Care-seeking among hepatocellular carcinoma patients in Uganda: health system navigation and its impact on diagnosis and survival in a prospective case-cohort study Authors: Sara K Nsibirwa, Katuku J Aizire, Fred Okuku, Emmanuel Ochola, Amir M Mohareb, Ponsiano Ocama, Gregory D Kirk Published: 2026 ABSTRACT AbstractBackgroundHepatocellular carcinoma (HCC) patients are often diagnosed with advanced-stage disease. We aimed to identify care-seeking patterns, assess correlates of earlier diagnosis and determine the impact of diagnostic delay on mortality among HCC patients.MethodsCross-sectional study, enrolled (March 2015 to August 2020), HCC patients in three Ugandan tertiary care hospitals. Time to HCC diagnosis was the duration between the onset of the first reported symptom(s) and ultrasound-based diagnosis. We used negative binomial regression to estimate incidence rate ratios (IRRs) for correlates of time to HCC diagnosis and Kaplan-Meier survival curves for time-to-death analysis.ResultsOf 651 HCC patients, 442 (67.9%) were male, median age 43 years (IQR 32–56). Majority 97.7% sought healthcare for HCC-related symptoms prior to enrolment; 248 (38.1%) once and 388 (59.6%) two or more times. These included public facilities (442, 69.4%), private facilities (493, 77.5%) and traditional and complementary medicine practitioners (140, 22%); 45 patients (6.9%) visited all three types at least once. Median time to HCC diagnosis was 90 days (IQR 58–173). Shorter time to HCC diagnosis was associated with male sex (IRR 0.84; 95% CI 0.71 to 0.99; p=0.04), chronic hepatitis B (IRR 0.78; 95% CI 0.66 to 0.93), HIV (IRR 0.64; 95% CI 0.51 to 0.79) and having at least secondary education (IRR 0.83; 95% CI 0.69 to 0.99). Longer time to HCC diagnosis was associated with age ≥50 years (IRR 1.29; 95% CI 1.02 to 1.64) and attending multiple facilities (IRR 2.20; 95% CI 1.54 to 3.00). Time to HCC diagnosis was not significantly associated with survival.ConclusionDiagnostic pathways for HCC in Uganda are characterised by repeated healthcare visits across multiple provider types. Strategies to improve outcomes should promote timely care-seeking, strengthen surveillance of high-risk individuals and improve diagnostic capacity and coordination of care within the pluralistic healthcare system.Keywords: HIV; Sociodemographic Factors; Confounding Factors, Epidemiologic WHAT IS ALREADY KNOWN ON THIS TOPIC Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death globally, with particularly poor outcomes in sub-Saharan Africa. Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death globally, with particularly poor outcomes in sub-Saharan Africa. WHAT THIS STUDY ADDS This study provides novel insights into the complex care-seeking journeys of HCC patients in Uganda, drawing attention to heterogenous patterns of care seeking from traditional medical facilities and non-traditional providers with multiple uncoordinated consultations leading to delays in diagnosis This study provides novel insights into the complex care-seeking journeys of HCC patients in Uganda, drawing attention to heterogenous patterns of care seeking from traditional medical facilities and non-traditional providers with multiple uncoordinated consultations leading to delays in diagnosis HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings emphasise that interventions to improve HCC outcomes need to target the pre-symptomatic periods of HCC disease progression. There is an urgent need for standardised HCC surveillance for high-risk individuals, enhancement of provider awareness and diagnostic capacity, integration of traditional and biomedical care pathways, and strengthening of referral systems to aid earlier detection and improve survival for HCC patients in Uganda. These findings emphasise that interventions to improve HCC outcomes need to target the pre-symptomatic periods of HCC disease progression. There is an urgent need for standardised HCC surveillance for high-risk individuals, enhancement of provider awareness and diagnostic capacity, integration of traditional and biomedical care pathways, and strengthening of referral systems to aid earlier detection and improve survival for HCC patients in Uganda. Introduction Hepatocellular carcinoma (HCC) is the most frequent primary malignancy of the liver and represents one of the few cancers with rising global incidence and mortality.1 HCC is the second leading cause of cancer death in high-incidence regions of East Asia and sub-Saharan Africa (SSA)2 with an estimated median survival time after HCC diagnosis of only 10 weeks.3 Clearly, HCC presents at advanced stages with limited effective treatment options due to failures in screening, early diagnosis and treatment.4 Healthcare delivery and care-seeking patterns in SSA are complex and not easily delineated. In Uganda, the healthcare ecosystem broadly comprises government-sanctioned conventional