Perceived barriers to primary healthcare access in a violence-affected community in Rio de Janeiro, Brazil: sociodemographic characteristics and self-reported health Authors: Ana Paula da Cunha, Jurema Corrêa da Mota, Francisco Inacio Bastos Published: 2026 ABSTRACT AbstractBackgroundThis study analysed perceptions of barriers to primary healthcare (PHC) access in a socially vulnerable and violence-affected community in Rio de Janeiro.MethodsA cross-sectional survey was conducted in the Manguinhos community, with approximately 1000 residents interviewed face to face in 2024. Data were collected using a structured questionnaire via REDCap, including sociodemographic variables, self-reported health conditions and Patient Health Questionnaire-2 and Generalised Anxiety Disorder-2 scales. Adapted narrative vignettes were used to describe profiles of individuals with or without mental health problems and socially stigmatised behaviours, allowing the assessment of perceived difficulty in accessing health services. Results showed no significant differences in perceived difficulty of access between Black and non-Black participants. However, men reported greater perceived difficulty in scenarios involving common mental disorders, suggesting the influence of internalised stigma. Younger participants perceived greater barriers in situations involving socially marginalised behaviours, reflecting a lack of trust in the responsiveness of health services. Conversely, formal diagnoses and medication use for diabetes and hypertension were associated with lower perceived barriers, indicating that prior engagement with the health system fosters greater familiarity and lower perceived difficulty. Item response theory analysis demonstrated good psychometric performance of the vignettes, confirming their methodological utility.ConclusionsThe findings highlight the need to strengthen PHC to address mental health and social vulnerability demands and to reduce subjective barriers, particularly for men and younger individuals. They also underscore the importance of targeted strategies to promote equity in access to care in vulnerable settings.Keywords: Primary Prevention, Health Services Accessibility, Violence WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS This study shows that vignette-based instruments can be validated and applied using item response theory in a low-income Brazilian community. It demonstrates that such approaches are feasible even in populations with very low literacy levels. This study shows that vignette-based instruments can be validated and applied using item response theory in a low-income Brazilian community. It demonstrates that such approaches are feasible even in populations with very low literacy levels. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Vignette-based methods can capture nuanced perceptions of healthcare access that conventional surveys miss. Their use may strengthen the assessment of inequalities and support more context-sensitive health policies. Vignette-based methods can capture nuanced perceptions of healthcare access that conventional surveys miss. Their use may strengthen the assessment of inequalities and support more context-sensitive health policies. Background Primary healthcare (PHC) plays a crucial role in the organisation of health systems, promoting timely, equitable and continuous access to essential services.1 However, access to PHC may be hindered by multiple factors, including sociodemographic characteristics, health conditions and contexts of social vulnerability and urban violence.2 In Brazil, difficulties in accessing PHC are particularly significant in communities exposed to violence, where structural and subjective barriers hinder the utilisation of health services.3 Access to PHC in socially vulnerable urban settings remains a major challenge in many low- and middle-income countries (LMICs), where structural inequalities, stigma and fragmented health systems shape patterns of utilisation and engagement with services.4 5 In such contexts, barriers to access are not solely determined by the availability of services, but also by social, cultural and relational factors that influence how individuals perceive and navigate health systems.6 Understanding these perceived barriers is essential for informing policies aimed at strengthening PHC systems and promoting equity in access. However, evidence on how different social groups interpret and respond to these barriers remains limited, particularly in urban territories marked by violence, poverty and social exclusion. Access to healthcare is influenced by both structural and individual-level factors.7 8 Structural or system-level barriers include aspects such as service organisation, availability of resources and territorial inequalities that shape the distribution and functioning of health services.9 10 In contrast, individual-level or perceptual barriers refer to how individuals experience and interpret access, including perceptions of discrimination, stigma, trust in services and previous interactions with the health system.11 12 Although analytically distinct, these dimensions are interrelated, as individual perceptions are shaped by broader structural conditions.7 13 However, conventional approaches to assessing access tend to prioritise structural indicators, often overlooking how barriers are perceived and navigated by individuals, particularly in socially vulnerable settings. In this study, we focus on perceived barriers to access, while recognising their structural determinants. In this context, vignette-based approaches offer a valuable methodological strategy to capture perceived barriers using standardised linguistic tools, allowing comparison across individuals while preserving contextual sensitivity. Despite challenges in their adaptation across different linguistic and cultural settings, this study demonstrates that vignettes can be successfully adapted to a socially vulnerable urban community, using rigorous