessed in the present study, they are consistent with our finding that patients referred through the PHS were more frequently diagnosed at advanced stages and experienced longer intervals before treatment initiation. Brazil is divided into five regions with socioeconomic and demographic disparities that influence healthcare access (online supplemental figure 1).15 In our study, the Mid-West region had the highest proportion of referrals through the PHS (92.7%), while the North region showed the highest percentage of private referrals (14.6%). However, the longest delays in treatment initiation were observed in northern states such as Roraima, Acre and Pará, all referred through the public system, with median intervals exceeding 140 days between diagnosis and the start of therapy. This pattern aligns with the findings of Borges et al,16 who identified Roraima and Amapá as Brazilian states with delayed treatment of oral cancer. Such delays may be attributed to structural challenges, including the shortage of specialised oncology services and the need for long-distance travel, which contribute to the prolonged intervals observed.17 The higher frequency of oral OCC and OPC among middle-aged men in Brazil observed in this study aligns with the oral cancer epidemiological profile widely reported in the literature, reflecting men’s historically greater exposure to risk factors such as excessive alcohol and tobacco consumption.18 Although women tend to use public health services more frequently than men,19 Raymundo et al20 found that men with oral cancer were 40% more likely to be referred for oncologic treatment through the PHS than women. Our findings also showed that men were associated with referral through the PHS. Age also appears to influence health service use. In Brazil, a study based on the National Health Survey found that over 70% of elderly men do not have private HIS.21 Wray et al22 reported that, in the USA, younger individuals are more commonly covered by private HIS. However, for oral cancer in Brazil, we identified a greater association of young individuals referred for treatment of the neoplasm through the public system compared with elderly, possibly explained by socioeconomic differences, with younger Brazilians with lower income being more dependent on the public system. Racial disparities in referrals for the treatment of OCC and OPC were evident in our study. This pattern is consistent with previous findings identifying skin colour as an important marker of inequity in access to Brazilian healthcare services. Raymundo et al20 demonstrated that non-white individuals had a 34% higher likelihood of being referred through the PHS for oral cancer treatment. These findings align with our results, in which black and brown individuals are associated with being referred through the PHS compared with whites. Structural racism, socioeconomic disadvantages, and geographic barriers are factors that contribute to limited and delayed access to healthcare services for the black population.23 These inequities may partially explain the worse prognosis, higher risk of distant metastases, and more frequent diagnosis at advanced stages observed among non-white patients with oral cancer.2324 Andrade et al12 reported that 40.3% of patients with breast cancer treated in a public hospital were diagnosed at an advanced stage, compared with only 17.5% in the private sector. Although our study focused on OCC and OPC, we similarly found that patients diagnosed at stage IV were associated with being referred through the PHS. This may reflect the higher clinical complexity of cases managed by PHS, which often demand a more prolonged, multidisciplinary approach before treatment can begin.25 Nevertheless, even among early-stage cases, our findings show that privately referred patients began treatment more promptly, indicating that delays may not be solely linked to disease severity but also to structural disparities between public and private healthcare systems such as resource availability, inequities in healthcare infrastructure and patient navigation support. Moreover, our results indicated that therapeutic approaches involving chemotherapy had the highest frequency of referrals by the PHS. This may be due to the higher concentration of cases with more advanced stages in the public system, often managed with fewer curative therapeutic options. Delay in initiating cancer treatment for OCC and OPC in Brazil is a critical factor that can negatively impact prognosis and overall survival among patients with oral cancer.23 Although Costa et al26 observed that treatment delays occurred regardless of whether patients had access to private healthcare services, Chiliti et al9 demonstrated that patients treated within the PHS in São Paulo waited, on average, 63 days to begin treatment, compared with 36 days for those treated in the private sector. In our study, patients who waited more than 60 days to initiate treatment were more likely to be referred through the PHS. These discrepancies highlight the urgent need for policies that ensure timely cancer care, in accordance with Brazilian legislation, which mandates a maximum interval of 60 days between cancer diagnosis and the initiation of treatment.27 This study has limitations inherent to analyses based on secondary population-level data. The lack of detailed clinical variables (such as comorbidities, performance status and therapeutic intent) limits the ability to assess case complexity and the clinical rationale behind treatment decisions. Additionally, the categorisation of referral type does not capture specific distinctions within public and private systems, such as regional differences or variability among private insurance providers. Potential under-reporting or inconsistent completion of variables, common in administrative databases, may also introduce bias into the findings. Despite these limitations, this study offers valuable insights into referral and treatment pathways for oral and OPC in Brazil, highlighting delays and disparities in clinical staging and treatment between private and public oncologic healthcare. By identifying regional and racial inequalities, as well as the sociodemographic profiles of patients who predominantly depend on the PHS, these findings can guide targeted public policies aimed at improving timely access, ensuring that vulnerable Brazilian populations receive equitable oncologic care and adequate resources. Supplementary material