Introduction

Warfarin is a class of anticoagulant drugs that are often used to treat diseases associated with thromboembolism, such as atrial fibrillation, venous thrombosis, and pulmonary thrombosis.^1,2 The main problem with the use of warfarin is that the variation in response between patients is very high.³ This causes difficulty in determining the initial dose of each patients appropriately, which will then result in the occurrence of DRP (drug-related problem) cases in the form of adverse drug reactions.^3–5 The high variation occurs due to the uniqueness of the drugs, which has the characteristics of a narrow therapeutic index. Therefore, underdose condition results in inadequate treatment or complications, while overdose leads to bleeding phenomena, ranging from severe instances such as cerebral hemorrhage to minor cases, namely ocular bleeding.^6–9

During the COVID-19 pandemic, the use of anticoagulants, including warfarin, gained significant attention due to the increased risk of thromboembolic complications in infected patients.^10–13 This highlights the critical need for precise warfarin dosing, as mismanagement could exacerbate complications related to both thromboembolism and bleeding. A previous study showed that 44% of patients who experienced bleeding had an INR value >3.0, whereas 48% of patients with thromboembolic events had an INR value <2.15.¹⁴ These findings highlight the significant risks associated with improper dosing and the need for careful monitoring of INR values in warfarin therapy.

Some of the factors that cause significant variations in response to warfarin use include clinical/demographic (age, weight, gender, body surface area, disease), non-clinical, and genetic factors (VKORC1, CYP2C9, CYP4F2).^15,16 Previous research has shown that genetic factors VKORC1 and CYP2C9 significantly influence variations in the pharmacokinetic and pharmacodynamic responses of warfarin.¹⁷ Patients carrying the homomutant VKORC1 gene type carrier (AA) show a low warfarin dose requirement, while the VKORC1 gene type (GG) tends to require a higher dose. Meanwhile, patients with homomutant (*3/*3) type carriers of CYP2C9 are at great risk of side effects in the form of bleeding. This condition necessitates the administration of warfarin at low doses. CYP2C9 wildtype (*1/*1) tends to require higher doses and risk disease complications when given standard doses.¹⁸

In recent research, another SNPs that could potentially influence warfarin therapy was found, namely CYP4F2 rs2108622. CYP4F2 catalyzes the conversion of vitamin K to its inactive metabolite, hydroxyvitamin K.¹⁹ The rs2108622 V433M variant results from a C > T nucleotide substitution, where the T allele replaces valine with methionine at position 433, reducing catalytic activity and potentially affecting blood clotting and warfarin response.¹⁷

A dosing algorithm model was needed to determine the appropriate initial and maintenance doses for patients receiving warfarin therapy. Several countries have developed algorithmic models to determine warfarin doses that are influenced by clinical, non-clinical, and genetic factors. Some of these models include Japan (Dose = 2.263 + 4.248 x (VKORC1 G/G) + 1.067 x (VKOCR1 A/G) − 2.416 x (CYP2C9*3/*3) − 0.864 (xCYP2C9*1/*3) + 1.308 x BSA + 0.025 x age), in China (Dose = 0.727–0.007 x age + 0.384 x BSA + 0.403 x (VKORC1 G/A) + 0.554 x (VKORC1 G/G) − 0.482 x (CYP2C9*1/*3) − 1.583 x (CYP2C9*3/*3), in Italy (Dose = 7.39764–0.02734 x age + 1.06287 x BSA − 1.04468 x VKORC1 A/G − 2.12117 x VKORC1), and USA (Dose = 3.52–0.006 x age + 0.38 x BSA − 0.15 x hypertension − 0.23 x (CYP2C9*1/*3 or *3/*3) − 0.24 x (VKORC1 A/G) − 0.48 x (VKORC1).^20–22 In Indonesia, there is still no development of this warfarin dosing algorithm model. Therefore, this research aimed to obtain a model of warfarin dosing algorithm or pattern according to the condition of each patient. The results can be applied as a guide in warfarin therapy in cardiac hospitals or clinics where cardiologists treat patients using warfarin.

Materials and Methods

Ethics Statement

This research complies with the principles outlined in the Declaration of Helsinki. Ethical approval was obtained from the West Java Health Ethics Commission-Faculty of Medicine, Universitas Padjadjaran with registration number 1342/UN6.KEP/EC/2019.

Subjects

The inclusion criteria were outpatients of the cardiac clinic who had been on warfarin therapy for ≥ 3 months, had Prothrombin Time-International Normalized Ratio (PT-INR) laboratory data available, had complete medical records, made routine medical visits, and were willing to participate. Similarly, the exclusion criteria were patients who took supplements containing vitamin K, and those who could not be followed up due to death, relocation of treatment, or inability to be contacted.