psychometric methods, including item response theory (IRT). This approach contributes to bridging narrative and quantitative perspectives in health research, particularly in contexts where language use may differ from formal standards.14 15 These dimensions of access have been widely conceptualised in the literature. The framework proposed by Levesque et al16 conceptualises access as the interaction between health system characteristics (approachability, acceptability, availability, affordability and appropriateness) and individuals’ abilities (ability to perceive, seek, reach, pay and engage). Similarly, Andersen,17 through the Behavioural Model of Health Services Use, highlights the role of predisposing, enabling and need factors in shaping health service utilisation. These frameworks reinforce that access is not only determined by the availability of services, but also by how individuals perceive, interpret and navigate health systems. In this study, perceived barriers captured through vignettes are understood as expressions of these broader mechanisms. To investigate perceptions and decision-making processes in different contexts, a widely adopted methodological strategy is the use of vignettes—brief narratives that describe hypothetical scenarios. Nevertheless, the application of this technique in Brazil has faced challenges and limitations, particularly in the field of public health. 18Although its use has advanced in specific areas, such as nursing19 and psychological research,20 its employment in public health remains scarce. Vignettes aim to convey information in a swift and accessible manner, suitable for audiences who prefer narrative (using a third, hypothetical individual, instead of the interviewee himself/herself)21 or audiovisual formats to understand relevant content.22 In contexts with limited engagement with written culture and/or stigmatisation of some behaviours in context, this approach proves particularly important, as well as among younger generations who have been socialised in a culture marked by audiovisual media. Such experiences appear to be a promising alternative in Brazil, as far as is currently known, based on preliminary assessments.23 In this context, understanding how different population groups perceive barriers to accessing PHC services becomes essential for the improvement of public policies. Investigating such perceptions in territories affected by urban violence and social vulnerability allows for the identification of persistent inequalities and supports the development of evidence-based policies. This study addresses this gap by examining perceived barriers to accessing PHC using vignette-based methods in a socially vulnerable urban community in Brazil. While grounded in a specific territorial context, the study seeks to contribute to a broader understanding of how social vulnerability, stigma and prior engagement with health services shape perceptions of access in LMIC settings. Therefore, this study aims to analyse perceptions regarding the difficulty of accessing PHC in a community in the municipality of Rio de Janeiro, using vignettes as a methodological tool, according to sociodemographic characteristics and self-reported health (highlighting health conditions amenable to primary health management and care). By exploring these dimensions, the study intends to contribute evidence to inform PHC organisation, support equity-oriented public policies and guide community-level interventions aimed at strengthening PHC in adverse contexts. Besides such broad objectives in the field of policies and community-level interventions, we aimed also to show that modern linguistic concepts and tools are fully compatible with the essential work on impoverished communities, overcoming challenging longstanding assumptions regarding the use of linguistic and psychometric tools in socially vulnerable populations. Methods Study design This is a cross-sectional study conducted in 2024 in the community of Manguinhos, located in the municipality of Rio de Janeiro, a territory characterised by high social vulnerability and frequent episodes of armed violence. The research employed a face-to-face survey using a structured questionnaire (via tablets using REDCap) based on contemporary methods in health surveys, including IRT and the use of narrative vignettes. The questionnaire and vignettes were specifically adapted for this study, based on a set of vignettes kindly sent to us by Nee and Witt24 (UK), and implemented using the REDCap electronic data capture platform, with data collected on palmtops by trained field researchers. Prior to data collection, the instrument was pilot tested to assess clarity, comprehension and adequacy to the local context. Adjustments were made based on pilot findings to improve question wording and the understanding of the vignettes. The use of electronic data collection contributed to standardisation, data quality control and reduction of data entry errors. Psychometric analyses, including IRT modelling, were conducted to evaluate the internal structure and performance of the instrument. Sampling A non-probability quota sampling method was adopted, stratified by sex (male and female) and age groups. The quotas were estimated using previous population-based surveys and data from the Pereira Passos Institute (https://ipp.prefeitura.rio/home/conheca-o-instituto/), aiming to reflect the basic demographics of the local population. The planned sample size was 1000 interviews (actually 1006 interviewees, in consequence of parallel interviewing process), modelled as large enough to ensure the statistical precision of analyses of the key variables25 in the context of the Manguinhos community. This non-probability quota sampling strategy was adopted due to the operational challenges of conducting probabilistic sampling in contexts marked by social vulnerability and urban violence, and continuous displacement and flux of people. The use of quotas allows for the inclusion of key sociodemographic