Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Health literacy, understood as the set of abilities that enable people to access, understand, appraise and use health information and services, plays a major role in diabetes management and social inequities in health. It is, however, most often approached in a global way, without considering its different dimensions, which limits the identification of the diverse skills, barriers and available resources across individuals.WHAT THIS STUDY ADDS This study identifies distinct multidimensional health literacy profiles among people living with type 2 diabetes, showing that although less favorable profiles often coincide with social disadvantage and poorer health, health literacy does not systematically follow social gradients, as some socially disadvantaged individuals display strong skills in specific dimensions while some socially advantaged individuals experience marked difficulties.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY By moving beyond a unidimensional and deficit-based view of health literacy, this profiling approach shows that similar combinations of overall abilities can mask very different barriers to care and opportunities for intervention, thereby informing proportionate and tailored strategies to reduce social inequities in diabetes care.Context Type 2 diabetes (T2D) is rising worldwide 1 and affects over 5% of the French population,2 with marked social inequities in prevalence and complications.3 4 Effective T2D management relies on patients’ active engagement in daily disease control and follow-up.5Managing T2D requires a wide range of complex skills, including proficient health literacy. The WHO defines health literacy as a multidimensional concept, which refers to people’s knowledge, confidence and comfort in accessing, understanding, appraising, remembering and using information about health and healthcare for their own health and well-being, and for others around them.6 7 Recognized as a key lever by the WHO6 for non-communicable diseases, including diabetes, health literacy acts both as a direct determinant of health and a bridge linking other determinants to health outcomes.8Health literacy in T2D is associated with glycemic control, knowledge, self-care, disease management.9 10 Key health literacy skills (eg, understanding health information, engaging with providers11) are associated with health behaviors12 and social position.13–15 Strengthening health literacy may therefore improve health outcomes and reduce inequities.Most studies assess health literacy as a single overall score or evaluate a single dimension and fail to capture people’s diverse health literacy skills. For example, two people may have the same mean health literacy score but have very different combinations of strengths and weaknesses. The WHO-endorsed Ophelia (Optimising Health Literacy and Access) process addresses this issue by identifying multidimensional health literacy profiles to inform codesigned, context-specific interventions.16 Specifically, these interventions combine health literacy skills with social and clinical characteristics to guide local actions that reduce social inequities in health (SIH). This approach has not yet been applied to T2D management in France, despite the wide variability in health literacy and marked SIH in this population.3 4 15In this context, we aimed to identify profiles with homogeneous skills and needs in terms of health literacy among a sample of individuals living with T2D in mainland France and to describe their socioeconomic and demographic characteristics, capacities and resources, diabetes history, healthcare utilization and health status, with a view to prioritizing needs for targeted interventions.Methods Data sources Entred 3 is the third edition of a nationally representative French diabetes population-based survey describing specifically the characteristics and health status of people with diabetes. In 2019, 8728 adults with diabetes residing in mainland France were randomly selected from national health insurance databases, using a validated algorithm based on reimbursements for hypoglycemic treatment. 17 18 Data on this sample were extracted from the French National Health Data System (Système National des Données de Santé, SNDS)—France’s health administrative data which include outpatient care reimbursements, hospitalizations, vital status, causes of death—and from other administrative databases.Moreover, selected participants were invited to complete the Entred 3 survey questionnaire, which collected data on socioeconomic status (SES), demographic characteristics, health literacy, healthcare use, health outcomes and behaviors. The questionnaire was available in three formats: a short telephone version, a detailed paper-mail version and a comprehensive online version. For the present study, analyses were restricted to the subsample of individuals with T2D who responded online, as this format included the most comprehensive set of questions related to health literacy. The types of data collected according to the data source (ie, online questionnaire and SNDS/other administrative database) are described in online supplemental table S1.SP110.1136/bmjdrc-2026-005977.supp1Supplementary dataTo assess health literacy, we used four (HLQ3, HLQ6, HLQ7 and HLQ9) of the nine scales of the Health Literacy Questionnaire (HLQ), a highly robust and reliable