medical facilities and others recognised by the government as traditional and complementary medicine practitioners (TCMPs). Conventional facilities include public (government provider) and private health facilities with relatively equal contributions nationally.5 A tiered public facility model ranges from the community health centres (levels 1–4) to more specialised facilities in those serving larger populations, namely general, regional and national referral hospitals. Private facilities include ‘not-for-profit’ organisations and private health providers. Through public–private partnerships, a minimum package of health services delivery was developed for all levels of healthcare for both public and private facilities, including curative, preventive, rehabilitative and specialist care.6 While TCMPs are broadly recognised by government, this sector remains largely unregulated with many unregistered and without a functional relationship with the rest of the health sector. TCMPs typically emphasise herbal medicines, and in some instances, spiritual healers focus on incorporating beliefs into divine healing rituals and divinations. Due to the limited survival time and complexity of healthcare delivery, patterns of care-seeking among African HCC patients have not been well characterised previously. Studies of care-seeking among cancer patients in Uganda have largely focused on cervical and breast cancers.7,9 We are not aware of prior Ugandan or SSA studies that systematically investigated care-seeking patterns among HCC patients. The objective of this study was to describe the care-seeking patterns of HCC patients as they navigate the Ugandan healthcare system. Further, we assessed care-seeking in relation to presenting symptom clusters and to time to HCC diagnosis and the impact of diagnostic delay on mortality. Methods Study population HCC patients were recruited as part of the H2U Consortium multi-site, hospital-based case–control study, described in detail elsewhere.10 HCC patients were recruited between March 2015 and August 2020 from specialised-care facilities in Uganda, including the gastrointestinal and oncology units of Mulago National Referral Hospital and the Ugandan Cancer Institute, both in Kampala, and St. Mary’s Lacor Hospital in Gulu. Patients were deemed eligible if they were 12 years of age or older and spoke and comprehended English or the local language in which consenting materials were administered. All adult patients provided written informed consent; assent and parental/guardian consent were obtained prior to enrolling 12–17-year-old individuals. Study procedures Sociodemographic, care-seeking behaviour and clinical data were collected using standardised questionnaires and procedures adapted from prior HCC and liver disease studies in West Africa.11 12 Patients were asked to describe symptoms and signs attributed to HCC and when the onset of these symptoms first occurred. Patients were asked about seeking care at a conventional medical facility (public or private), or alternative care (traditional healers, herbalists or spiritual healers). In addition, we reviewed hospital medical records and abstracted pertinent data such as clinical and referral details. Laboratory testing was performed at a single laboratory (Infectious Diseases Institute; Kampala, Uganda) for hepatitis B virus (HBV) surface antigen, anti-HBV core antibodies, anti-HBV surface antibodies, anti-hepatitis C antibodies (anti-HCV), HIV antibody, alpha-fetoprotein, alanine aminotransferase, aspartate aminotransferase, total bilirubin, international normalised ratio and a complete blood cell count. Ultrasound examination was performed by trained radiographers/radiologists with data collected on a standardised ultrasound form. Study analysis Care-seeking encounters were enumerated and categorised as visits to public medical facilities, private medical facilities or to TCMPs, which included herbalists, traditional healers and religious healers.13 Symptoms were classified into clusters based on previous studies14,17 that showed common HCC symptom clusters can identify patients at risk for worse outcomes. Clusters included pain, fatigue, anorexia-cachexia, obstructive syndrome, gastrointestinal symptoms and miscellaneous symptoms. Time to HCC diagnosis was calculated as the number of days from the reported onset of the initial HCC-attributed symptoms to the date of initial ultrasound-based HCC diagnosis. Ultrasonographic diagnosis date was selected due to widespread availability of this diagnostic modality and its earlier utilisation compared with biochemical or pathological confirmation of HCC. A majority of HCC cases (443, 68%) had received an ultrasonographic HCC diagnosis prior to study enrolment (median time to HCC diagnosis of 88 (IQR, 51–154) days). For 208 (31.9%) cases, their initial ultrasonographic HCC diagnosis occurred at the study enrolment