groups and ensures diversity within the sample, although it does not aim to provide optimal statistical inference.26 Quota sampling for mobile populations is a non-probability sampling method designed to ensure that specific subgroups of a moving, transient, or hard-to-reach population are represented in a study, even when a complete sampling frame (a list of all members) is unavailable. This technique is frequently used when working with fragmented or highly diverse populations to gather structured, comparative data quickly and cost-effectively. Quota sampling has been used in similar contexts and populations.27 The sample size was defined based on precision criteria as defined by Rothman and Greenland.25 Empirical assessment of precision using the ‘Precisely’ R package28 indicated narrow CIs for key outcomes. For example, the proportion of respondents reporting high perceived difficulty in a vignette scenario was estimated at 30.9% (95% CI 28.1% to 33.8%), corresponding to a margin of error of approximately ±3 percentage points for a sample size of 1000 participants. This level of precision supports reliable estimation and subgroup analyses (sex, age and race/skin colour), as well as stable parameter estimation for IRT models. This approach is consistent with recommendations that prioritise estimation precision in studies focused on measurement and modelling of complex constructs.29 Inclusion and exclusion criteria Participants were eligible if they were residents of the Manguinhos community, aged 18 years or older, and able to understand and respond to the interview questions. Individuals were excluded if they had significant cognitive or communication impairments that prevented comprehension of the vignettes or the questionnaire, or if they declined to provide informed consent. Recruitment followed the predefined quota criteria to ensure diversity across key sociodemographic groups. Efforts were made to minimise selection bias through the combined use of (1) the positive aspects of quota sampling (ie, its flexibility to recruit mobile and hard-to-reach subpopulations); (2) the recruitment of a team of native interviewers (ie, people from the community under study), who have been comprehensively trained and were included in a pilot study and (3) continuous field coordination via first-hand observation and continuous monitoring, using a network of palmtops using REDCap (https://project-redcap.org/). There is a myth about impoverished communities in Brazil. There is no doubt that many essential items are lacking, such as formal education and specialised healthcare. Notwithstanding, the number of mobile phones is huge. Official figures document ~90% Brazilians have a cell phone.30 Of course, there is digital exclusion in rural areas and faraway regions such as the Tropical Rain Forest, but not in urban areas, even in impoverished communities. The vast majority of people, including people with very low education levels, are familiar with mobile phones and basic apps. In the context of an urban economy where services are a transient but essential powerhouse, people (especially poor people) must be connected to be included in the formal and informal work market. Brazilian sociologists coined the expression ‘Uberisation’ to describe this new and thriving market and communication network.31 Instruments and data collection Data were collected through in-person interviews using a structured questionnaire built on the REDCap platform. The instrument included questions on sociodemographic characteristics such as age, sex at birth, and self-reported race/skin colour, in addition to questions on physical health conditions such as medical diagnoses, follow-up care and medication use for hypertension and diabetes. It also included questions on previous mental health diagnoses and follow-up by mental health professionals, as well as the administration of the Patient Health Questionnaire-2 and Generalised Anxiety Disorder 2-item scales to screen for symptoms of depression and anxiety. Furthermore, narrative vignettes adapted from Nee and Witt24 were included. These described different profiles of individuals with or without mental disorders or histories of socially stigmatised behaviours. The vignettes served as a methodological strategy to assess participants’ perceptions of the difficulty in accessing a PHC Unit (Unidade Básica de Saúde—thereafter UBS) for hypothetical individuals facing those situations. Responses were recorded on a five-point Likert scale ranging from ‘very easy’ to ‘extremely difficult’. The instrument, including the vignettes, was culturally adapted to the Brazilian context, with attention to language, terminology and contextual relevance. The wording of the questions and scenarios was refined to ensure clarity and appropriateness for the target population. Linguistic validity was supported through translation and back-translation (by two bilingual translators), analysis using the theory of item response and pilot testing, which allowed the identification and correction of ambiguities and comprehension difficulties. The interviewer team was composed of professionals with prior experience in the territory and links to local health services, such as the ‘Consultório na Rua’ (roughly translated as Street Outreach Team, but without a precise translation into English) programme. All interviewers received specific training and a standardised manual for administering the instrument. A pilot study was conducted before the main data collection phase to ensure question clarity and proper field performance. Throughout the process, data confidentiality, participant anonymity and ethical conduct of interviews were ensured. Brief description of the adaptation of original vignettes for the sake of the survey As previously mentioned, the six original vignettes had been developed for other purposes by Nee and Witt.24 The original vignettes were written in British English and were initially translated (and then back-translated) into