Portal vein thrombosis and associated factors among patients admitted to Tibebe Ghion hospital, Bahir Dar, Ethiopia: a cross-sectional study, 2021–2024

Abstract

Background

Portal vein thrombosis (PVT) is a Liver vascular disease characterized by the development of a blood clot inside the portal vein’s main branches or trunk. There are limited studies on PVT and associated factors in Africa, with no studies from Ethiopia. Therefore, this study was conducted to assess the prevalence of portal vein thrombosis and associated factors among patients at the TGSH medical ward in Bahir Dar, Ethiopia, from January 1, 2021, to December 31, 2024.

Method

A retrospective cross-sectional study was conducted involving 422 patient charts selected through simple random sampling. Data were collected using structured tools, entered into EpiData v4.6, and analyzed using SPSS v25. Bivariate and multivariable logistic regression analyses were performed to identify factors associated with PVT, with statistical significance set at p < 0.05. Ethical clearance was obtained from the Institutional Review Board of Bahir Dar University.

Result

A total of 407 study subjects were included, and 64 (15.7%) had portal vein thrombosis. Having a diagnosis of chronic liver disease(CLD), the AOR is 2.139 (95% CI: 1.017–4.499) with a p-value of 0.045, patients with hepatocellular carcinoma (HCC) with an AOR of 3.912 (95% CI: 1.609–9.512) and a significant p-value of 0 003, Platelet count > 450 K, with an AOR of 4.574 (95% CI: 1.989–10.519) and a p-value of 0.001, were significantly associated with portal vein thrombosis.

Conclusion and recommendations

This study found a notably high prevalence of portal vein thrombosis (15.7%) among hospitalized patients, while chronic liver disease, hepatocellular carcinoma, and elevated platelets significantly increasing risk. These findings underscore the need for heightened clinical vigilance and targeted screening in high-risk populations. Future research should focus on elucidating underlying causal mechanisms, including potential thrombophilic conditions, and should incorporate more diverse and representative patient cohorts to inform comprehensive preventive and therapeutic strategies.

Introduction

Portal vein thrombosis (PVT) is a clinically significant hepatic vascular disorder characterized by thrombus formation within the portal vein or its major branches. As the portal vein supplies approximately 75% of hepatic blood flow, its obstruction can critically impair liver function and exacerbate underlying hepatic pathology.

PVT occurs in both cirrhotic and non-cirrhotic populations and is associated with a variety of local and systemic risk factors. In cirrhosis, hepatocellular carcinoma (HCC) and portal hypertension are predominant contributors, whereas in non-cirrhotic patients, malignancies, inflammatory abdominal conditions, and systemic prothrombotic states—such as inherited thrombophilias and myeloproliferative disorders—play key roles [1–4]. Although often asymptomatic, PVT may manifest as abdominal pain, gastrointestinal bleeding, or ascites. Diagnosis typically relies on Doppler ultrasonography, with contrast-enhanced CT or MRI providing greater anatomical detail [2, 5–8].

The global prevalence of PVT varies widely depending on population and diagnostic criteria, ranging from 5 to 64% in cirrhotic patients and up to 50% in those with HCC [9–13]. Importantly, PVT significantly worsens clinical outcomes, increasing the risk of gastrointestinal bleeding, hepatic encephalopathy, and mortality. Mortality rates in affected individuals vary between 7% and 50%, influenced by comorbid liver disease, malignancy status, and timeliness of intervention. Additionally, the involvement of mesenteric veins may lead to ischemic complications, further compounding the clinical burden [14–16]. Despite its clinical importance, epidemiological data on PVT remain limited in low-resource settings. In particular, there is a dearth of evidence from African nations—including Ethiopia—where the burden of chronic liver disease is increasing due to viral hepatitis and limited access to preventive care [9, 11, 17, 18]. The limited local data hinder early detection, risk stratification, and timely intervention.

Efforts to reduce the burden of venous thromboembolism (VTE), including PVT, in Ethiopia have focused largely on general DVT prevention protocols supported by governmental and non-governmental stakeholders. These include clinical guidelines for thromboprophylaxis in hospitalized patients, capacity building for imaging diagnostics, and improved access to anticoagulant therapy. However, these interventions have yet to be tailored specifically to PVT and its hepatic risk spectrum.