instrument, developed using a validity-driven approach.11 The HLQ was translated and culturally adapted to French using a standardized forward–backward method.14 19 The selected scales (table 1) were closely linked to diabetes control and core competencies required for healthcare participation10 20 and included in the online questionnaire only. Qualitative interviews previously conducted in the French context highlighted that these scales cover a variety of skills relevant to diabetes management.10HLQ3 ‘Actively managing my health’ (five items) assesses individuals’ ability to take charge of their health by seeking relevant information and implementing appropriate strategies.HLQ6, ‘Ability to engage with healthcare providers’ (five items) assesses individuals’ ability to convey their health situation to healthcare professionals, to discuss their concerns with professionals and to ask them questions, and to understand everything they feel they need to know.HLQ7, ‘Navigating the healthcare system’ (six items) assesses individuals’ ability to access, understand and effectively use healthcare services.HLQ9, ‘understanding health information well enough to know what to do’ (five items) assesses people’s comprehension of prescriptions, of healthcare professionals’ instructions, and of healthcare products as well as their ability to complete medical forms.Table 1Description and interpretation of the four Health Literacy Questionnaire (HLQ) scales used in the Entred 3 studyHLQ scaleItems, score range and interpretationPart 1: How strongly do you disagree or agree with the following statements? (strongly disagree/disagree/agree/strongly agree)—score range 1–4HLQ3:actively managing my healthItems*:I spend quite a lot of time actively managing …I make plans for what I need to do to be …Despite other things in my life, I make time …I set my own goals about health and fitness.There are things that I do regularly …High score: individuals recognize the importance of taking responsibility for their own health. They actively engage in their own care, seek relevant health information, and make informed decisions regarding their well-beingLow score: individuals do not perceive their health as their own responsibility. They are disengaged from healthcare processes and view healthcare as something that happens to them rather than something they actively participate in.Part 2: How easy or difficult are the following tasks for you to do now? (Cannot do or always difficult/usually difficult/sometimes difficult/usually easy/always easy)—score range 1–5HLQ6:ability to engage with healthcare providersItems*Make sure that healthcare providers understand …Feel able to discuss your health concerns with a …Have good discussions about your health…Discuss things with healthcare providers…Ask healthcare providers questions to get…High score: individual is proactive about their health, feels in control of their interactions with healthcare providers, and is empowered to seek advice from additional professionals until their needs are met.Low score: passive approach to healthcare—where the individual does not actively seek or clarify information, accepts what is offered without question, and is unable to share concerns or ask for necessary clarifications—potentially impeding shared decision-making and optimal care management.HLQ7:navigating the healthcare systemItems*Find the right healthcare.Get to see the healthcare providers you need.Decide which healthcare provider you need…Make sure you find the right place to get…Find out what healthcare services you are…Work out what is the best care for you.High score: Individuals are capable of identifying available healthcare services and support systems to meet their needs. They can effectively navigate the healthcare system and advocate for themselves at both the system and service levels.Low score: Individuals struggle to advocate for themselves and cannot find someone to assist them in accessing appropriate healthcare services. They tend to rely only on the most obvious resources and have a limited understanding of available services and their entitlements.HLQ9:understanding health information well enough to know what to doItems*Confidently fill medical forms in the correct …Accurately follow the instructions from …Read and understand written health …Read and understand all the information on …Understand what healthcare providers …High score: individual is able to understand all written information (including numerical data) related to their health and can complete medical forms appropriately.Low score: individual has difficulty understanding any written health information or instructions about treatments or medications and is unable to read or write sufficiently well to complete medical forms.