Brazilian Portuguese. Considering its application in a study of a totally different nature, targeting a different population, that is, the general population of a deprived and underserved favela community (Manguinhos), the Oswaldo Cruz Foundation research team decided to double-check its comprehensibility using two complementary strategies. We used resources from Artificial Intelligence, submitting the translated vignettes to a trained neural network. We submitted the question as follows: Are such texts comprehensible by individuals with just fundamental education or illiterate? We have also carried out a pilot study with 20 individuals from different ages and gender, all residing at the Manguinhos favela, addressing the same question. The results from both queries were practically identical: a new version using a less complex syntax and simple words. Outcome variable The study’s outcome variable was the perception of difficulty in accessing the UBS, measured using a five-point Likert scale in response to the different scenarios presented in the vignettes. Statistical analysis Statistical analyses were conducted according to the nature and distribution of the variables. Descriptive analyses were used to characterise the study population. Given the ordinal nature of the outcome variables derived from Likert scales, non-parametric tests were selected for group comparisons. Psychometric analyses, including IRT modelling using both compensatory (high skill in one dimension offsets low skill in another) and non-compensatory (mastery of all relevant skills is required), were employed to assess the internal structure and performance of the instrument, as appropriate for latent constructs measured through ordered categorical responses. Further details are available from the first authors. The methodological procedures were conducted in accordance with the guidelines presented in Ayala’s seminal textbook.32 IRT models were applied to analyse perceived difficulty in accessing PHC as a latent construct underlying participants’ responses to the vignettes. IRT is particularly suitable for this context as it allows the modelling of unobservable traits, such as perceived access barriers, based on ordered categorical responses. In this study, the latent trait represents the level of perceived difficulty in accessing PHC services. The models estimate parameters that describe how each vignette relates to this latent construct. The discrimination parameter indicates how well each vignette differentiates between individuals with different levels of perceived difficulty, while the difficulty parameters represent the thresholds at which respondents are more likely to endorse higher levels of perceived difficulty. Higher difficulty values indicate that greater levels of the latent trait are required for respondents to report more severe access barriers. This approach allows for a more nuanced understanding of how different scenarios capture variations in perceived access, providing insights beyond traditional descriptive analyses. Initially, the sample was described using absolute and relative frequencies, means and SD. Subsequently, Mann-Whitney and Kruskal-Wallis tests were applied to compare perceptions of access difficulty across sociodemographic groups (sex, age group and race/skin colour) and health-related variables (diagnosis, follow-up and medication use for hypertension and diabetes). Race/skin colour was self-reported based on IBGE census categories: White, Black, Brown, Yellow and Indigenous. For analysis, Black and Brown were combined as ‘Black population’, following IBGE and Ministry of Health guidelines.33 This categorisation, common in public health research, reflects shared experiences of discrimination and structural racism. Analyses were conducted comparing ‘Black’ and ‘non-Black’ groups.34 Subgroup analyses were conducted according to key sociodemographic characteristics (sex, age group and race/skin colour) and health-related variables, given their relevance to understanding inequalities in access to PHC. These analyses aimed to explore potential differences in perceived access barriers across population groups in a context marked by social vulnerability and structural inequities. Given the exploratory nature of the study and the use of non-parametric methods for subgroup comparisons, the results are primarily presented through point estimates and CIs. Where appropriate, estimates derived from modelling approaches were interpreted considering their analytical robustness and precision rather than exclusively relying on formal hypothesis testing. This approach is aligned with statistical traditions emphasising estimation and precision, including Neyman’s contributions to sampling and statistical inference, and later developments in epidemiological methodology.25 29 35 Missing data were minimal due to the use of electronic data collection with built-in validation procedures. Analyses were conducted using available data for each variable, and cases with missing values were excluded on a case-by-case basis. All analyses were conducted using R libraries for the analysis of Surveys36 and the IRT.37 Patient and public involvement Patients and members of the public were not involved in the design, conduct, reporting, or dissemination plans of this research. Ethical considerations All participants were informed about the study objectives, the voluntary nature of participation and guarantees of confidentiality. They were also informed of their right to refuse or withdraw from the study at any time, without any consequences. They signed the informed consent form, in accordance with Resolution No. 466/2012 of the Brazilian National Health Committee on Ethics (CONEP). All procedures were conducted in accordance with the ethical principles of the Declaration of Helsinki. Participants’ capacity to provide informed consent was assessed by trained interviewers at the time of recruitment. Only individuals who were able to understand the study