Despite global advances, PVT remains underdiagnosed and undertreated in Ethiopia. To address this gap, the present study aimed to assess the prevalence of PVT and identify associated clinical and demographic risk factors among patients admitted to the the medical ward of Tibebe Ghion Specialized Hospital (TGSH), Bahir Dar. We hypothesize that the prevalence of PVT in this setting is significant and influenced by key hepatic comorbidities such as CLD and HCC. The findings are intended to inform clinical decision-making and stimulate further research in similar low-resource environments.

Methods

Study Design and Setting

A retrospective cross-sectional study was conducted at Tibebe Ghion Specialized Hospital (TGSH), a tertiary teaching hospital with 450 beds providing inpatient and outpatient services in Bahir Dar, Ethiopia, covering admissions from January 1, 2021, to December 31, 2024.

Study Population

The source population consisted of all patients admitted to the TGSH medical ward during the study period. The study population included all adult patients admitted to the medical ward of the TGSH medical ward between January 1, 2021, and December 31, 2024.

Inclusion criteria

All adult patients (age ≥ 18 years) admitted to the TGSH medical ward within the study period.

Exclusion criteria and handling of missing data

Patients with medical records missing critical clinical, laboratory, or imaging information essential for confirming a diagnosis of portal vein thrombosis (PVT) were excluded from the analysis. Specifically, 15 records (4.3% of the initial dataset) were removed due to the absence of indispensable diagnostic or demographic variables, such as Doppler ultrasound confirmation or baseline liver function data.

For variables with partial missingness that did not meet the exclusion threshold, missing data were assumed to be Missing at Random (MAR) based on the absence of systematic patterns when cross-checked against observed covariates. The variables with incomplete data included platelet count (2.1% missing), serum albumin (1.8%), INR (1.6%), and Child–Pugh score components (< 1.5% each). To minimize loss of statistical power, person mean substitution was applied.

At the time of analysis, a sensitivity assessment was undertaken to evaluate the potential impact of imputation. The primary analyses were repeated using both (i) complete-case data and (ii) the imputed dataset. Differences in adjusted odds ratios between the two approaches were minimal (absolute variation ≤ 0.04), with no change in the direction or statistical significance of the main findings. These results indicated that the imputation method had negligible influence on the study conclusions. Although the original sensitivity table cannot be reproduced due to the time elapsed since data collection, the stability of the estimates across methods supports the robustness of our findings.

We acknowledge, however, that person mean substitution can reduce variance and potentially attenuate associations. Future studies in this field would benefit from prospectively planned multiple imputation or Bayesian approaches to optimize handling of missing data.

Sample Size Determination and Sampling Procedure

The sample size was calculated using the single population proportion formula, assuming a prevalence of portal vein thrombosis (PVT) of 50% due to a lack of prior local studies. With a 95% confidence level and 5% margin of error:

n = (Zα/2)² × p(1 - p)/w²Where;

n = minimum sample size.

Z(α⁄2) = The desired level of statistical significance (1.96).

p = estimated prevalence of portal vein thrombosis (assumed to be 0.5).

p(1-p) = A measure of variability.

W = the margin of error = 5%.

Given the absence of prior studies on PVT prevalence in the local setting, a conservative estimate of 50% was used to maximize sample size and ensure adequate statistical power. Substituting the values:

n = (1.96)² × 0.5(1–0.5)/(0.05)² = 384.

To account for potential incomplete records or non-response, a 10% contingency was added, resulting in a final sample size of 422 records. A simple random sampling technique was applied to select medical records from the hospital’s admission registry using a computer-generated random number list. In cases where a selected record was incomplete or missing key variables, the next eligible record was included (Fig. 1).

Fig. 1.

Flow diagram illustrating the patient selection process for the study titled: “Portal Vein Thrombosis and Associated Factors among Patients Admitted to Tibebe Ghion Hospital, Bahir Dar, Ethiopia: A Cross-Sectional Study, 2025

Variables of the Study

Dependent variable

Presence of Portal vein thrombosis (yes/no).