The sample size required for this study was calculated using the Lemeshow formula based on the allele prevalence:

Explanation of variables:

n: required sample size

d: margin of error (5%)

N: population size

Prev: prevalence of the CYP4F2 polymorphism (31.45% in the Asian population, as reported by Singh et al, 2011)¹⁷

Z: confidence level (95%, corresponding to 1.96)

Given that the population of warfarin therapy patients at Hasan Sadikin Hospital, Bandung, was 100, and the polymorphism prevalence (C > T) was 31.45%, the calculation is as follows:

All patients provided informed consent, then clinical characteristics, medical history, medications used, and daily warfarin doses were recorded. Clinical data were collected by reviewing medical records and direct inquiry during regular scheduled clinic visits. The clinical data included age, height, weight, gender, target INR, concomitant diseases, combined medications, and warfarin dosage.

Blood Sampling

A 3 mL blood sample was collected into marked EDTA tubes and stored at −20°C. The design of gene-specific primers for CYP4F2 rs2108622 was carried out by downloading the gene sequence from the National Center for Biotechnology Information (NCBI). After obtaining the sequence, the nitrogenous base sequence was input into the Primer-BLAST tool on the NCBI website (www.ncbi.nlm.nih.gov/tools/primer-blast/). The primers were then verified using the online OligoCalc software (http://biotools.nubic.northwestern.edu/OligoCalc.html). The primers are shown in Table 1, respectively.

Table 1 Primer

Deoxyribonucleic Acid (DNA) Extraction and Genotyping

A total of 200 μL of blood was placed in a 1.5 mL Eppendorf tube and 20 μL of proteinase K and 20 μL of Ribonuclease (RNAse) A solution were added. The mixture was homogenized by vortexing, then 200 μL of lysis solution C was added to the Eppendorf tube, and the tube was vortexed again for 15 seconds. The mixture was then incubated for 10 minutes at 55°C. After incubation, 200 μL of 95% ethanol was added to the lysate, and the mixture was homogenized by vortexing for 10 seconds.

DNA purification was performed using GenElute™ miniprep binding columns. The lysates, previously mixed with 95% ethanol, were transferred into the columns and centrifuged at 6,500 x g for one minute. The liquid in the collection tubes (2.0 mL) was discarded and replaced. The next step in the DNA purification process was the washing stage, using a wash solution concentrate that had been diluted with 95% ethanol. The DNA extraction process was concluded with the elution stage, where 100 μL of elution solution was added to the column and centrifuged at 6,500 x g for one minute, and the process was repeated twice.

The Polymerase Chain Reaction (PCR) process consists of three stages, namely denaturation, annealing, and extension. Several temperature variations were used to determine the optimal primer annealing temperature, including 55.4°C, 56.4°C, 57.4°C, 58°C, 59°C, 60°C, 61°C, 62°C, 63.4°C, and 64.4°C. The total reaction volume was 25 μL, comprising 2 μL of DNA template, 1 μL of forward primer, 1 μL of reverse primer, 12.5 μL of PCR Master Mix, and 8.5 μL of nuclease-free water. The PCR product was then electrophoresed on a 2% agarose gel at 80 volts for 90 minutes. The electrophoresis results were visualized under UV light at 312 nm using a fluorescence scanner. The PCR products were then sent to Humanizing Genomics Macrogen (https://www.macrogen.com/en/main/index.php), Korea, for sequencing. Sequencing was performed using the Sanger method, which relied on DNA synthesis with chain termination.

Statistical Analysis

The characteristics of the data were assessed to determine the normality using the D’Agostino or Kolmogorov–Smirnov tests. Based on the results, appropriate statistical test methods were applied. For normally distributed data, ANOVA or Student’s t-test was used for analysis, at a significance level of α = 0.05. Otherwise, the Kruskal–Wallis or Mann–Whitney U-test was applied.

Univariate analysis was conducted for descriptive analysis to determine the characteristics of each research variable, presented as number and percentage (n, %). Bivariate analysis was conducted to identify variables that could be included in the multivariate model, with a p-value < 0.05. Furthermore, the multivariate regression analysis (logistic regression) was used to examine the correlation and develop warfarin dosing model, considering both clinical and non-clinical factors, with a p-value < 0.05.

Results

A total of 77 patients participated in this research from March to December 2021. Demographic data and clinical characteristics of patients were obtained by reviewing medical records. Table 2 shows the description of patients demographic characteristics.

Table 2 Baseline Demographic, Clinical Characteristic and Mean INR Value

The average weekly dose based on age, Body Mass Index (BMI), and CYP4F2 rs 2108622 genotype are shown in Table 3. The results showed that the required dose decreases with increasing age. Specifically, patients aged 70–79 required a weekly dose of 16.17 mg, which is 27.33% lower than the highest average dose for patients aged 30–39, while patients aged 80–89 required a significantly lower dose of 7 mg (3 times smaller than the largest dose).

Table 3 Mean Weekly Doses (in Mg) for Age, BMI, and CYP4F2 Rs 2108622 Genotype

Bivariate Analysis

The results of the bivariate analysis between patients demographics and genotypes on warfarin dose are shown in Table 4. Variables with a p-value <0.25 in the bivariate analysis are eligible to enter the multivariate model.