visit (median time to HCC diagnosis 120 (IQR, 60–180) days). Based on the median time to HCC diagnosis of ~90 days, time to HCC diagnosis was dichotomised into delayed or not; biological data suggest that the time period of 90 days approximates the estimated HCC tumour doubling time.18,20 The term diagnostic delay is used to describe the prolonged time to HCC diagnosis while early diagnosis is used to refer to detection of disease at a stage when curative treatment may still be possible. The definition of onset of symptoms, time to HCC diagnosis and the measurement approach were informed by recommendations from the Aarhus statement guidelines for early cancer diagnosis research21 to enhance methodological rigour. Pearson χ2 and Mann-Whitney tests were used to compare categorical and continuous variables, respectively. To identify correlates of time to HCC diagnosis, negative binomial regression was used to estimate the incidence rate ratio (IRR) and corresponding 95% CI. We assessed the following factors based in part on Andersen’s model of healthcare utilisation,22 sociodemographic: age, sex, marital status, residence, education, employment; behavioural: alcohol use, cigarette smoking; and laboratory: HBV, HCV and HIV infections. Variables of a priori importance, including age, sex and alcohol use and those with a significant statistical association in bivariable analyses were included in multivariable models. Kaplan-Meier survival curves depicted time to death following HCC diagnosis stratified by time to HCC diagnosis, early (less than or equal to 90 days) and delayed (greater than 90 days). All p values were based on a two-sided hypothesis test with a type-1 error (α=0.05). Stata V.17.0 (StataCorp. 2021, Stata Statistical Software: Release 17, College Station, Texas, USA) was used for statistical analysis. Patient and public involvement statement Patients and the public were not involved in the initial conception of this study; however, the research questions were informed by prior clinical and research experience with HCC patients in Uganda. Patient experiences shared during routine clinical care helped shape the focus on health-system navigation. Clinicians and community representatives contributed to the development of the data collection tools to ensure feasibility and minimise participant burden. Study findings have been shared with participating clinical sites and have informed creation of a national steering committee focused on interventions to improve HBV and HCC outcomes. Results Participant characteristics at study enrolment are summarised in table 1. Of 651 participants, the median age was 43 years (IQR 32–56), the majority were male (67.9%), married (69.6%), engaged in informal employment (74.5%) and 42.2% had attained a post-primary education. The prevalence of chronic viral infections was high, HBV (46.0%), HCV (10.4%) and HIV infection (15.9%). Table 1Characteristics of 651 Ugandan HCC patients at study enrolment, stratified by time to HCC diagnosisVariablen (%)Time to HCC diagnosis≤90 days (n=336)n (%)>90 days (n=315)n (%) Sociodemographic Age, median (IQR), years43 (32–56)42 (30–56)45 (33–56)Age ≤30146 (22.4)85 (25.3)61 (19.4) 31–49253 (38.9)129 (38.4)124 (39.4) ≥50252 (38.7)122 (36.3)130 (41.3)Male sex442 (67.9)248 (73.8)194 (61.6)Region of residence Central263 (40.5)139 (41.5)124 (39.4) Eastern71 (10.9)33 (9.9)38 (12.1) Northern238 (36.6)126 (37.6)112 (35.6) Western72 (11.1)33 (9.9)39 (12.4) Foreign6 (0.9)4 (1.2)2 (0.6)Marital status Married453 (69.6)234 (69.6)219 (69.5)Education status Primary or below376 (57.8)180 (53.6)196 (62.2) Secondary182 (27.9)102 (30.3)80 (25.4) Tertiary93 (14.3)54 (16.1)39 (12.4)Employment status Unemployed76 (11.7)37 (11.0)39 (12.4) Informal486 (74.7)254 (75.6)232 (73.7) Formal89 (13.7)45 (13.4)44 (13.9)Family history of HCC Definite27 (4.2)15 (4.5)12 (3.8) Probable28 (4.3)15 (4.5)13 (4.1) Don’t know39 (5.9)17 (5.1)22 (6.9) Behavioural factors Cigarette smoking136 (20.9)69 (20.5)67 (21.3) Alcohol use340 (52.2)170 (50.6)170 (53.9) Laboratory findings HBsAg, positive284 (45.7)159 (49.8)125 (41.4) Anti-HCV, positive64 (10.3)33 (10.3)31 (10.2) HIV, positive99 (15.8)53 (16.5)46 (15.0) Staging of disease Multifocal disease*470 (73.4)246 (75.7)224 (71.3)Care-seeking behaviour Visited public facilities only128 (19.7)68 (20.2)60 (19.1)Visited private facilities only171 (26.3)111 (33.0)60 (19.1)Visited TCMPs only2 (0.3)2 (0.6)0Public and private facilities259 (39.8)115 (34.2)144 (45.7)Public and TCMPs10 (1.5)5 (1.49)5 (1.59)Private and TCMPs18 (2.8)9 (2.7)9 (2.9)Visited all three types45 (6.9)16 (4.8)29 (9.2)Did not seek prior care18 (2.8)10 (2.9)8 (2.5)*Staging of disease as per abdominal ultrasound scan.anti-HCV, hepatitis C virus antibody; HBsAg, hepatitis B surface antigen; HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. Characteristics of 651 Ugandan HCC patients at study