objectives, the nature of their participation, and to provide voluntary consent were included. Individuals with significant cognitive or communication impairments that could compromise their ability to provide informed consent were excluded from the study. Given the sensitive nature of some topics addressed in the study, including mental health and social vulnerability, interviewers were trained to identify signs of emotional discomfort and to interrupt or discontinue the interview if necessary. Participants were also informed that they could skip any question or withdraw from the study at any time without any consequences. When appropriate, participants reporting distress were advised to seek care at local PHC services, and information about available services in the territory was provided. All data were handled in accordance with data protection and confidentiality standards. Personal identifiers were removed from the analytical dataset, and all information was stored securely using password-protected systems with restricted access limited to authorised members of the research team. Data were collected using encrypted electronic devices and managed through a secure data capture platform, ensuring data integrity and confidentiality throughout the research process. Data anonymisation procedures were applied prior to analysis, including the removal of personal identifiers and the use of non-identifiable codes for each participant. No directly identifiable information was included in the analytical dataset. Results No statistically significant differences were observed across racial groups for diabetes-related outcomes. In contrast, differences were identified by sex and age. Women presented higher frequencies of diagnosis compared with men (p=0.047), and a clear age gradient was observed, with increasing diagnosis across age groups (p<0.001), indicating a substantial rise over the life course (table 1). Table 1Distribution of absolute and relative frequencies of sociodemographic variables according to diagnosis, follow-up and treatment of diabetes, Manguinhos Rio de Janeiro, 2024VariableCategoryHave you ever been diagnosed with diabetes by a doctor?NoYesP valuen%n%Race/colourNon-Black23680.85619.20.532Black56182.711717.3SexFemale41079.810420.20.047Male38784.96915.1Age group18–29 years238961040.00030–44 years32183.86216.245 years or older23870.210129.8VariableCategoryHas any doctor ever prescribed you an oral medication for diabetes?NoYesP valuen%n%Race/colourNon-Black23981.85318.20.764Black56282.911617.1SexFemale41480.510019.50.092Male38784.96915.1Age group18–29 years24096.883.20.00030–44 years31983.36416.745 years or older24271.49728.6VariableCategoryDo you regularly visit a doctor/health service for diabetes follow-up?NeverOnly when a problem occursRegularlyP valuen%n%n%Race/colourNon-Black14750.38629.55920.20.529Black35151.817726.115022.1SexFemale26451.412123.512925.10.003Male23451.314231.18017.5Age group18–29 years15160.97731208.10.00030–44 years2035310527.47519.645 years or older14442.58123.911433.6 Distribution of absolute and relative frequencies of sociodemographic variables according to diagnosis, follow-up and treatment of diabetes, Manguinhos Rio de Janeiro, 2024 A similar pattern was observed for the prescription of oral medication and medical follow-up. While no differences were identified by race, significant differences emerged by sex and age, with women more frequently reporting regular follow-up (p=0.003) and younger individuals showing lower engagement in follow-up compared with older age groups (p<0.001) (table 1). These findings contribute to the study objective by indicating that differences in access to diagnosis, treatment and follow-up of diabetes were observed according to age and sex, while no statistically significant differences were identified by race in this sample. These variables were interpreted as indicators of prior engagement with health services, which may influence individuals’ perceptions of access to PHC. For hypertension, no statistically significant differences were observed between racial groups for diagnosis, treatment, or follow-up, with the exception of recent medication use. Differences were consistently identified by age group (p<0.001), with a marked increase in diagnosis, prescription and regular follow-up among older individuals (table 2). Similarly, these variables were interpreted as indicators of prior engagement with health services, which may shape individuals’ perceptions of access to care. Table 2Distribution of absolute and relative frequencies of sociodemographic variables according to diagnosis, follow-up and treatment of hypertension, Manguinhos, Rio de Janeiro, 2024VariableCategoryHave you ever been diagnosed with hypertension by a doctor?NoYesP valuen%n%Race/colourNon-Black19867.19732.90.116Black42261.626338.4SexFemale31560.120939.90.033Male30566.915133.1Age group18–29 years23193.5166.50.00030–44 years24362.814437.245 years or older14642.220057.8VariableCategoryHas any doctor ever prescribed you medication for hypertension?NoYesP valuen%n%Race/colourNon-Black19867.19732.90.219Black43062.825537.2SexFemale32361.620138.40.101Male30566.915133.1Age group18–29 years23595.1124.90.00030–44 years24864.113935.945 years or older14541.920158.1VariableCategoryDo you regularly visit a doctor/health service for hypertension follow-up?NeverOnly when a problem occursRegularlyP valuen%n%n%Race/colourNon-Black12743.16722.710134.20.506Black30544.613319.424636SexFemale22042.19718.520639.40.019Male21246.510322.614130.9Age group18–29 years15060.77229.12510.10.00030–44 years17144.28421.713234.145 years or older11132.24412.819055.1VariableCategoryIn the past 2 weeks, have you taken medication to control your hypertension (high blood pressure)?NoSome daysEvery dayP valuen%n%n%Race/colourNon-Black19164.7268.87826.40.035Black39657.89814.319127.9SexFemale30257.67013.4152290.301Male28562.55411.811725.7Age group18–29 years22591.1124.91040.00030–44 years23159.76617.19023.345 years or