Table 4 Results of Bivariate Analysis Between Patients Demographics and Genotype on Warfarin Dose

The Kruskal–Wallis test on genotype showed a p-value of 0.02 (<0.05), suggesting that the CC, CT, and TT genotypes have a significant association with warfarin dosage. Meanwhile, the Mann–Whitney test on gender had a p-value of 0.16 (>0.05). This result showed that gender does not have a significant relationship with warfarin dosage. However, gender was included in the multivariate analysis (p < 0.25) as a confounding factor.

The results of the Spearman Rank correlation analysis for age (p = 0.02) and BMI (p = 0.03) showed p-values <0.05. This implies that age and BMI have a significant relationship with warfarin dosage. The correlation coefficient values from this analysis were −0.28 for age and 0.25 for BMI. These results suggest that the strength of the relationship between age, BMI, and warfarin dosage is very weak (correlation coefficient: 0.00–0.30).²³ Specifically, as age increases, the required dose of warfarin decreases. Conversely, as BMI increases, the required dose of warfarin also increases.

Multivariate Analysis

Multivariate analysis aimed to determine the factors associated with warfarin dosing. Multiple linear regression was used to select age, BMI, sex, and CYP4F2 genotype for the creation of warfarin dosing formula. The results of the multiple linear regression analysis are shown in Table 5.

Table 5 Multiple Linear Regression Analysis Between Age, BMI, Gender, Genotype, and Warfarin Dose

Quality of Life

Quality of life of warfarin therapy patients in Dr. Hasan Sadikin Central General Hospital is presented in Table 4, with categories. The lower score showed a better quality of life and the higher score showed worse conditions. In addition, the results showed that the highest percentage score was included in the category < 56,266. This showed that most patients on warfarin therapy had a better quality of life.

The principle of multiple linear regression analysis used was backward elimination. In the initial model, all variables were entered simultaneously, and those with a significance value >0.05 were excluded. The final model of this regression analysis included three variables, namely age, BMI, and genotype. Table 5 shows that the final model analysis has a significance value of <0.01 for each variable. This result suggests that age (p = 0.01), BMI (p = 0.01), and genotype (p = 0.01) have a significant influence on the determination of warfarin dose.

Based on Table 5, the regression model can be expressed as y = 12.736–0.160×1 + 0.540×2 + 3.545X3, or dose = 12.736–0.16*age + 0.54*BMI + 3.55*CYP4F2 genotype, where 1 = CC, 2 = CT, and 3 = TT. The constant 12.736 represents warfarin dose in mg/week when age, BMI, and genotype are not considered. The regression coefficient of −0.16 (β1) shows that for every decrease in age, warfarin dose increases by 0.16 mg/week. The regression coefficient of 0.54 (β2) shows that each unit increase in BMI will raise warfarin dose by 0.54 mg/week. Finally, the regression coefficient of 3.55 (β3) suggests that the presence of the CYP4F2 C > T polymorphism increases warfarin dose by 3.55 mg/week.

The result in Table 5 showed an R-squared value of 0.25, showing that 25% of the variance in warfarin dose was explained by age, BMI, and CYP4F2 genotype, while the remaining 75% was determined by other factors not included in this research. The effective contribution of each variable was 8.76%, 8.29%, and 7.95% for age, CYP4F2 gene polymorphism, and BMI. The effective contribution can be calculated using the formula SE% = βx × rxy × 100%.

Discussion

In this research, 77 patients met the inclusion criteria, consisting of 37 men and 40 women, with an average BMI of 23.63 kg/m². The CYP4F2 rs2108622 gene polymorphism profile included 47 patients with the CC genotype, 27 with CT, and 3 with the TT. Table 3 shows that the older patients, the lower the dose required. The results of this research are consistent with previous reports that patients with middle and old age require warfarin doses 10.60% lower than young age, as the age of patients decreases the weekly dose by 0.40 mg per year of age.²⁴ In addition, in old age, there are many hemorrhagic events due to the use of drugs that can increase the risk of bleeding, such as antiplatelets, anticoagulants, statins, and amiodarone.²⁵ The low dose of warfarin in elderly patients was attributed to decreased activity of the vitamin K redox recycling system, which was affected by age-related physiological changes. These changes included alterations in body composition, an increase in fat tissue (leading to an increased volume of distribution for fat-soluble drugs), slowing of metabolic processes, and reduced blood perfusion to the intestinal region.^26,27