enrolment, stratified by time to HCC diagnosis Staging of disease as per abdominal ultrasound scan. anti-HCV, hepatitis C virus antibody; HBsAg, hepatitis B surface antigen; HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. Care-seeking patterns At enrolment, study participants reported the type and frequency of prior healthcare encounters. Figure 1 illustrates the pattern of care-seeking pathways by participants across public, private and traditional providers prior to diagnosis, demonstrating overlapping use of multiple healthcare facilities. Participants reported directed referrals from public (39%) or private (35%) facilities while 26% represented self-referrals to the tertiary referral centres where recruitment occurred. Overall, 171 (26.3%) reported previously seeking care at private facilities only, 128 (19.7%) at public facilities only and 2 (0.3%) from TCMPs only; 259 (39.8%) reported visiting both private and public facilities, while 45 (6.9%) reported visiting all three types of facilities and 18 (2.8%) reported no prior healthcare visits. A majority of HCC patients (388, 59.6%) reported seeking healthcare two or more times prior to diagnosis. Among those seeking care, 442 (67.9%) attended a public facility, 493 (75.7%) a private facility and 140 (21.5%) a traditional health facility at least once. Female participants (72/209 (34.4%)) were more likely than males (113/442 (25.6%)) to report multiple visits to public health facilities prior to diagnosis (p=0.001). Patterns of care-seeking by type and frequency of health facilities visited by HCC patients prior to study enrolment. Columns represent facility types (public, private and alternative (TCMPs)). Colours indicate visit frequency categorised as 0 (blue), 1 (orange) or≥2 visits (green). The alluvial plot illustrates the flow and frequency of patient visits across facility types and demonstrates overlapping use of services and fragmented care-seeking pathways. Encounters are non-sequential and do not reflect the order of care seeking. HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. Care-seeking patterns across different sociodemographic, behavioural and clinical characteristics of HCC patients are presented in online supplemental table 1. Patterns varied by age, sex, education and region of residence, with older individuals and those with higher levels of education more likely to seek care from multiple types of facilities. Notably, patients with HIV (48.5%) and HBV infections (38.4%) frequently accessed both public and private facilities. Across all sub-groups, multifocal disease was prevalent, with no major differences by care-seeking patterns. Reported HCC-associated symptom clusters and patterns of care-seeking are depicted in figure 2. Overall, there was a significant association between the type of HCC-associated symptom clusters reported and the type of health facility visited (p=0.012). Public and private facilities were the most frequently visited overall, with roughly equal proportions of patients across most symptom clusters. Pain was the predominant symptom cluster reported and among patients who reported pain, a larger proportion initially sought care at private facilities or with TCMPs. In contrast, patients who reported obstructive symptoms (such as jaundice or abdominal swelling) were more likely to have first presented to public health facilities than those reporting pain or other symptoms. Pattern of health facility visits by HCC patients. Horizontal bars show the number of patients visiting each type of facility, stratified by reported HCC-symptom cluster. HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. We also evaluated what services were provided during the reported 1454 healthcare encounters. Of the medical interventions documented, 1099 visits (76%) involved prescribing of medications, of which 25% were antimicrobial drugs, 20% were pain medications and 10% were anti-ulcer medicines. Diagnostic and therapeutic procedures, such as exploratory laparotomy, abdominal paracentesis and biopsies, accounted for only 2.3% of all reported interventions. When comparing public and private facilities, the difference in the proportion of HCC patients who underwent diagnostic (1.8% vs 2.2%; p=0.5) and therapeutic procedures (67.1% vs 71.3%; p=0.07) was negligible. The use of chemotherapy was reported with a similar distribution across facilities despite differences in resource availability or patient demographics. Time to HCC diagnosis The median time from reported symptom onset to ultrasound-based HCC diagnosis was 90 days (IQR, 58–173). Female patients experienced longer delays to diagnosis than males (median 119 vs 89 days, p=0.001). Furthermore, a higher proportion of late diagnosis occurred among patients who were HBV uninfected (p=0.004). Care-seeking patterns were also strongly associated with diagnostic delay; those who had sought care multiple