Independent variables

Sociodemographic characteristics (age, sex, residence, education level).

Clinical conditions (cirrhosis, malignancy, myeloproliferative disorder, coagulation abnormalities).

Local risk factors (abdominal surgery, intra-abdominal infection)

Operational definitions

PVT: Presence of thrombus in the portal vein or its branches, confirmed by imaging (Doppler ultrasound(US), contrast CT, or MRI) [5].

Ultrasound with Doppler: Absence of flow or echogenic material in the portal vein.

Contrast-enhanced CT or MRI: Intraluminal filling defects in the portal vein or its branches, with or without cavernous transformation. All imaging studies were conducted or validated by senior radiologists during hospitalization, with no significant discrepancies observed in PVT diagnoses among the various modalities employed. Notably, these imaging techniques demonstrate high sensitivity and specificity for diagnosing PVT, even in resource-limited settings, with research indicating that each modality reliably detects PVT while offering distinct advantages based on clinical circumstances [19, 20].

Cirrhosis: Chronic liver damage marked by fibrosis and nodular regeneration, diagnosed with the use of clinical, laboratory, and imaging findings [21].

Malignancy: Includes both primary hepatobiliary cancers and liver metastases [22].

Myeloproliferative disorder: A group of diseases causing abnormal blood cell production [23].

Coagulation abnormality: Genetic or acquired conditions that predispose to excessive clot formation [24].

Abdominal surgery: Any surgical intervention involving the abdominal cavity.

Intra-abdominal infection: Infections originating within the gastrointestinal tract or abdominal cavity.

Data collection procedure

We retrieved medical records from the hospital archive following the random sampling procedure. Trained data collectors (three medical interns and two nurses) used a structured and pretested checklist to extract data on sociodemographic characteristics, comorbidities, and clinical, laboratory, and imaging findings. To minimize the risk of extraction bias, data collectors were blinded to the specific study hypothesis and research objectives. All records were documented in English, eliminating the need for translation. The principal investigator supervised the entire data collection process.

Data quality assurance

A structured data extraction checklist was developed based on standardized chart review tools and relevant literature, and its content validity was confirmed by expert review. To assess clarity and feasibility, a pretest was conducted on 22 randomly selected records (5% of the sample), leading to minor revisions. We evaluated inter-rater reliability by having two independent data collectors extract data from 10% of the records, yielding a Cohen’s Kappa value > 0.80, indicating strong agreement. The principal investigator reviewed completed forms daily to ensure accuracy, consistency, and completeness, with discrepancies and missing data addressed promptly.

Data management and analysis

Collected data were cleaned, coded, and entered into EpiData version 4.6 and exported to SPSS version 25 for analysis. Descriptive statistics (frequencies, means, and proportions) were used to summarize variables. Bivariate analysis was first performed to identify candidate variables (p < 0.25), followed by multivariable logistic regression to determine independent predictors of PVT. Model fitness was assessed using the Hosmer–Lemeshow test. A p-value of < 0.05 and a 95% confidence interval were used to declare statistical significance. Crude and adjusted odds ratios (COR, AOR) were calculated to estimate the strength of associations. The result of the analysis was presented using texts and tables.

Results

Socio-demographic and clinical characteristics of study participants

A total of 407 patients were included in the final analysis, yielding a response rate of 96.4% from the initially targeted sample size of 422. Among these, 64 patients were diagnosed with portal vein thrombosis (PVT), corresponding to a prevalence of 15.7% (95% CI: 12.2–19.3%). The median age of participants was 38 years (IQR: 28–49 years), with an age range of 18 to 85 years, consistent with the inclusion criteria. The majority (44.5%, n = 181) were aged between 21 and 40 years. Males accounted for 68.3% (n = 278), giving a male-to-female ratio of 2.12:1 (Table 1).

Table 1.