older13137.94613.316948.8 Distribution of absolute and relative frequencies of sociodemographic variables according to diagnosis, follow-up and treatment of hypertension, Manguinhos, Rio de Janeiro, 2024 Differences by sex were also observed for diagnosis (p=0.033) and follow-up (p=0.019), with women more frequently reporting diagnosis and regular monitoring. In contrast, men more often reported seeking care only when experiencing health problems. Use of antihypertensive medication in the 2 weeks prior to the survey showed a statistically significant difference by race (p=0.035), with slightly higher frequencies of regular use among Black individuals, as well as a strong gradient by age (p<0.001) (table 2). These findings indicate that differences in diagnosis, treatment, and follow-up of hypertension were observed according to age and sex, while no statistically significant differences were identified by race. These variables were interpreted as proxies of prior engagement with health services, suggesting that previous contact with care may influence how individuals perceive access to PHC (table 2). A multidimensional IRT model with four factors was fitted using oblique (oblimin) rotation.38 The rotated factor loadings indicated that each item was predominantly associated with a single factor, with communalities ranging from 0.533 to 0.999 (table 3), suggesting a good representation of the latent structure. Table 3Rotated factor loadings (Oblimin) and communalities (h²) of items from the four-factor multidimensional IRT modelItemF1F2F3F4h²Typical history of depression and anxiety0.00 0.998 0.000.000.999Typical history of alcohol dependence0.000.00 0.997 0.000.993Disadvantaged background, without mental health problems 0.946 0.000.000.000.858Control - no distinctive characteristics0.6540.0700.0410.000.533The bold values indicate the primary (highest) factor loading for each item in the four-factor multidimensional IRT model.IRT, item response theory. Rotated factor loadings (Oblimin) and communalities (h²) of items from the four-factor multidimensional IRT model The bold values indicate the primary (highest) factor loading for each item in the four-factor multidimensional IRT model. IRT, item response theory. The items related to depression/anxiety and alcohol dependence showed high loadings on distinct factors, indicating that these vignettes capture different dimensions of perceived difficulty, particularly those related to mental health conditions. In contrast, vignettes without mental health components were associated with a separate factor, suggesting a different pattern of perceived access (table 3). Inter-factor correlations ranged from –0.004 to 0.792, indicating moderate relationships between factors. Overall, these findings support the multidimensional nature of perceived barriers to accessing PHC services (table 3). The item probability curves (figure 1) from the unidimensional IRT model (graded response model, GRM) demonstrated good performance of the four vignettes. Response categories were well distributed along the latent trait continuum (θ), indicating that participants were able to discriminate between levels of perceived difficulty in accessing health services. Item probability curves in the unidimensional graded response model . The vignette referring to typical histories of depression and anxiety showed a higher probability of selecting more negative responses (‘extremely difficult’) as the latent trait increased, suggesting greater perceived barriers associated with common mental disorders. Similarly, the vignette on alcohol dependence displayed clear transitions between response categories, with a predominance of intermediate responses, indicating greater variability in perceptions regarding this condition (figure 1). In contrast, vignettes describing individuals without mental health problems, whether in a socially vulnerable context or a stable profile, were associated with lower levels of the latent trait. In these cases, the probability of selecting ‘extremely easy’ was higher among respondents with lower levels of judgement, while ‘extremely difficult’ responses were concentrated at the upper end of the continuum. This pattern suggests lower levels of stigmatising attribution associated with these profiles (figure 1). Figure 2 displays the total information curve derived from the unidimensional GRM fitted to the vignette items. The curve shows that the instrument provides the highest measurement precision between approximately θ = −3 and θ = +2, indicating that the items are most informative for individuals with low to slightly above-average levels of the latent trait. Information decreases outside this range, suggesting reduced precision at the extremes. Overall, the model demonstrates greater capacity to differentiate individuals at lower levels of the latent trait. Total information curve from the unidimensional graded response model fitted to the vignette items. These results provide additional insight into how perceptions of access to health services vary according to different social and clinical profiles. Taken together, the findings suggest that both individual characteristics and prior engagement with health services play an important role in shaping perceived access barriers. Discussion The findings of this study highlight the complexity of perceptions regarding access to PHC in contexts marked by social vulnerabilities and urban violence, such as the Manguinhos community in Rio de Janeiro. Although situated in a specific urban territory, these findings reflect broader challenges observed in socially vulnerable settings in LMICs, where structural inequalities, stigma and fragmented health systems shape access to care. The use of vignettes as a methodological tool enabled the capture of important nuances in participants’ perceptions of barriers to accessing health services. Initially, no statistically significant differences were observed in perceived difficulty in accessing Basic Health Units between