times before study enrolment had the longest median time to HCC diagnosis (180 days), while those with no or only one prior visit had shorter intervals (52 and 89 days, respectively). HCC patients who attended only private facilities were more likely to receive early diagnosis, whereas those visiting multiple facility types, especially both public and private, experienced greater delays (p<0.001). Those seen exclusively in public facilities showed a more even distribution of time to HCC diagnosis (online supplemental table 2). Table 2 shows the association between socio-demographic, behavioural, clinical factors, care-seeking patterns and time to HCC diagnosis. On bivariable analysis, age of 50 years and above, alcohol use, residing within the Western region of Uganda and care-seeking patterns involving visiting both private and public facilities were associated with longer time to HCC diagnosis, while males, HBV and HIV infections were associated with a shorter duration to HCC diagnosis. On multivariate analysis, age of 50 years and older (IRR 1.29; 95% CI 1.02 to 1.64), HBV (IRR 0.78; 95% CI 0.66 to 0.93), HIV (IRR 0.64; 95% CI 0.51 to 0.79) infections, having a secondary education as highest level (IRR 0.83; 95% CI 0.68 to 0.99), residing in the western region (IRR 1.33; 95% CI 1.02 to 1.72), seeking care at public and private facilities (IRR 1.38; 95% CI 1.13 to 1.68) and seeking care at all types of facilities (IRR 2.16; 95% CI 1.54 to 3.04) were independently associated with time to HCC diagnosis. Table 2Incidence rate ratios (95% CI) estimates of time to HCC diagnosisVariable*†Bivariable analysisP valueMultivariable analysisP value Sociodemographic Age categories ≤30Ref 31–491.21 (0.98 to 1.48)0.151.23 (0.99 to 1.54)0.06 ≥501.56 (1.27 to 1.92) <0.001 1.29 (1.02 to 1.64) 0.04 Male sex0.84 (0.71 to 0.98) 0.04 0.89 (0.74 to 1.06)0.19Current region of residence CentralRef Eastern1.16 (0.89 to 1.51)0.291.14 (0.87 to 1.49)0.36 Northern1.07 (0.90 to 1.28)0.441.00 (0.83 to 1.21)0.97 Western1.49 (1.15 to 1.95) <0.001 1.33 (1.02 to 1.72) 0.04 Foreign0.67 (0.32 to 1.50)0.350.62 (0.26 to 1.48)0.28Education status Primary and belowRef Secondary0.83 (0.69 to 0.99) 0.04 0.83 (0.69 to 0.99) 0.05 Tertiary0.97 (0.77 to 1.22)0.790.97 (0.75 to 1.24)0.80Occupation UnemployedRef Formal1.03 (0.81 to 1.32)0.79–– Informal1.20 (0.88 to 1.65)0.24––Family hx of HCC NoRef Definite0.94 (0.64 to 1.39)0.74–– Probable1.18 (0.80 to 1.74)0.41–– Don’t know1.09 (0.78 to 1.53)0.61–– Behavioural Alcohol use, yes1.18 (1.01 to 1.38) 0.04 1.11 (0.94 to 1.31) 0.21 Cigarette smoking, yes1.11 (0.91 to 1.34)0.30–– Laboratory HBsAg, positive0.68 (0.58 to 0.79) 0.001 0.78 (0.66 to 0.93) 0.005 Anti-HCV, positive1.08 (0.83 to 1.40)0.59––HIV, positive0.63 (0.50 to 0.78) <0.001 0.64 (0.51 to 0.79) <0.001 Care-seeking Did not seek care1.07 (0.66 to 1.75)0.781.24 (0.76 to 2.04)0.39Visited public facilities only1.15 (0.91 to 1.45)0.231.20 (0.95 to 1.52)0.13Visited private facilities onlyRef–– Visited TCMPs only0.53 (0.13 to 2.17)0.380.69 (0.18 to 2.69)0.59Public and private facilities1.28 (1.06 to 1.56) 0.01 1.38 (1.13 to 1.68) 0.001 Public and TCMPs0.85 (0.45 to 1.62)0.631.09 (0.56 to 2.14)0.79Private and TCMPs1.12 (0.69 to 1.83)0.651.12 (0.69 to 1.80)0.65Visited all three types2.09 (1.50 to 2.91) <0.001 2.16 (1.54 to 3.04) <0.001 Highlighted p values are statistically significant.*Missing data <5% for all variables unless otherwise specified.†Variables included were based on Andersen’s behavioural model of health services utilisation.anti-HCV, hepatitis C virus antibody; HBsAg, hepatitis B surface antigen; HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. Incidence rate ratios (95% CI) estimates of time to HCC diagnosis Highlighted p values are statistically significant. Missing data <5% for all variables unless otherwise specified. Variables included were based on Andersen’s behavioural model of health services utilisation. anti-HCV, hepatitis C virus antibody; HBsAg, hepatitis B surface antigen; HCC, hepatocellular carcinoma; TCMPs, traditional and complementary medicine practitioners. Correlates with survival The overall median (IQR) survival after HCC diagnosis was 42 (20–106) days. When stratified by diagnostic delay or not, survival was similar; 46 (39–53) days versus 55 (43–62) days respectively, log-rank test p=0.24 (figure 3). There is no significant difference in the median survival times across the different care-seeking patterns, Kruskal-Wallis rank test p=0.61 (online supplemental table 3). Survival by time to hepatocellular carcinoma (HCC) diagnosis. Kaplan-Meier curves estimated the survival probability over time for HCC patients stratified by time to diagnosis (≤90 days vs >90 days). In the analysis censored at 100 days, there was no statistically significant difference between groups. Overall, with extended follow-up, the resulting p value was 0.24. Discussion