Socio-demographic and clinical characteristics of adult patients (n = 407) admitted to TGSH medical ward, Bahir dar, amhara, ethiopia, 2021–2024

Characteristics		N = 407(100%)
Age (years)	≤ 20	20 (4.9%)
	21–40	181(44.5%)
	41–60	137 (33.7%)
	> 60	69 (16.9%)
Sex	Male	278 (68.3%)
	Female	129 (31.7%)
Residency	Urban	135 (33.2%)
	Rural	272 (66.8%)
Admission Diagnosis	CLD	199 (48.9%)
	HCC	37 (9.1%)
	Pancreatitis	8 (2%)
	GI TB	37 (9.1%)
	IBD	22 (5.4%)
	Others	143(35.1%)
Comorbidities	Renal Disease	10(2.5%)
	Diabetes Mellitus	9(2.2%)
	Hypertension	9(2.2%)
	Respiratory disorders	5(1.2%)
	Cardiac diseases	2(0.5%)
Biochemical tests	INR(mean)	1.4
	Bilirubin(mean)	1.2 mg/dL
	Albumin(mean)	2.4 mg/dL
	HGB(mean)	11 mg/dL
	AST(mean)	48IU/L
	ALT(mean)	39IU/L
	Platelet count(mean)	168
	Sodium(mean)	137
	Random Blood sugar(mean)	105

At least one comorbidity was identified in 8.6% (n = 35) of participants. The most commonly reported comorbid conditions included renal diseases (2.5%, n = 10), diabetes mellitus (2.2%), hypertension (2.2%, n = 9), and respiratory disorders (1.2%, n = 5). Cardiac abnormalities were observed in 0.5% (n = 2) of patients, while no cases of rheumatologic disorders were documented in the study population. (Table 1).

Clinical characteristics of patients with PV

Among the 64 patients diagnosed with portal vein thrombosis (PVT), the most frequently reported clinical symptoms were abdominal pain, observed in 81.3% of cases (n = 52), and gastrointestinal bleeding, reported in 18.8% (n = 12). On physical examination, ascites was present in 48.4% of patients (n = 31), while splenomegaly was identified in 26.6% (n = 17).

Abdominal ultrasonography was the initial diagnostic modality in all patients, confirming PVT in 54 individuals (84.4%). Computed tomography (CT) scans were performed in 44 patients (68.8%) either as a primary diagnostic tool or as a follow-up to ultrasonography. In total, 10 patients (15.6%) were diagnosed with PVT based on CT findings alone. On Imaging, complete thrombosis was the most frequent finding, observed in 36 patients (56.3%), while partial obstruction was seen in only 3 patients (4.7%). Portal cavernoma was present in 17 patients (26.6%), suggesting chronic thrombosis. Splenic varices and superior mesenteric vein (SMV) extension were each identified in 10 patients (15.6%), and splenic vein thrombosis was found in 8 patients (12.5%). Upper gastrointestinal (GI) endoscopy was performed in 25 patients (39.1%). Among these, esophageal varices were detected in 20 patients (31.3%), portal hypertensive gastropathy in 5 patients (7.8%), and normal endoscopic findings in 3 patients (4.7%) (Table 2). These findings support the presence of clinically significant portal hypertension in a subset of patients with PVT.

Table 2.

Clinical characteristics of the patients with PVT (n = 64) admitted to TGSH medical ward, Bahir dar, amhara, ethiopia, 2021–2024

Characters		N = 64(100%)
Diagnostic modality used	Ultrasound	54 (84.4%)
	CT scan	44 (68.8%)
	Both	34 (53.1%)
Imaging Report	Complete thrombosis	36 (56.3%)
	Partial obstruction	3 (4.7%)
	Portal cavernoma	17 (26.6%)
	Splenic varices	10 (15.6%)
	Superior mesenteric vein extension	10 (15.6%)
	Splenic vein thrombosis	8 (12.5%)
Primary cause	CLD	45 (70.3%)
	Malignancy	21 (32.8%)	HCC	18 (28.1%)
			Other malignancies	3 (4.7%)
	GI TB	2 (3.1%)
	Pancreatitis	2 (3.1%)
	Abdominal surgery	3 (4.7%)
	Abdominal infection	3 (4.7%)
	Cause not identified	8(12.5%)
Presenting symptoms and signs	Abdominal pain	52 (81.3%)
	Splenomegaly	17 (26.6%)
	Jaundice	11 (17.2%)
	Fever	11 (17.2%)
	Digestive bleeding	12 (18.8%)
	Ascites	31 (48.4%)

There was no discernible relationship between portal vein thrombosis and clinical features, age, residency, pancreatitis, or test results, according to the study. Hepatitis virus-associated cirrhosis and cirrhosis severity (Child-Pugh class) were found to be significantly correlated (Table 3).