Dosing based on BMI classification showed that the higher the BMI index, the greater the weekly dose required. The average weekly dose for obese patients was 24 mg, which was 26.38% greater than the underweight and 5 mg higher than normal-weight patients (Table 3). This result was consistent with previous research showing a correlation between weekly dose and BMI. Research by Alshammari et al (2020) and Mueller et al (2014) showed significant results that obese patients require weekly doses 20% higher than those of normal and overweight.^28,29 According to Yoo et al (2012), an increase in body weight was directly proportional to the required warfarin dose and INR value. Patients over 80 years old and weighing less than 55 kg needed a maintenance dose of 3 mg. Meanwhile, those under 55 years old and weighing more than 50 kg required a dose of 10 mg. Patients within these two age and weight ranges needed a dose of 3–7 mg.³⁰ This is due to differences in pharmacokinetics in obese patients, specifically, in drug distribution within tissues, volume of distribution (Vd), blood flow, plasma protein binding, and drug elimination. The absorption process remains similar to that of normal-weight patients. Obese patients have greater absolute body and fat mass, and the hemodynamic conditions can enhance drug kinetics. Changes in plasma protein-binding concentrations can impact the movement of drugs into tissue compartments, influencing therapeutic effects. Furthermore, the need for larger weekly doses in obese patients was attributed to increased body weight, which affected the volume of distribution and clearance of warfarin, leading to elevated coagulation factors.³¹

Dosing based on the CYP4F2 rs2108622 genetic polymorphism showed that patients with CC, CT, and TT genotypes required doses of 19 mg, 21 mg, and 33 mg, respectively. The weekly dose for TT patients was significantly greater than CC and CT, as shown in Table 3. Several countries have conducted research on CYP4F2 polymorphism and the effect on warfarin dosing. Research in China,³² Iran,³³ Italy,³⁴ and India¹⁷ showed that patients with the CYP4F2 polymorphism required higher warfarin doses. However, research conducted on populations in the UK,³⁵ Japan,³⁶ and Norway³⁷ suggested that CYP4F2 polymorphism had no significant influence on warfarin dosing.

The CYP4F2 gene expression catalyzes the hydroxylation of vitamin K1 (VK1) into an inactive form, hydroxyvitamin K. This gene served as an important negative regulator of vitamin K levels, thereby affecting blood clotting.³⁸ The CYP4F2 rs2108622 V433M variant arises from a polymorphism including the C > T nucleotide substitution. The T allele in rs2108622 replaced a valine residue with a methionine residue at position 433 in the coding region. This change impacted enzyme activity, and drug metabolism, as well as physiological and pathophysiological processes. The increase in warfarin dose for CT and TT genotypes was consistent with the observed rise in plasma concentration.

Molecular dynamics (MD) research showed that the CYP4F2 V433M variant was associated with a decrease in protein stability, as evident by free energy values. Free energy values below zero suggested low stability. Destabilization of the protein structure could alter biological function and disrupt signal cascades and normal protein pathways. The V433M variant impacted the physicochemical characteristics, intermolecular interactions, as well as functional and structural properties of the protein. Furthermore, the mutant amino acid (methionine) was larger than the wild-type (valine), leading to structural mismatches within the protein. The wild-type amino acid was located in a critical position for interacting with other molecules that are essential for protein activity. Mutations could disrupt these interactions, affecting the signaling cascade from the binding to the activity domain.¹⁹

Research by McDonald et al in 2009 showed the participation of CYP4F2 in the oxidative degradation of vitamin K and oxidative activity. The protein encoded by the rs2108622 T allele had reduced activity compared to the wild-type in the genotyping of liver microsomal enzymes, with the TT phenotype showing a 75% reduction in vitamin K oxidative activity. The CYP4F2 rs2108622 V433M variant had a diminished ability to metabolize VK1 to hydroxyvitamin K1, resulting in reduced steady-state hepatic enzyme concentration. Consequently, patients with the rs2108622 polymorphism tend to have elevated hepatic VK1 levels, leading to a requirement for higher warfarin doses to achieve the same anticoagulant response.¹⁹

Based on the INR values obtained in this study, the majority of patients with CYP4F2 genotypes CC, CT, and TT had INR values within the target therapeutic range of 2–3. Among the CC genotype group, only 4 patients had INR values exceeding 3, while 3 patients in the CT group exhibited similar results. Notably, no patients with the TT genotype had INR values above 3. These findings suggest that most patients across all genotypes were effectively managed within the desired therapeutic range, reducing the risk of adverse outcomes such as bleeding. Furthermore, there were no reports of major bleeding events among the study participants, further supporting the safety of the dosing regimens utilized in this population (Table 2).

The algorithm model obtained was y = 12.736–0.160×1 + 0.540×2 + 3.545X3, or dose = 12.736–0.16*age + 0.54*BMI + 3.55*CYP4F2 genotype, where 1 = CC, 2 = CT, and 3 = TT. The results of this algorithm are consistent with several models developed in various countries, such as in Japan (Dose = 2.263 + 4.248 x (VKORC1 G/G) + 1.067 x (VKOCR1 A/G) − 2.416 x (CYP2C9*3/*3) − 0.864 (xCYP2C9*1/*3) + 1.308 x BSA + 0.025 x age), China (Dose = 0.727–0.007 x age + 0.384 x BSA + 0.403 x (VKORC1 G/A) + 0.554 x (VKORC1 G/G) − 0.482 x (CYP2C9*1/*3) − 1.583 x (CYP2C9*3/*3), Italia (Dose = 7.39764–0.02734 x age + 1.06287 x BSA − 1.04468 x VKORC1 A/G − 2.12117 x VKORC1), and USA (Dose = 3.52–0.006 x age + 0.38 x BSA − 0.15 x hypertension − 0.23 x (CYP2C9*1/*3 or *3/*3) − 0.24 x (VKORC1 A/G) − 0.48 x (VKORC1).^22,39,40