Despite representing the third most common cause of cancer death worldwide, there is a remarkable paucity of data describing the healthcare-seeking experiences of patients with HCC in SSA. This study on the healthcare-seeking patterns of Ugandan HCC patients highlights the urgent need for improved interventions along the spectrum of HCC prevention and care; including provider and patient awareness of HCC, cancer interception strategies, appropriate screening of high-risk individuals, earlier HCC diagnosis and broader availability of therapeutic modalities. Our findings reinforce earlier reports that many patients with HCC in the region present late to specialist care and are diagnosed with advanced-stage disease, which collectively contribute to the poor overall prognosis. Only a few patients (<10%) presented with limited disease that was still amenable to curative therapy.23 And despite recent advances in treatment of advanced HCC and palliative care being available, the short time intervals between diagnosis and death provided a limited opportunity for intervention. Studies looking at the factors affecting care-seeking among patients of other cancer types in SSA24,27 have shown varying results and it is therefore important to explore the reasons for this care-seeking behaviour28 and address the unique factors related to HCC in any attempt to improve disease outcomes. Evidence from cancer-screening literature has applied theoretical models of behaviour change and suggests that patients’ poor knowledge, negative perceptions or perceived barriers to cancer screening and treatment may adversely influence care-seeking behaviour. For those patients who did seek healthcare, there are heterogenous patterns of seeking care seen, similar to many SSA countries.4 28 Although conventional health services are available, up to 12% of our participants reported seeking care from TCMPs at least once. Traditional medicine is typically viewed as widely accessible and relatively inexpensive. It has been reported that over 50% of the Ugandan population seek care from TCMPs before and after visiting the conventional health facilities.6 TCMPs are available in both rural and urban areas; within easy reach of communities; providing cheaper and readily accessible services that are not regulated nationally. Most TCMPs have no functional relationship with public and private health providers and as a result, this may lead to delayed referrals, poor management and high morbidities and mortalities. Collaboration of biomedical practitioners with TCMPs by educating them on signs and symptoms of HCC can be undertaken as a strategy to improve quick and earlier referrals of patients to specialist care.29 The healthcare system operates on a referral basis, with lower-level facilities directing patients needing more advanced care to higher-level facilities. Within the public health delivery system, specialist-level care is provided only at the regional and national referral hospitals while within the private facilities, specialist services are usually offered only within urban areas.5 However, patients may also visit a specialist physician or medical centre directly. A large proportion (86%) of Ugandan households are within 5 km from a health facility, with services in public facilities provided by the government at no cost; however, drug stock-outs and the need for specialised tests lead to patients paying for their care out-of-pocket. In our study, patients residing in the Western region experienced longer diagnostic intervals. This may reflect the diagnostic capacity of proximal lower-level facilities which provide primary care but lack imaging, laboratory or specialist expertise required to confirm HCC. Consequently, patients may require multiple referrals across different levels of the health system before a definitive diagnosis is established, prolonging diagnostic pathways despite apparent geographic access to care. This highlights how both health-system constraints and social determinants may shape diagnostic pathways and contribute to delays in the diagnosis of HCC.30 However, limited investments could potentially remedy this gap. Based on our analysis of HCC diagnostic criteria, we advocate that quality ultrasound services could be readily available across the health system and would greatly advance pragmatic HCC diagnosis.10 Furthermore, multiple visits to different types of providers were made by patients without clear diagnosis or referral being made, alluding to the inefficiency of the pluralistic healthcare delivery system. Approximately 40% of patients reported multiple healthcare visits prior to receiving a definitive diagnosis. This pattern likely reflects system-level challenges in addition to patient-level delays, including inadequate suspicion for HCC by providers, limited diagnostic capacity at lower-level facilities and fragmented referral