Table 3.

Correlation between different factors and portal vein thrombosis (n = 64) admitted to TGSH medical ward, Bahir dar, amhara, ethiopia, 2021–2024

Characters		PVT	P value
Age mean age (years)		41.48	0.446
Residency	Urban	19	0.519
	Rural	45
Pancreatitis		2	0.467
Others		17	0.118
Cause of cirrhosis	Alcohol	2	0.205
	Viral hepatitis	17	0.034
	Auto-immune	1	0.344
Child class	A	0
	B	7	0.001
	C	20	0.001
Complications	Upper GI Bleeding	11	0.302
	Encephalopathy	8	0.27
	Spontaneous Bacterial Peritonitis	7	0.716
	Hepato-Renal Syndrome	2	0.317
	Esophageal varices	20	0.329
Laboratory value	INR (mean)	1.42	0.073
	Bilirubin (mean) (mg/dL)	2.61	0.528
	ALT(mean)	89.26	0.103
	Sodium (mean)	135.15	0.317
	Random Blood Sugar(mean)	101.62	0.2

Factors associated with portal vein thrombosis

Binary logistic regression was initially employed to assess the association between various sociodemographic and clinical variables and the occurrence of portal vein thrombosis (PVT). Variables with a p-value < 0.25 in the bivariate analysis were selected as candidates for multivariable logistic regression. These included: sex, chronic liver disease (CLD), hepatocellular carcinoma (HCC), gastrointestinal tuberculosis (GI TB), platelet count (PLT), hemoglobin level (HGB), and aspartate aminotransferase (AST). Model adequacy was confirmed using the Hosmer–Lemeshow goodness-of-fit test (p = 0.175), indicating a good fit.

Multivariable logistic regression analysis revealed that male sex was significantly associated with lower odds of PVT. Males had 53.3% reduced odds compared to females (AOR = 0.467; 95% CI: 0.22–0.992; p = 0.048). Patients diagnosed with CLD had significantly higher odds of developing PVT compared to those without CLD (AOR = 2.139; 95% CI: 1.017–4.499; p = 0.045). Although HCC was found to be a strong independent predictor of PVT, the wide confidence interval (AOR = 3.912; 95% CI: 1.609–9.512; p = 0.003) suggests limited statistical precision, likely due to the small number of patients with both HCC and PVT. This limitation restricts the robustness of the estimate, and the findings should be interpreted cautiously.

Elevated platelet count (> 450,000/µL) was significantly associated with PVT, with patients having more than four times higher odds compared to those with normal platelet levels (AOR = 4.574; 95% CI: 1.989–10.519; p < 0.001).

No statistically significant association was observed between GI TB and PVT (AOR = 0.432; 95% CI: 0.092–2.018; p = 0.286).

Elevated AST levels did not demonstrate a statistically significant association with PVT across all categories (p > 0.05). Although patients with AST > 300 IU/L had an AOR of 1.51, this was not statistically significant (95% CI: 0.505–4.521; p = 0.461).

Similarly, hemoglobin levels were not significantly associated with PVT (AOR = 2.992; 95% CI: 0.696–12.865; p = 0.141) (Table 4). Some variables showed wide confidence intervals, reflecting reduced precision. This is likely due to smaller subgroup sizes.

Table 4.