The similarity of the algorithm obtained in this research with those from several other countries was in the inclusion of age and BMI or BSA as factors in the dosing model. The correlation between age and dose was negative across research, namely Japan (+0.025 x age), China (−0.007 x age), Italy (−0.02734 x age), America (−0.006 x age), and Indonesia (−0.16 x age). This result showed that as age increases, the required dose tends to decrease. In contrast, BMI showed a positive correlation, suggesting that the higher the BMI, the greater the required dose. A key difference between the algorithm developed in this research and models from other countries was the genetic factors. While previous investigation focused on VKORC1 and CYP2C9, this research emphasized CYP4F2, due to its crucial role in the vitamin K cycle, which was directly related to the vitamin K intake.

The results of this study align with previous findings indicating that age and BMI significantly influence warfarin dosing. For example, Khoury et al (2014) demonstrated that warfarin dosage decreases with age, consistent with our findings.⁴¹ Similarly, the observed correlation between higher BMI and increased warfarin requirements corresponds with results reported by Alshammari et al (2020) and Mueller et al (2014).^28,29 However, our study highlights CYP4F2 as a genetic factor in warfarin dosing, diverging from studies in other countries that emphasize VKORC1 and CYP2C9. This underscores the importance of considering population-specific genetic variations, such as CYP4F2 in Indonesia, in developing dosing algorithms.

The limitations of this research include the relatively small sample size, which may not accurately represent the broader population, thereby limiting the generalizability of the results to all patients with similar conditions. Future research with larger sample sizes is needed to validate these results. Additionally, this research was conducted at only one hospital within a specific geographical area, which could introduce location and population bias, as patients from other regions or hospitals may exhibit different characteristics. Comprehensive analyses that incorporate more genetic factors, as well as other non-clinical variables, are necessary for a more thorough understanding of these issues.

Conclusion

In conclusion, the factors that influenced warfarin dose adjustment in cardiovascular patients in Indonesia were age, BMI, and the CYP4F2 gene polymorphism rs2108622. Specifically, as age increased, the required dose decreased. The CYP4F2 rs2108622 gene polymorphism also affected warfarin dose variation, with patients carrying the TT polymorphism requiring higher doses. The percentage contributions of each factor to warfarin dose adjustment included 8.76%, 7.95%, and 8.29% for age, BMI, and gene polymorphism, respectively. The total contribution of age, BMI, and CYP4F2 genotype to warfarin dose adjustment was 25%. Finally, the linear regression model for predicting warfarin dose was represented by the equation y = 12.736–0.16Age + 0.54 BMI + 3.55*Genotype. In addition, further exploration of International Normalized Ratio (INR) data could provide more insights into the warfarin response, as INR is a key parameter for monitoring warfarin therapy. The relationship between INR levels and the influencing factors identified in this study may help optimize dosing strategies for cardiovascular patients in Indonesia.

Funding

The authors are grateful to the Rector of Universitas Padjadjaran for funding this study (RKDU grant No 1918/UN6.3.1/PT.00/2024).

Disclosure

The authors report no conflicts of interest in this work.

References

1. Martin J, Somogyi A. Pharmacogenomics and warfarin therapy. therapeutic drug monitoring: newer drugs and biomarkers. Therape Drug Monitor Newer Drugs Biomark. 2012:161–173. doi:10.1016/B978-0-12-385467-4.00008-7

2. Daly AK. Pharmacogenomics of Warfarin. Handbook Pharmacogenomics Strat Med. 2014;497–507. doi:10.1016/B978-0-12-386882-4.00024-4

3. Putriana NA, Destiani DP, Putri AN, Latarissa IR. Quality of life of patients receiving warfarin therapy at a tertiary care centre in Indonesia using DASS (duke anticoagulation satisfaction scale). Vasc Health Risk Manag. 2024;20:403–413. doi:10.2147/VHRM.S467656

4. Loebstein R, Yonath H, Peleg D, et al. Interindividual variability in sensitivity to warfarin–Nature or nurture? Clin Pharmacol Ther. 2001;70(2):159–164. doi:10.1067/MCP.2001.117444

5. Van Spall HGC, Wallentin L, Yusuf S, et al. Variation in warfarin dose adjustment practice is responsible for differences in the quality of anticoagulation control between centers and countries: an analysis of patients receiving warfarin in the randomized evaluation of long-term anticoagulation therapy (RE-LY) trial. Circulation. 2012;126(19):2309–2316. doi:10.1161/CIRCULATIONAHA.112.101808/ASSET/2B596CB7-59AE-4100-A2E8-97818129888B/ASSETS/GRAPHIC/ZHC0441213230001.JPEG