pathways. These repeated consultations not only impose a higher financial burden on patients and the health system but also prolong the time to definitive diagnosis reduce the opportunity for curative treatment for a disease with a very poor prognosis.31 Even in the instances where a diagnosis had been made, patients reported visiting other facilities before seeking specialist care; which contributed to the late presentation. This also implies an unclear mechanism in transitioning across the levels of healthcare. We observed that female patients experienced significantly longer diagnostic intervals despite reporting multiple healthcare visits, particularly to public healthcare facilities. This pattern may reflect socio-cultural and structural barriers affecting women’s access to specialised care, including financial constraints and reduced autonomy in health-related decision making. Similar gender-related barriers to timely cancer diagnosis have been described in other low- and medium-income settings and may contribute to prolonged diagnostic pathways.32 33 Although services in public facilities are intended to be provided at no cost, patients are frequently required to pay for diagnostic investigations or purchase essential medications, which may contribute to delays in diagnosis. In our cohort, informal employment was common, suggesting that many patients may lack stable income or formal health coverage, which would make healthcare costs financially draining. Financial constraints may influence initial provider choice, with patients opting for lower-cost or more accessible providers, including TCMPs, before seeking care at facilities capable of diagnosing HCC. Such patterns may also contribute to repeated healthcare visits and fragmented care pathways. The absence of an integrated national health insurance system in Uganda likely amplifies these challenges and is similar to what has been seen among uninsured populations globally.34 35 Accurate identification of patients who are at high risk for HCC is a key step for delivery of prevention and earlier diagnosis programmes.36 Globally, surveillance programmes have been particularly encouraged in the high-risk group patients and have led to an increasing number of earlier detections of HCC37 38; this has also proven effective in our population which has shown that high-risk populations, such as those with HBV infection and HIV infection had lower incidence rates of diagnostic delays. However, HCC surveillance rates remain low even among high-risk persons despite stipulated clinical guidelines of many countries in the region.39 In Uganda, HCC screening is recommended biannually for people living with HBV.40 Individuals aged 50 years and above experienced a longer time to HCC diagnosis, suggesting potential delays in detection within this age group. This highlights the need for improved screening and early diagnostic strategies tailored to older populations to facilitate timely diagnosis and intervention. HCC is asymptomatic in the early stages and may be difficult to diagnose in a timely fashion. When symptoms and signs do appear, over 90% present with a triad of abdominal pain, anorexia and hepatomegaly41 at which point disease is likely to be advanced. However, even as symptoms appear, these tend to be of a wide variety and overlapping which can present diagnostic difficulty at first presentation leading to multiple consultations with investigations into each symptom and thus delayed referral as seen elsewhere in SSA.23 42 43 Healthcare providers may lack current knowledge about HCC surveillance and diagnostic guidelines.44,46 Therefore, interventions that will successfully improve mortality from HCC will have to target pre-symptomatic periods as opposed to waiting for patients to develop symptoms and these should be targeted to improve care-seeking behaviour including health education of patients about HCC and associated symptoms, standardised educational training for healthcare providers, improved access to and quality of diagnostics to ensure timely diagnostic follow-up. Robust referral mechanisms should be established at primary health level involving both public and private health facilities and a population cancer registry would be beneficial in monitoring and evaluation. Similar to a study done in the USA,47 we did not observe any significant difference in overall survival when stratified by duration to diagnosis. It is likely that the patients with delayed diagnosis were less sick or had slower-growing tumours leading to a comparable 1-year survival as those who had an earlier diagnosis. Aggressive tumours could lead to advanced disease or death before a definitive HCC diagnosis can be made unlike slow-growing tumours, which can allow for repeated diagnostic testing. However, this finding should be interpreted with consideration of the waiting time paradox, where patients with more