Factors associated with portal vein thrombosis among patients (n = 407) admitted to TGSH medical ward, Bahir dar, ethiopia, 2021–2024

Variable	Category	PVT		COR (95% CI)	AOR (95%CI)	P value
		Yes	No
Sex	Male	53	225	0.396 (1.99 − 0.786)	0.467 (0.22–0.992)	0.048
	Female	11	118	1	1
Diagnosis of CLD	Yes	45	154	2.907 (1.633–5.178)	2.139 (1.017–4.499)	0.045
	No	19	189	1	1
HCC	Yes	18	19	6.673 (3.265–13.638)	3.912 (1.609–9.512)	0.003
	No	46	324	1	1
GI TB	Yes	2	35	0.284 (0.067–1.211)	0.432 (0.092–2.018)	0.286
	No	62	308	1	1
Platelet	< 150 × 10³/µL	24	156	1	1
	150-450 × 10³/µL	22	154	0.929 (0.5-1.726)	0.987 (0.476–2.046)	0.972
	> 450 × 10³/µL	18	33	3.545 (1.73–7.264)	4.574 (1.989–10.519)	< 0.001
AST	< 40IU/L	23	142	1	1
	> 40-100IU/L	21	126	1.029 (0.543–1.948)	0.665 (0.323–1.369)	0.268
	> 100-300IU/L	11	55	1.237 (0.564–2.702)	0.771 (0.308–1.93)	0.579
	> 300IU/L	9	19	2.924 (1.18–7.245)	1.51 (0.505–4.521)	0.461
Hemoglobin	< 12 mg/dL	32	218	1	1
	12-16 mg/dL	28	117	1.63 (0.36–2.839)	1.366 (0.713–2.618)	0.347
	> 16 mg/dL	4	8	3.406 (0.97-11.964)	2.992 (0.696–12.865)	0.141

Discussion

This study investigated the prevalence and associated factors of portal vein thrombosis (PVT) among hospitalized patients in a tertiary care setting in Ethiopia. The observed prevalence of 15.7% is substantially higher than rates reported in similar studies, such as the 1.9% in Senegal. This discrepancy may be attributed to differences in study populations, as our cohort included high-risk individuals, notably those with chronic liver disease (CLD) and hepatocellular carcinoma (HCC), who are known to have a higher risk of PVT. A significant proportion (22.6%) of CLD patients in our study had PVT, reinforcing established evidence of the association between cirrhosis and thrombotic events. Comparable prevalence rates of PVT among cirrhotic patients were reported in Malmö (28%) and JIPMER (17.2%), suggesting a consistent trend globally. The mechanisms underlying this association likely involve portal hypertension, venous stasis, and a prothrombotic shift in cirrhosis-related hemostasis [13, 25, 26].

In our cohort, males constituted the majority (82.8%) of patients, consistent with reports from Qatar, Mexico, and Italy, likely reflecting the higher prevalence of chronic liver disease (CLD) and hepatocellular carcinoma (HCC) among men in our setting. However, despite this male predominance, multivariable analysis revealed that female sex was independently associated with higher odds of portal vein thrombosis (PVT). At first glance, this appears counterintuitive, as male sex is often linked with an increased risk of venous thromboembolism in general populations. Yet, sex-related differences in thrombosis risk are complex and appear to be highly context-dependent, influenced by the nature of the underlying disease, comorbidities, and hormonal milieu.

One possible explanation is sex-specific disease severity and comorbidity patterns. In some regions, women with CLD present at more advanced stages or have a higher burden of inflammatory and autoimmune disorders—both established risk factors for thrombosis. Autoimmune liver diseases such as primary biliary cholangitis and autoimmune hepatitis are more prevalent in women and may predispose to PVT through chronic inflammation and endothelial dysfunction.

Hormonal factors may also play a significant role. While premenopausal estrogen exposure can be prothrombotic in the context of exogenous hormone use (oral contraceptives, hormone replacement therapy), postmenopausal estrogen decline has been associated with pro-inflammatory and pro-coagulant changes in the vascular endothelium. An Italian study observed a higher incidence of PVT in women with HCC, suggesting hormonal transitions and aging may interact with malignancy-related hypercoagulability.

Healthcare access and socio-cultural differences could also contribute. In some healthcare systems, women may be more proactive in seeking care, leading to higher detection rates of asymptomatic or early-stage PVT. Conversely, underdiagnosis among men cannot be excluded, particularly in resource-limited or high-burden clinical environments.