6. Cross B, Turner RM, Zhang JE, Pirmohamed M. Being precise with anticoagulation to reduce adverse drug reactions: are we there yet? Pharmacogenomics J. 2024;24(2):1–23. doi:10.1038/s41397-024-00329-y

7. Petty GW, Brown RD, Whisnant JP, Sicks JRD, O’Fallon WM, Wiebers DO. Frequency of major complications of aspirin, warfarin, and intravenous heparin for secondary stroke prevention. A population-based study. Ann Intern Med. 1999;130(1):14–22. doi:10.7326/0003-4819-130-1-199901050-00004

8. Gulløv AL, Koefoed BG, Petersen P. Bleeding during warfarin and aspirin therapy in patients with atrial fibrillation: the AFASAK 2 study. Atrial fibrillation aspirin and anticoagulation. Arch Intern Med. 1999;159(12):1322–1328. doi:10.1001/ARCHINTE.159.12.1322

9. Kimmel SE. Warfarin therapy: in need of improvement after all these years. Expert Opin Pharmacother. 2008;9(5):677. doi:10.1517/14656566.9.5.677

10. Barnes GD, Burnett A, Allen A, et al. Thromboembolism and anticoagulant therapy during the COVID-19 pandemic: interim clinical guidance from the anticoagulation forum. J Thromb Thrombolysis. 2020;50(1):72. doi:10.1007/S11239-020-02138-Z

11. Latarissa IR, Barliana MI, Meiliana A, et al. Efficacy of quinine sulfate in patients with mild-to-moderate COVID-19:A randomized controlled trial. Indones Biomedl J. 2023;15(6):366–374. doi:10.18585/INABJ.V15I6.2543

12. Latarissa IR, Meiliana A, Sormin IP, et al. The efficacy of herbal medicines on the length of stay and negative conversion time/rate outcomes in patients with COVID-19: a systematic review. Front Pharmacol. 2024;15:1383359. doi:10.3389/FPHAR.2024.1383359

13. Latarissa IR, Rendrayani F, Iftinan GN, et al. The efficacy of oral/intravenous corticosteroid use in COVID-19 patients: a systematic review. J Exp Pharmacol. 2024;16:321–337. doi:10.2147/JEP.S484596

14. Putriana NA, Rusdiana T, Rostinawati T, Akbar MR, Destiani DP. Evaluation of adverse drug reaction in patients warfarin therapy. J Adv Pharm Technol Res. 2022;13(4):291–295. doi:10.4103/JAPTR.JAPTR_439_22

15. Yoshizawa M, Hayashi H, Tashiro Y, et al. Effect of VKORC1-1639 G>A polymorphism, body weight, age, and serum albumin alterations on warfarin response in Japanese patients. Thromb Res. 2009;124(2):161–166. doi:10.1016/J.THROMRES.2008.11.011

16. Patel S, Singh R, Preuss CV, Patel NW. Hemostasis and thrombosis: fourth edition. 2023. Published online March 24.

17. Singh O, Sandanaraj E, Subramanian K, Lee LH, Chowbay B. Influence of CYP4F2 rs2108622 (V433M) on warfarin dose requirement in Asian patients. Drug Metab Pharmacokinet. 2011;26(2):130–136. doi:10.2133/DMPK.DMPK-10-RG-080

18. Rusdiana T, Araki T, Nakamura T, Subarnas A, Yamamoto K. Responsiveness to low-dose warfarin associated with genetic variants of VKORC1, CYP2C9, CYP2C19, and CYP4F2 in an Indonesian population. Eur J Clin Pharmacol. 2013;69(3):395–405. doi:10.1007/S00228-012-1356-9

19. McDonald MG, Rieder MJ, Nakano M, Hsia CK, Rettie AE. CYP4F2 is a vitamin K1 oxidase: an explanation for altered warfarin dose in carriers of the V433M variant. Mol Pharmacol. 2009;75(6):1337–1346. doi:10.1124/MOL.109.054833

20. Pei L, Tian X, Long Y, et al. Establishment of a Han Chinese-specific pharmacogenetic-guided warfarin dosing algorithm. Medicine. 2018;97(36):e12178. doi:10.1097/MD.0000000000012178

21. Cho EH, Lee K, Yang M, et al. Development and validation of a novel warfarin dosing algorithm for Korean patients with VKORC1 1173C. Ann Lab Med. 2020;40(3):216. doi:10.3343/ALM.2020.40.3.216

22. Ramirez AH, Shi Y, Schildcrout JS, et al. Predicting warfarin dosage in European–Americans and African–Americans using DNA samples linked to an electronic health record. Pharmacogenomics. 2012;13(4):407. doi:10.2217/PGS.11.164

23. Mukaka MM. A guide to appropriate use of correlation coefficient in medical research. Malawi Med J. 2012;24(3):69.

24. Garcia D, Regan S, Crowther M, Hughes RA, Hylek EM. Warfarin maintenance dosing patterns in clinical practice: implications for safer anticoagulation in the elderly population. Chest. 2005;127(6):2049–2056. doi:10.1378/CHEST.127.6.2049