*The HLQ is protected by copyright; the full version of the items comprising each dimension can be obtained from Swinburne University at the following address: HLQ-info@swin.edu.au.The HLQ3 score ranges from 1 to 4, corresponding to the average scores of the five items, using a 4-point scale (1=strongly disagree, 2=disagree, 3=agree, 4=strongly agree). The HLQ6, HLQ7 and HLQ9 scores range from 1 to 5, again corresponding to the average score, but using a 5-point difficulty scale (1=cannot do or always difficult, 2=usually difficult, 3=sometimes difficult, 4=usually easy, 5=always easy). For each scale, if more than half of the items are missing, the score is not calculated. If half or fewer responses are missing, then the missing items are imputed based on the mean score of the non-missing items for that scale.Additional self-reported skills and resources related to health literacy but not captured by the HLQ were also assessed, including whether the questionnaire was completed with or without assistance, living alone and satisfaction with social support received from loved ones related to diabetes management (no need for this support, satisfied or not satisfied).Diabetes history: hypoglycemic treatment was assessed from treatment reimbursements extracted from the SNDS; it was categorized into three classes (oral antidiabetic drug (OAD) monotherapy, OAD multitherapy, insulin therapy alone or combined), used as a proxy for diabetes severity. Diabetes duration was self-reported.Demographic and socioeconomic indicators: country of birth (France, other), age, considered as a continuous variable, and sex were extracted from administrative databases. Education level and perceived financial situation were self-reported and each was dichotomized as at least a high school diploma versus less than high school diploma, and a favorable financial situation (ie, ‘You are financially comfortable/things are financially ok’) versus an unfavorable situation (‘You just get by/It’s difficult to get by/You cannot get by without going into debt’).Perceived health status and quality of life were self-reported, using the EQ-5D-5L questionnaire, which provides a utility score (from 0 to 1, the latter reflecting the best score) based on five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression and includes a visual analog scale (score from 1 to 100).21 22The following diabetes-related complications (self-reported or hospitalization in the preceding 10 years extracted from the SNDS) were assessed: renal complications (hospitalization for renal transplant or dialysis procedure); lower limb complications (self-reported current or past foot ulcers, hospitalizations for foot ulcers or amputations); stroke (self-reported or hospitalization); coronary complications (self-reported or hospitalization for myocardial infarction); eye complications (self-reported retinopathy); dental complications (self-reported periodontal disease). Finally, we created a composite indicator of any existing complication at the time of the survey, whether self-reported or recorded in the SNDS.The following risk factors for diabetes-related complications were assessed: self-reported overweight or obesity (body mass index (BMI) ≥25 kg/m2), alcohol consumption, assessed using the self-reported Alcohol Use Disorders Identification Test–Consumption (AUDIT-C) questionnaire (moderate consumption defined as a score of 0–3 for men and 0–2 for women; high to severe consumption defined as a score of 6–12 for both sexes), self-reported tobacco smoking (defined as current smoker or having quit less than 3 years prior to the study) and both hypertension and hyperlipidemia, identified either through self-report or via treatment reimbursements from the SNDS in the past year. Low treatment adherence was assessed using the Girerd scale, a validated self-reported questionnaire used to identify levels of adherence among patients with chronic conditions (score ≥3, indicating low adherence).23Healthcare utilization for diabetes control was defined as having had, during the previous year (1) at least three visits to a general practitioner, (2) at least one visit to an endocrinologist and (3) at least 3 glycated haemoglobin (HbA1c) test reimbursements, as recorded in the SNDS.Statistical analyses We used several key aspects of the Ophelia process 16 to identify similar health literacy profiles. The Ophelia process specifies that all nine HLQ scales must be used in the cluster analysis to capture the full breadth of the health literacy construct. Given that Entred 3 only collected four HLQ scales, we sought to supplement the missing constructs with other data as noted above.For the cluster analysis, we employed agglomerative hierarchical clustering (AHC) on the scores of the four scales of the HLQ included (see above). All scores were standardized using z-scores to account for differences in scale. In line with Ophelia recommendations, we used Ward’s method to perform the AHC. This method is particularly suited for exploratory purposes; it groups individuals into clusters based on their similarity across different health literacy scales without requiring a priori assumptions about the number of clusters. Furthermore, it ensures that the number of clusters balances statistical indicators with the potential for each cluster to generate clinically and socially relevant insights, with each additional split required to produce a new cluster that is both informative and clearly distinct from its parent group in terms of health literacy, clinical and social characteristics.We also used the Dunn index—which is not part of the Ophelia process—to provide a complementary, objective measure of the clustering solution in order to support our Ophelia-based findings. While this index reflects the overall clustering structure and does not ensure that all clusters are equally well separated, the selected solution optimized the balance between within-cluster homogeneity and between-cluster separation.24 25 Dunn index values for candidate solutions are provided in online supplemental table S2. Once defined, clusters