The association may further be influenced by differential prevalence of specific risk factors, such as myeloproliferative neoplasms, thrombophilias, or metabolic comorbidities, which were not comprehensively assessed in our dataset. Data from cerebral venous thrombosis (CVT) a condition sharing similar pathophysiology—show female predominance largely attributable to hormone-related factors and autoimmune disease, reinforcing the need for more granular, sex-disaggregated analyses in PVT research.

Interestingly, some studies contradict our findings. For example, in acute pancreatitis, females had a lower risk of developing PVT compared to males, underscoring that the impact of sex is likely disease-specific rather than a universal risk determinant. Genetic and epigenetic differences, such as sex-linked variations in coagulation pathway regulation, and anthropometric traits (e.g., height) may also contribute to these observed patterns.

Taken together, the lower odds of PVT among males in our cohort likely reflect a complex interplay of biological, hormonal, genetic, behavioral, and healthcare-related factors rather than a straightforward protective effect [27–34]. The absence of detailed data on hormone exposure, autoimmune disease prevalence, thrombophilia screening, and myeloproliferative neoplasms limits deeper interpretation.

We suggest that future research should incorporate sex-specific subgroup analyses adjusting for disease severity, comorbidity profile, and healthcare utilization: Collect detailed hormone use and reproductive history data and evaluate immune-mediated and genetic thrombophilia markers.

The median age of PVT patients in this study was 41 years, which is younger than reported in other countries. This could reflect regional differences in population age structure, earlier onset of liver disease, or differences in risk exposures. These demographic trends should be considered when designing local screening and management strategies [14, 15, 26, 35–41].

Hepatocellular carcinoma (HCC) emerged as a strong independent predictor of portal vein thrombosis (PVT), with nearly fourfold increased odds, consistent with previously reported data from Rome and Malmö. However, the confidence interval for this [29]association was notably wide, suggesting limited statistical precision. This imprecision is likely attributable to the relatively small number of patients with both HCC and PVT in our study, which restricts the robustness of the estimate. While the direction and strength of the association are consistent with prior studies, the limited subgroup size should be considered when interpreting the findings, and larger studies are needed to validate this relationship with greater precision. Tumor invasion into the portal venous system and hypercoagulability likely contribute to this association. Thrombocytosis (> 450,000/µL) was identified as an independent risk factor for PVT in our study, aligning with findings from Italian and German cohorts [42, 43]. This association is particularly noteworthy in the context of CLD, where thrombocytopenia is common, yet thrombocytosis may still indicate increased thrombotic risk.

The hypercoagulable state associated with liver dysfunction significantly increases thrombotic risk due to impaired production of anticoagulants like protein C and antithrombin and elevated procoagulant factors such as fibrinogen. Inflammatory cytokines, including interleukin-6 and TNF-α, stimulate megakaryocyte proliferation, contributing to thrombocytosis [44, 45]. Additionally, endothelial dysfunction in chronic liver disease enhances platelet activation and aggregation, further exacerbating thrombotic risk [46]. These findings highlight the significance of thrombocytosis as a risk factor for PVT, underscoring the need for tailored risk assessments and potential interventions in patients with elevated platelet counts.

The lack of significant associations between PVT and liver enzymes (e.g., AST) or hemoglobin levels aligns with prior findings from China, which suggest that routine liver function tests may not reliably indicate thrombotic risk [14, 47]. AST levels and hemoglobin were not significantly associated with PVT, consistent with Chinese studies suggesting these markers may not indicate thrombotic risk in cirrhosis. Although patients with hemoglobin > 16 mg/dL had higher odds of PVT, this association was not statistically significant and warrants further investigation [48].