aggressive disease are diagnosed sooner but have poorer outcomes. This may conceal true associations between time to HCC diagnosis and prognosis. While sensitivity analysis by multifocality did not show any difference in care-seeking patterns or prognosis, residual confounding by disease severity is possible, and this justifies the need for future studies using analytical methods that are adjusted by more refined disease severity assessment. Our study has provided a descriptive analysis of care-seeking behaviour in HCC and its impact on diagnosis in SSA. A major strength of this study was the availability of rigorously collected information on initial symptoms and the time from initial symptoms to diagnosis in a large cohort of patients with HCC. However, our results should be considered in light of the limitations of this research. We relied on patient recall for the information related to initial symptoms and care-seeking behaviour. To improve the accuracy of key clinical timelines, patient-reported information was supplemented where possible with data obtained from medical records and chart extraction documenting diagnosis and presentation of HCC. However, medical records for these Ugandan patients were extremely limited and would almost never refer to the timing of onset of symptomatology. Similarly, characterisation of the types of traditional providers is based on the patient’s perception. These self-reported data may be subject to inaccuracies. Furthermore, we did not collect contextual information (eg, family preferences, religious beliefs) or individual factors (eg, perceptions of illness, trust in biomedical care) that might help explain the reasons for the care-seeking behaviour exhibited and there was also limited information in the medical records to explain reasons for delays in referral and diagnosis. These gaps highlight the need for further qualitative and longitudinal investigation to fully explore the drivers of diagnostic delay and better understand the decision-making processes behind HCC care-seeking behaviour. Conclusion Understanding how and when patients with HCC seek care is essential for designing effective interventions that promote earlier diagnosis and improve outcomes. Insights from this study can inform strategies to strengthen public awareness, provider education and the capacity of healthcare systems to recognise and manage liver disease promptly. Addressing both patient- and system-level barriers to timely diagnosis will be critical to reducing HCC-related mortality in SSA. Earlier diagnosis of HCC remains a key goal in improving the poor prognosis associated with this cancer. Our findings highlight substantial delays along the diagnostic pathway, underscoring the need to strengthen both provider and community awareness and the efficiency of diagnostic and referral processes. Strengthening the surveillance of high-risk individuals as well as improving access to affordable healthcare through a national health insurance scheme could improve opportunities for earlier detection and treatment as a public health strategy to improve disease outcomes. Further studies are warranted to characterise the reasons and improve understanding of the current barriers and delays along the healthcare pathway for HCC cases, if a reduction in HCC mortality is to be achieved. Pluralism in the healthcare delivery system will be beneficial only if we manage to incorporate and regulate similar standards of diagnostics, treatment and referral across the health system, supported by a vigorous monitoring and evaluation system at all levels to identify gaps in early diagnosis. Supplementary material Acknowledgements We thank the study participants and their caregiving families who agreed to participate in this study, and the study teams in Kampala and Gulu who have shown great dedication and resilience over the years. We wish to acknowledge the support from the Infectious Diseases Institute in Kampala, Uganda Cancer Institute, Mulago National Referral Hospital and St. Mary’s Hospital, Lacor, Uganda. Footnotes Funding: This work was supported by grant U54-CA190165 from the National Cancer Institute. GDK was supported by grant K24-AI118591 from the National Institute of Allergy and Infectious 18 Diseases. AMM is supported by K01AI166126 from the National Institute of Allergy and Infectious Diseases at the NIH. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants and was approved by Makerere University School of Medicine (IRB approval number 2014-117); Johns Hopkins Medicine (IRB: approval number IRB00044309); Uganda National Council for Science and Technology. Participants gave informed consent to participate in the study before taking part. Data availability free text: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. 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. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.