were described according to their socioeconomic and demographic characteristics, capacities and resources, diabetes history, healthcare utilization and health status. To further identify the most clearly distinct, clinically and socially interpretable and actionable profiles, we conducted a visual inspection of the clustering structure using principal component analysis (PCA), focusing on clusters showing minimal overlap in the projection space (online supplemental methods, table S3 and figure S1). All analyses were unweighted and performed using R (V.4.1.2) software.Results Population description A total of 510 individuals completed all four HLQ scales in the online questionnaire. The mean HLQ score for HLQ3 was 2.84 (SD=0.64), while for HLQ6, HLQ7 and HLQ9, it was 4.10 (SD=0.71), 3.65 (SD=0.74) and 4.05 (SD=0.68), respectively. Mean age was 62 years; 71% were men. Median duration of diabetes was 10 years, and 18% of the study sample was treated with insulin ( table 2). Online supplemental table S4 describes the characteristics of the online subsample used for the cluster analysis and the overall Entred 3 population with T2D in mainland France. Compared with the full study population, the online subsample was younger, more often male, and differed across several socioeconomic characteristics.Table 2Table showing each cluster profile according to health literacy skills and resources, and social characteristics among the 510 online participants who completed the four HLQ dimensions of the Entred 3 survey questionnaireClusterAllABCDEFGHIJKLMNumber of persons51012244156744547501656173933Proportion of the sample (%)1002.44.78.011.014.58.89.29.83.111.03.37.66.5HLQ scores Actively managing their health (HLQ3)Mean2.843.452.662.471.913.082.493.042.763.903.092.002.963.96SD0.640.550.390.490.380.350.170.160.270.160.480.500.250.09 Ability to actively engage with healthcare providers (HLQ6)Mean4.102.353.133.423.923.663.954.054.244.104.884.994.964.87SD0.710.690.760.330.380.400.200.110.540.210.170.050.090.27 Navigating the healthcare system (HLQ7)Mean3.652.132.273.043.413.303.703.943.663.753.794.574.854.57SD0.740.660.480.330.440.380.290.120.480.230.480.320.200.45 Understanding health information well enough to know what to do (HLQ9)Mean4.052.533.653.034.003.823.783.954.833.943.954.924.954.85SD0.681.030.560.510.350.310.260.130.200.160.310.140.100.26Skills and resources (%) Completed questionnaire without assistance84.058.379.255.392.784.788.984.189.693.383.9100.081.690.0 Household size>181.554.587.573.282.180.388.984.885.787.578.688.278.978.8 Diabetes related social support No need17.60.012.517.123.219.224.46.414.018.817.941.212.818.8 Quite satisfied57.645.545.851.250.052.153.361.758.068.864.341.274.475.0Social characteristics (% if not specified) Men71.650.075.058.564.370.373.376.686.075.075.088.271.860.6 Born in France84.358.387.575.691.189.284.487.280.081.382.182.492.378.8 Mean age (years)61.968.060.859.358.662.261.262.762.963.460.764.663.264.5 At least high school diploma56.125.050.039.069.650.768.945.776.040.043.676.561.560.6 Favorable perceived financial situation68.616.766.765.956.467.171.168.173.575.067.976.584.681.3Reading guide: Clusters are ordered by increasing mean score across the four HLQ scale scores. The color gradient (from red to orange to yellow to green) reflects the score level: red reflects the lowest scores; green reflects the highest scores.HLQ, Health Literacy Questionnaire.Clusters Cluster analysis identified 13 distinct health literacy profiles ( table 2), each with a specific combination of HLQ scale scores, highlighting respective strengths and needs. Five of these clusters exhibited distinct, clinically and socially coherent profiles (tables 2 and 3, online supplemental methods and figure S1), described below. The remaining clusters are summarized in online supplemental results.Cluster M: several strengths, no difficulty (6.5% of the overall sample). Persons in Cluster M had strong abilities across all four HLQ dimensions. Forty-nine percent were women. Individuals in this cluster had more favorable socioeconomic characteristics than the overall sample (81% reported a favorable financial situation; 61% had at least a high-school diploma). They did not report low treatment adherence and had a very high quality of life (utility score 0.93). They also had the lowest exposure to diabetes-related risk factors (59% with BMI ≥25 kg/m², 6% were smokers, none had high or severe alcohol consumption, and 68% and 63%, respectively, had hypertension and hyperlipidemia). To summarize, cluster M combined the broadest set of health literacy strengths with the lowest risk profile.Cluster A: active role in health management but struggled with both understanding information and engagement with health services (2.4% of the overall sample). Persons in cluster A had strong skills in active health management (HLQ3) but faced more challenges in the three other scales. Fifty percent were women. This cluster had the longest median diabetes duration (19 years), the least favorable socioeconomic characteristics (17% had a favorable financial situation and only 25% had at least a high school diploma) and the highest proportion of individuals born outside France (58%). Health service use was low (the lowest rate of endocrinologist consultations; 25% did not have an annual HbA1c test). Cardiovascular and cerebrovascular complications were common (33%), as were foot