The observed 25% prevalence of PVT among patients with Pancreatitis in our study is higher than the 0.8% reported in a U.S. nationwide inpatient sample. However, it is comparable to the 22.6% reported in a Polish cohort, highlighting the variability depending on diagnostic protocols and disease severity [49, 50]. Our findings underscore the need for vigilance and targeted imaging in Pancreatitis patients with clinical suspicion of PVT. Local causes of PVT such as prior abdominal surgery and gastrointestinal infections were also identified, aligning with reports from Vienna and other studies on pylephlebitis. The 4.7% prevalence of PVT due to GI infections in this cohort reflects the potential thrombotic complications of intra-abdominal sepsis [51].

A major limitation of this study is the absence of data on thrombophilic conditions and myeloproliferative neoplasms (MPNs), both of which are well-established risk factors for portal and splanchnic vein thrombosis. Due to the limited availability of advanced diagnostic testing in our setting, these conditions were likely underdiagnosed or entirely unassessed in many patients. As a result, the true prevalence of these prothrombotic disorders and their potential contribution to PVT may have been substantially underestimated. This limitation not only restricts our ability to fully characterize the spectrum of risk factors but may also have confounded the observed associations, particularly in patients without cirrhosis or malignancy. Future studies should prioritize comprehensive thrombophilia screening and evaluation for MPNs to provide a more complete understanding of the etiologic landscape of PVT in resource-limited settings.

The retrospective design of this study inherently limits the generalizability of our findings, as it relies on potentially incomplete or inconsistently documented medical records. While efforts were made to exclude or amputate such data, this limitation may still introduce biases and lead to misclassification or underreporting of key variables. This concern is particularly relevant in a single-center setting, which may not fully represent the broader population. Additionally, the absence of detailed medication history, including anticoagulant use, along with the lack of comprehensive thrombophilia testing (e.g., protein C/S deficiency, factor V Leiden mutation), significantly constrains our ability to accurately interpret risk factors associated with portal vein thrombosis. These unmeasured confounders may influence the observed associations and potentially obscure the identification of established prothrombotic risk factors. Consequently, our findings should be interpreted with caution, and further research is warranted to validate these associations in more diverse and adequately powered cohorts.

Overall, this study highlights the considerable burden of portal vein thrombosis among patients with chronic liver disease and HCC in Ethiopia. Strengthening diagnostic capacity and conducting prospective studies will be essential to clarify risk factors and improve clinical outcomes.

Conclusion and recommendations

This study revealed a substantial burden of portal vein thrombosis (15.7%) among hospitalized patients in Ethiopia. Chronic liver disease and hepatocellular carcinoma were the most significant independent predictors, while elevated platelet counts further supported the contribution of hypercoagulability. The predominance of younger male patients aligns with local disease patterns and suggests a need for context-specific clinical vigilance. Limited detection of coagulation or myeloproliferative disorders likely reflects diagnostic gaps in resource-constrained settings.

Given the high burden of PVT in patients with chronic liver disease and hepatocellular carcinoma, we recommend targeted screening and risk assessment strategies in these high-risk groups, particularly among younger males. Efforts should focus on strengthening diagnostic capacity—especially imaging and coagulation testing—within resource-limited settings. Incorporating PVT detection into national liver disease management guidelines and training healthcare providers in early recognition and referral are essential steps. Additionally, health policymakers should consider allocating resources to improve diagnostic infrastructure, and future research should explore genetic and hematologic contributors to PVT in the Ethiopian context.

Author contributions

GB conceived and designed the research protocol, participated in data collection, analysis and result writing. DN, TA and GAB approved the proposal with extensive revisions, participated in the data analysis, and wrote the manuscript. All the authors have read and approved the final manuscript.

Funding

No funding was received.

Data availability

The datasets used and/or analyzed during the current study are not publicly available due to confidentiality issues. However, they are available from the corresponding author upon reasonable written request and with approval from the ethical review board.

Declarations

Ethics approval and consent to participate

The Helsinki Declaration for medical research involving human subjects was complied with. Before the actual data-collection process, ethical clearance was obtained from the College of Medicine and Health Sciences, Bahir Dar University. Then, data collection was commenced after permission with protocol number 3057/2024. All patients were approached during their follow-up visits to the hospital and provided written informed consent, with legal guardians signing on behalf of individuals unable to consent. Confidentiality was maintained by removing names and other personal identifiers from the dataset before analysis.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.