25. Shendre A, Parmar GM, Dillon C, Beasley TM, Limdi NA. Influence of age on warfarin dose, anticoagulation control, and risk of hemorrhage. Pharmacotherapy. 2018;38(6):588–596. doi:10.1002/PHAR.2089

26. Miura T, Nishinaka T, Terada T, Yonezawa K. Relationship between aging and dosage of warfarin: the current status of warfarin anticoagulant therapy for Japanese outpatients in a department of cardiovascular medicine. J Cardiol. 2009;53(3):355–360. doi:10.1016/J.JJCC.2008.12.003

27. Aktan A, Güzel T, Aslan B, et al. Comparison of the real-life clinical outcomes of warfarin with effective time in therapeutic range and non-vitamin K antagonist oral anticoagulants: insight from the AFTER-2 trial. Kardiol Pol. 2023;81(2):132–140. doi:10.33963/KP.A2022.0287

28. Alshammari A, Altuwayjiri A, Alshaharani Z, Bustami R, Almodaimegh HS. Warfarin dosing requirement according to body mass index. Cureus. 2020;12(10). doi:10.7759/CUREUS.11047

29. Mueller JA, Patel T, Halawa A, Dumitrascu A, Dawson NL. Warfarin dosing and body mass index. Ann Pharmacother. 2014;48(5):584–588. doi:10.1177/1060028013517541

30. Yoo SH, Kwon SU, Jo MW, Kang DW, Kim JS. Age- and weight-adjusted warfarin initiation nomogram for ischaemic stroke patients. Eur J Neurol. 2012;19(12):1547–1553. doi:10.1111/J.1468-1331.2012.03772.X

31. Cheymol G. Effects of obesity on pharmacokinetics implications for drug therapy. Clin Pharmacokinet. 2000;39(3):215–231. doi:10.2165/00003088-200039030-00004

32. Li JH, Ma GG, Zhu SQ, Yan H, Wu YB, Xu JJ. Correlation between single nucleotide polymorphisms in CYP4F2 and warfarin dosing in Chinese valve replacement patients. J Cardiothorac Surg. 2012;7(1):97. doi:10.1186/1749-8090-7-97

33. Khosropanah S, Faraji SN, Habibi H, Yavarian M, Mansoori R, Haghpanah S. Correlation between Rs2108622 locus of CYP4F2 gene single nucleotide polymorphism and warfarin dosage in Iranian cardiovascular patients. Iran J Pharm Res. 2017;16(3):1238.

34. Borgiani P, Ciccacci C, Forte V, et al. CYP4F2 genetic variant (rs2108622) significantly contributes to warfarin dosing variability in the Italian population. Pharmacogenomics. 2009;10(2):261–266. doi:10.2217/14622416.10.2.261

35. Zhang JE, Jorgensen AL, Alfirevic A, et al. Effects of CYP4F2 genetic polymorphisms and haplotypes on clinical outcomes in patients initiated on warfarin therapy. Pharmacogenet Genomics. 2009;19(10):781–789. doi:10.1097/FPC.0B013E3283311347

36. Harada T, Ariyoshi N, Shimura H, et al. Application of Akaike information criterion to evaluate warfarin dosing algorithm. Thromb Res. 2010;126(3):183–190. doi:10.1016/J.THROMRES.2010.05.016

37. Kringen MK, Haug KBF, Grimholt RM, et al. Genetic variation of VKORC1 and CYP4F2 genes related to warfarin maintenance dose in patients with myocardial infarction. J Biomed Biotechnol. 2011;2011. doi:10.1155/2011/739751

38. Antman EM. Cardiovascular therapeutics: a companion to braunwald’s heart disease: fourth edition. Elsevier. 1–807.

39. Lei X, Guo Y, Sun J, et al. Accuracy assessment of pharmacogenetic algorithms for warfarin dose prediction in Chinese patients. Am J Hematol. 2012;87(5):541–544. doi:10.1002/AJH.23151

40. Cho HJ, On YK, Bang OY, et al. Development and comparison of a warfarin-dosing algorithm for Korean patients with atrial fibrillation. Clin Ther. 2011;33(10):1371–1380. doi:10.1016/J.CLINTHERA.2011.09.004

41. Khoury G, Sheikh-Taha M. Effect of age and sex on warfarin dosing. Clin Pharmacol. 2014;6(1):103–106. doi:10.2147/CPAA.S66776

Background

Adherence to COVID-19 prevention guidelines and vaccination in most low-income countries is challenging due to widespread negative information dissemination.^1–3 A variety of factors influence the adherence to COVID-19 protocols and vaccine acceptance across different populations, resulting in varying uptake rates.⁴