ulcers (16%). Quality of life was the lowest among all 13 clusters (utility score 0.46), although alcohol and tobacco use risks were also the lowest. To summarize, although this cluster combined strong self-management abilities, their impact seemed to be reduced by needs in other dimensions of health literacy.Cluster K: high health literacy scores apart from active role in health management (3.3% of the overall sample). Persons in cluster K were diametrically opposed those in cluster A. It comprised 88% men; 41% of the cluster reported not needing support from others. SES was high (77% had a favorable financial situation, and 77%—the highest proportion among all 13 clusters—had at least a high school diploma). Despite these advantages, this cluster was very exposed to diabetes-related risk factors (47% had moderate to severe alcohol consumption, 19% were smokers, 71% had hypertension and 71% hypercholesterolemia). To summarize, this cluster paired strong socioeconomic advantages and cognitive strengths with low engagement in health management and high exposure to diabetes-related behavioral risks.Cluster D: moderate health literacy scores, no active engagement in health management (11.0% of the overall sample). People in cluster D had limited engagement in health management (ie, HLQ3) and mixed abilities in the other three HLQ scales. This cluster had the most recent diabetes diagnosis (median duration 7.5 years). SES was higher than the average across all clusters in education level (70% holding a high school diploma) but lower regarding perceived financial situation (56% favorable). They were very exposed to diabetes-related risk factors (91% had overweight or obesity, 44% had moderate, high or severe alcohol consumption and 21% were smokers). In addition, 31% reported low adherence to diabetes treatment. They had among the highest rates of lower limb complications (14%) and coronary complications (20%) in the whole sample. Interestingly, their mean quality of life was higher (utility score 0.92) than for the 13 clusters combined. To summarize, cluster D combined recent diabetes diagnosis and moderate health literacy abilities with a high prevalence of both metabolic risk factors and early complications.Cluster L: moderate engagement in health management but high scores in other health literacy dimensions (7.6% of the overall sample). People in cluster L moderately engaged in health management (ie, HLQ3 scale) but had much stronger skills for the other three scales. They had a higher SES but were very exposed to diabetes-related risk factors (90% had a BMI ≥25 kg/m², 87% reported hypertension, 82% reported hypercholesterolemia and 13% were smokers). Only 8% reported low treatment adherence. To summarize, cluster L was marked by strong relational and cognitive health literacy skills but a high prevalence of risk factors combined with moderate engagement in health management.Table 3Table showing each cluster profile according to diabetes history, health status and health behaviors among the 510 online participants who completed the four HLQ dimensions of the Entred 3 survey questionnaireClusterAllABCDEFGHIJKLMHLQ scores Actively managing their health (HLQ3)Mean2.843.452.662.471.913.082.493.042.763.903.092.002.963.96 Ability to actively engage with healthcare providers (HLQ6)Mean4.102.353.133.423.923.663.954.054.244.104.884.994.964.87 Navigating the healthcare system (HLQ7)Mean3.652.132.273.043.413.303.703.943.663.753.794.574.854.57 Understanding health information well enough to know what to do (HLQ9)Mean4.052.533.653.034.003.823.783.954.833.943.954.924.954.85Diabetes history Diabetes duration (median, years)10.019.012.016.07.511.08.011.510.09.59.012.09.011.0 On insulin therapy at time of study (%)18.633.312.541.512.521.611.125.58.025.012.523.510.324.2Health status Quality of life (score/1)0.890.460.810.760.920.870.940.960.890.970.930.950.940.93 Self-perceived health (/100)71.954.265.657.869.967.876.075.773.478.874.682.178.079.7 Diabetes complications (%) At least one complication39.472.740.941.543.643.720.037.030.653.350.029.434.237.9 Lower limb7.318.24.519.514.34.24.44.32.00.08.96.35.33.3 Stroke3.216.74.27.30.02.84.54.40.07.13.66.30.00.0 Coronary disease15.925.012.514.619.621.14.513.016.021.419.612.518.46.9 Retinopathy6.78.320.84.910.710.82.24.36.112.53.60.02.63.0 Kidney disease0.40.00.00.00.01.40.02.20.00.00.00.00.00.0 Periodontal disease15.416.712.57.314.520.58.914.916.025.021.45.910.321.2Health behavior: treatment adherence, diabetes control, exposure to risk factors, and healthcare use (%) Low adherence to treatment (%)15.533.326.136.630.916.712.22.18.50.014.311.87.90.0 Healthcare use during previous year (%) At least three visits to a general practitioner80.991.791.782.982.180.679.580.468.085.785.775.089.580.0 At least one visit to an endocrinologist14.68.38.314.68.919.415.915.26.07.117.925.015.823.3 At least three HbA1c tests64.341.762.565.950.072.256.867.472.064.366.168.860.573.3Diabetes risk factors (%) BMI ≥2582.475.091.778.090.977.885.491.387.260.080.882.489.759.4 Alcohol consumption (moderate+high + severe)32.18.341.729.343.631.047.732.630.013.328.647.123.119.4 tobacco smoker13.80.012.514.621.49.513.617.412.212.514.318.817.96.1 High blood pressure (declared or treated)75.975.075.078.076.880.075.071.768.081.373.282.487.267.7 Hyperlipidemia (declared or treated)75.3100.070.878.972.777.884.171.768.073.370.987.582.162.5Reading guide: clusters are