Black and non-Black individuals. This result may suggest that in highly vulnerable contexts, structural barriers common to the territory may override perceived racial inequalities in access.39 However, studies indicate that barriers to healthcare access can be categorised as geographical, financial, informational and organisational, affecting different groups in distinct ways.40 These findings indicate that, in such contexts, interventions aimed at improving access to PHC should prioritise territorial and structural barriers affecting the entire population, while still recognising that racial inequalities may operate through other dimensions not captured by perceived access alone. It is important to emphasise that the findings of this study should be interpreted as associations rather than causal relationships. Given the cross-sectional design and the use of perception-based measures derived from vignettes, it is not possible to establish causal links between sociodemographic characteristics, contextual factors and perceived barriers to access. The observed differences between groups may reflect underlying structural conditions, but causal inferences cannot be drawn from the data. The data made evident that men reported greater difficulty in access when the vignettes described scenarios involving depression and anxiety. This finding may reflect a component of internalised stigma towards mental disorders, especially among men, a group for whom psychological suffering is still socially less acceptable and often overlooked in PHC services. The literature highlights that stigma associated with mental health remains a significant barrier to access and continuity of care.41,43 These results suggest the need for PHC strategies specifically aimed at reducing mental health stigma among men and promoting gender-sensitive approaches to engagement and care. The influence of age on access perceptions was also evident. Younger individuals tended to perceive greater difficulty in scenarios lacking clinical conditions or involving behaviours considered deviant. This may reflect a generalised distrust among youth regarding the responsiveness of health services to non-biomedical or socially marginalised demands, indicating a misalignment between the expectations of younger users and the current PHC offer. Studies indicate that adolescents face specific barriers in accessing PHC, including lack of confidentiality, moral judgement and service hours incompatible with their routines.44 45 These findings highlight the need to adapt PHC services to better address the expectations and specific barriers faced by younger populations, including improving accessibility, confidentiality and responsiveness to non-clinical demands. The data revealed that the perception of lower difficulty in access was associated with having a formal diagnosis and prescribed medication, both for diabetes and hypertension. These findings reinforce the idea that a connection with the health system, formalised through diagnosis and treatment, may contribute to greater familiarity with services and, consequently, to a lower perception of barriers. Conversely, individuals without diagnosis, follow-up or continuous medication use reported greater difficulty in access, suggesting that the absence of prior engagement in care may compromise trust in the accessibility of services. This result underscores the importance of strengthening continuity of care and proactive engagement strategies, particularly for individuals without prior diagnosis or regular follow-up, as identified in this study. The literature points out that continuity and consistency of care over time are essential attributes of PHC that positively influence the perception of access.46,49 These findings have important implications for health policy and the organisation of PHC services. While some of the observed patterns are likely shaped by the specific socio-territorial context of Manguinhos, characterised by high levels of social vulnerability and urban violence, other findings may have broader relevance to similar urban settings in LMICs, where structural barriers, stigma and fragmented care also influence access to PHC. In socially vulnerable territories, such as Manguinhos, strengthening PHC requires not only expanding service availability but also improving the capacity of services to engage populations with different needs and levels of prior connection to care. Evidence suggests that barriers to access in PHC are strongly influenced by stigma, fragmentation of care and insufficient continuity of services, particularly for individuals with mental health conditions.50,52 Strategies aimed at enhancing continuity of care, particularly for individuals without previous diagnosis or regular follow-up, may contribute to reducing perceived barriers to access, especially in contexts marked by social vulnerability and violence.53 In particular, the finding that individuals without prior diagnosis or follow-up reported greater perceived barriers highlights the need for strategies that actively promote entry into and continuity within the health system. These findings can be interpreted in light of established frameworks on access to care, which emphasise the interaction between structural conditions and individual capacities in shaping health service utilisation. In addition, the findings highlight the need for PHC services to incorporate approaches sensitive to gender and age-related differences, as evidenced by the greater perceived difficulties reported by men in mental health-related scenarios and by younger individuals in specific contexts. This includes actions to address stigma associated with mental health and to improve responsiveness to younger populations, who often face additional barriers related to recognition and help-seeking.52 From a system-level perspective, these results reinforce the importance of adapting PHC organisation to territorial and social contexts, ensuring that services are accessible, welcoming and capable of addressing both clinical and psychosocial demands. In the Brazilian context, this includes recognising structural inequalities and the role of social determinants in shaping access to care.54 The application of the IRT showed that the vignettes used had good psychometric performance, with probability curves consistent across the latent judgement trait. In this context, higher probabilities of selecting categories indicating greater difficulty reflect higher levels of the latent perception of access barriers. Conversely, lower probabilities indicate scenarios perceived as more accessible. The vignette addressing common mental disorders (depression and anxiety) concentrated the highest proportion of responses in the category of greatest perceived difficulty in access, reflecting persistent stigma towards mental health. The vignette related to alcohol dependence presented more ambiguous response patterns, suggesting more heterogeneous perceptions among participants. In contrast, vignettes unrelated to mental health problems were less associated with difficulty in access, possibly reflecting the normalisation of access for profiles perceived as ‘non-problematic’ or neutral. The use of vignettes in health research has proven effective in assessing subjective perceptions and identifying stigma associated with specific health conditions.18 55 These findings reinforce the relevance of addressing stigma as a key barrier within PHC services. While grounded in a specific territorial context, this study contributes to a broader understanding of how perceived barriers to access are shaped by social and structural conditions in vulnerable settings. By highlighting the role of stigma, prior engagement with health services, and sociodemographic factors, the findings offer insights that may inform the design of more equitable and context-sensitive PHC strategies in LMICs. Conclusions These findings underscore the need to strengthen the training of PHC teams to better address mental health and social vulnerability demands, while also tackling the stigma and subjective barriers that continue to hinder access. They also highlight the importance of targeted strategies for young people and men, groups that perceive greater difficulty in access, particularly in situations involving psychological distress. Understanding the barriers perceived by users is essential for the improvement of public policies and the promotion of equity in access to healthcare services. Such evidence should inform health managers and policymakers in the implementation of more responsive and context-sensitive actions, tailored to the specific needs of different population groups. Investing in better alignment between service provision and user expectations is crucial for consolidating a truly universal, equitable, and comprehensive PHC system. This study presents some limitations that should be acknowledged. First, the use of a non-probability quota sample limits the generalisability of the findings to the broader population of the community or to other settings. While the sampling sought to reflect key demographic characteristics, potential selection biases may remain. Notwithstanding, in a deprived and underserved community, classic household surveys usually miss people living in non-standard households, those in provisional shelters or individuals experiencing homelessness who move across the territory under analysis. Subgroup analyses should be interpreted with caution. Given the non-probability sampling design, the results are not statistically representative of the broader population. In addition, some subgroup analyses may have been limited by reduced sample sizes within categories, which may affect the stability and precision of estimates. These analyses were primarily exploratory and intended to identify patterns rather than establish definitive differences between groups. Furthermore, as the study relied on self-reported perceptions derived from vignette scenarios, the findings may be subject to response bias, including social desirability bias. Participants may have provided answers they perceived as more socially acceptable, particularly in relation to sensitive topics such as mental health and access to care. Although the use of vignettes reduces direct personal exposure, respondents’ interpretations may still be influenced by their own experiences and beliefs. In this sense, recall bias cannot be ruled out, as participants may have drawn on previous experiences with health services when evaluating the scenarios. Second, the cross-sectional design limits causal inference. Third, self-reported data on health and behaviours may involve bias, including underreporting or social desirability. Although the vignettes were culturally adapted and piloted, interpretations may vary by education or experience. Also, as the study focused on one urban community affected by violence and vulnerability, findings may not generalise to other settings. Nonetheless, the study offers insights into perceived barriers to PHC and points to relevant paths for future research and action. Footnotes Funding: This study was supported by the Carlos Chagas Filho Foundation for Research Support of the State of Rio de Janeiro (FAPERJ), grant number E-26/010.002428/2019. Data availability free text: The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request, subject to ethical and data protection considerations. Patient consent for publication: Not applicable. Provenance and peer review: Not commissioned; externally peer reviewed. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Ethics approval: The study was approved by the Research Ethics Committee of the Oswaldo Cruz Foundation (FIOCRUZ), under CAAE number 78822424.9.0000.5241. All participants provided written informed consent prior to participation. Data availability statement Data are available on reasonable request. References Associated Data Data Availability Statement Data are available on reasonable request.