ordered by increasing mean score across the four HLQ scale scores. The color gradient (from red to orange to yellow to green) reflects the score level: red reflects the lowest scores; green reflects the highest scores.BMI, body mass index; HbA1c, glycated haemoglobin; HLQ, Health Literacy Questionnaire.Discussion The objective of this study was to identify profiles with similar strengths and weaknesses in terms of health literacy among a sample of individuals living with T2D in mainland France, with a view to informing future-targeted interventions. Of the 13 health literacy profiles, we identified based on four dimensions of the HLQ using the Ophelia process, we selected five because of their clinical and public health relevance. Our findings show that health literacy abilities can vary widely between individuals managing the same condition, leading to distinct barriers to and levers for care. Our insights can guide targeted interventions to maximize inclusion, reach and impact, ensuring no subgroup in our sample is left behind.Implications and actionable strategies Health literacy influences both health status and relationship between social position and health, 8 underscoring the need for tailored public health interventions. This is particularly true for France, where there are substantial social inequities in diabetes.3 4 In line with the principle of proportionate universalism,26 interventions must be accessible to all, but adapted in their intensity and form to each group’s needs. The Ophelia process16 offers a structured approach to do this in the context of public health intervention development: once profiles are mapped, priority groups are identified and their needs are assessed using quantitative and/or qualitative data. Tailored interventions are then codesigned with local stakeholders (eg, patients, healthcare professionals, health authorities).In the present study, people represented in clusters A and D, combining limited health literacy, socioeconomic vulnerability and poorer health, warrant particular attention. People in cluster A may benefit from enhanced support and navigation tools while for cluster D, people may need interventions to strengthen general health literacy and engagement. People in cluster K had strong cognitive and relational skills but little engagement in health management. Tools such as digital apps and targeted messaging might help to foster greater engagement in this cluster by leveraging existing strengths. All these hypotheses will be further explored in the second phase of our project, named Alisa-Diab project, which is part of the French contribution to the European Commission co-funded Joint Action project to reduce cardiovascular diseases and diabetes (JACARDI).27 Conducted in two very different French sites, this phase will include a contextual assessment of health literacy using qualitative methods, in collaboration with local stakeholders including people living with T2D.Our findings have significant implications at multiple levels. Clinically, they can be used to develop more personalized care by enabling healthcare professionals, who know their patients’ socioeconomic background and health status, to recognize likely health literacy profiles and to adapt their communication, education and follow-up strategies accordingly, not only addressing limitations but also leveraging patients’ existing strengths. At the health service level, they can be used to guide the development of structured patient education programs, navigation support services, and multiprofessional coordination targeting the needs of specific groups. At the policy level, they can inform national and regional strategies, by identifying some priority health literacy needs in France. Importantly, our results challenge common assumptions that equate health literacy level solely with socioeconomic and health status. For example, people in clusters A and K show that a high SES does not necessarily translate into strong health literacy across all dimensions, and that socioeconomically disadvantaged groups may have strong health literacy competencies. Recognizing this complexity can help to avoid oversimplified judgments, foster more equitable care and ultimately improve outcomes in diabetes management.Added value of this study Unlike most studies to date, which either assess health literacy in a unidimensional way or without considering interdimensional interactions, our approach integrated four HLQ scales to produce multidimensional profiles. These profiles, which would have been invisible had each domain been analyzed separately, capture both relatively homogeneous strengths-based patterns (eg, cluster M, with strong abilities across all four scales, favorable health outcomes and high SES) and heterogeneous combinations of strengths and challenges (eg, clusters A, D, K and L). In addition to mapping health literacy patterns, our cluster analysis enabled a nuanced description of each profile’s socioeconomic, clinical and behavioral characteristics, which in turn highlighted that specific combinations of health literacy abilities often coexist with particular social and health-related factors. While no causal inference can be drawn from this cross-sectional study, the observed associations are consistent with the existing literature, 8–10 12–15 and with other studies using similar approaches (ie, multidimensional Ophelia-based analysis of health literacy in various settings such as the general population,28 29 hospitals30 and persons with specific chronic conditions such as kidney disease31 32). However, the profiles differed across studies, highlighting the contextual nature of health literacy. The concept of health literacy should not be regarded as a discrete individual skill; rather it should be considered as a multifaceted construct that is profoundly influenced by the social, economic, cultural and organizational contexts in which individuals and communities live.7 33Strengths and limitations This study has several methodological strengths. First, the HLQ, a validated tool used in diverse settings, 34 guaranteed that we could make a reliable, multidimensional assessment of health literacy. Second, we applied a rigorous and transparent exploratory strategy: we used the Ophelia process to implement hierarchical clustering using Ward’s method. This identified five relevant clusters for guiding future interventions. We also assessed the distinctiveness of these clusters using PCA, confirming their separation and coherence in multidimensional space. Finally, the study leveraged a number of rich data sources: individual medicoadministrative records provided objective measures of health status and care use, while self-reported data offered complementary insights into participants’ SES, resources and perceived health.The study also has limitations. First, only four of the nine HLQ scales were used in Entred 3, limiting full conceptual coverage. Scientifically robust cluster analysis requires a well-grounded a priori understanding of the construct under investigation. The selected scales were chosen for their relevance to diabetes care,8 15 but others—such as feeling understood and supported by healthcare providers (HLQ1) and the ability to find accurate information (HLQ8)—were not included and may have strengthened the identified profiles. Additional or different profiles might have emerged had the complete HLQ been used. Nonetheless, the selected clusters appeared to be clinically and socially coherent. Although some dimensions were not explored, for example, social support (HLQ4), they were nonetheless reflected in other Entred 3 variables and informed the interpretation of the clusters. The internet-based questionnaire format was the only one which included all four HLQ dimensions. One might question whether relying solely on the online subsample—excluding paper and telephone respondents—may have led to additional health literacy profiles being missed. Individuals with lower health literacy or limited digital access may also have been less likely to participate in the online questionnaire, which could have influenced the cluster solution and comparisons. Although the online subgroup is not fully representative of the French population living with T2D (online supplemental table S4), 15% of online respondents reported completing the questionnaire with assistance, suggesting that individuals who might not usually be able to complete an online questionnaire independently were nevertheless included in the sample. Since cluster analysis can identify distinct profiles even when represented by relatively small numbers of individuals, this may have helped capture a broader range of health literacy profiles.More broadly, the limitation of using only four scales, instead of all nine, reflects the inherent tension between population-wide health surveys and targeted studies. While comprehensive instruments such as the HLQ offer valuable insights, it is often challenging to include them in their entirety in large-scale surveys due to respondent burden and feasibility issues. Our survey, which was designed primarily for national surveillance rather than clustering or intervention development, reflects this trade-off. However, the systematic development of interventions for diverse population groups would benefit from more granular data, as shown in recent national and comparative health literacy surveys.29 35 The Alisa-Diab study (JACARDI) aims to address this issue by qualitatively exploring all nine dimensions of the HLQ in people with T2D in two contrasting French settings, thus complementing our work.A second limitation is that the cross-sectional nature of the data does not allow for causal inference between health literacy profiles, SES and health outcomes. Moreover, the interpretation of the clusters is based on an exploratory classification method, which involves a degree of subjectivity in the choice of the number of clusters retained. This was partially addressed by relying on statistical criteria (Dunn index) and visual cluster separation inspection (PCA).Conclusion By identifying different health literacy profiles among a sample of individuals living with T2D in France, this study highlights the complexity and diversity of the health literacy skills and resources people possess. The profiles that emerged show that health literacy cannot be reduced to a single unidimensional score. Instead, it should be seen as a set of interrelated, specific strengths and challenges that are closely linked to individuals’ social contexts and health outcomes. Addressing these different profiles is crucial for developing effective targeted strategies to combat SIH. Our findings support the need for contextualized interventions tailored to real, diverse needs, which are codesigned with the people they are intended to serve.