COVID-19 vaccines became available to a broader range of people over time, beyond those initially targeted by vaccination campaigns in most countries. However, with only 12% COVID-19 full vaccination rates by March 2022, it was estimated that Sub-Saharan Africa would need to increase its vaccination efforts by a factor of six in order to meet its mid-year vaccination targets.⁵ COVID-19 vaccination uptake is influenced by acceptance, trust, and willingness to receive vaccines.⁶ It has been proposed that in order to promote COVID-19 services response and vaccine uptake, it is necessary to assess the targeted populations’ knowledge of ways to reduce the risk of contracting COVID-19, vaccine uptake, willingness and hesitancy to accept vaccination, and the factors influencing such decisions.²

The COVID-19 vaccination program began in Uganda on March 3, 2021, nearly five months after the developed world began vaccination, and there has been no assessment of the COVID-19 response or vaccination status throughout the country.⁴ The purpose of this community-based survey was to determine adherence to COVID-19 standard operating procedures, the status of COVID-19 vaccination, and the reasons for vaccine acceptance and hesitancy in order to plan interventions to increase COVID-19 vaccine uptake in eight districts in central Uganda. The study also looked into what influenced respondents to accept or reject COVID-19 vaccination. All of this information is intended to guide the districts’ ongoing and future COVID-19 and other epidemic response planning of related nature.

The COVID-19 response and vaccination campaign are affected by a variety of factors, some of which are complex based on geographic, cultural, and settlement context, affecting vaccine coverage and other COVID-19 response services.⁷ These complex factors influencing the pandemic responses necessitate refining and contextualizing COVID-19 mitigation plans to the specific needs of geographical units identified as underperforming. As a result, evaluating existing response plans and determining the factors influencing response in the targeted communities is critical to informing any evidence-based changes needed to effectively address the pandemic. The Lot Quality Assurance Sampling (LQAS) provides for differentiating between good and poor performance geographical areas, the reason for its choice in this study. This evaluation method was previously used to track the performance of routine immunization and other health services.^8,9 It has also been used to assess factors influencing COVID-19 mitigation in Nigerian communities.¹⁰ This LQAS survey was employed to track the COVID-19 response on the assumption that the COVID-19 pandemic would impact on the HIV/AIDS pandemic response. As a result, Mildmay Uganda found it necessary to strengthen the districts’ COVID-19 response in order to avoid losing the gains made in districts where it has been implementing HIV/AIDS response interventions.

Methods

Study Design and Sampling

A cross-sectional community-based household survey was conducted in the districts of Kiboga, Kyankwanzi, Mubende, Kasanda, Mityana, Luwero, Nakaseke, and Nakasongola using the binomial LQAS methods. By combining geographical regions known as sub-counties, town councils (TC), or divisions, we stratified each district into five supervision areas (SAs), yielding 40 SAs (Table 1). The study targeted women aged 15–49 years and men aged 15 years or older (15+ years). Based on the classical LQAS principles¹¹ with each district stratified into five supervision areas (SAs), a two-stage sampling plan was used to randomly select 19 villages/interview locations from each SA, yielding a district sample size of 95. A sample of 190 respondents was generated for each district for the two respondent groups, totaling 1,520 respondents for the eight districts.

Table 1 The Supervision Areas (SA) for All the 8 Districts

A random sample of 19 interview locations was drawn from each SA using a probability proportionate to size (PPS) based on projections from the 2014 Uganda population and housing census. This method ensured that the likelihood of sampling a village was proportional to the size of its population. We began by generating a list of villages from each SA, as well as the population of each village, and then calculated the cumulative population. A sampling interval (Si) was obtained by dividing the total SA population by 19 (the SA-level sample size). A random number between 1 and Si was chosen to determine the starting village. To select the second, third, until 19 interview locations in each SA, Si was added to the random number. We used segmentation sampling to identify the random starting point, ie, the reference household, in order to select households in the sampled interview location/village. Segmentation was done by mapping, sub-dividing the village into segments of approximately equal household numbers before randomly selecting one segment. The segmentation process was repeated until selection of a segment with manageable number of households (15-<30) was selected. At this point, the households were listed and a reference household randomly selected. Segmentation was done with the help of a village guide. No interview was conducted in the reference household, but the nearest household to the reference household’s front door was identified where the search for eligible respondents (women 15–49 years and men 15+ years) started.

We used a parallel sampling approach to select respondents from the households in a “next nearest” household sequence until two interviews (ie, one questionnaire set) were completed in each interview location. Administering only one questionnaire set in each interview location aids in avoiding clustering and reduces the survey design effect to close to one. To ensure independence and avoid clustering, a new random starting household was selected through re-segmentation for each questionnaire set in villages sampled more than once. Indicators were chosen because they were found to be useful in informing interventions aimed at improving adherence to the COVID-19 standard operating procedures as well as COVID-19 vaccination.

Data Analysis

Data for each indicator was analyzed using percentage coverage and 95% confidence intervals for each district separately, as well as for all the districts combined. SA performance was evaluated by comparing the SA’s coverage to the overall coverage estimate for each indicator using the LQAS decision rule (DR). A DR in this study refers to the minimum number of respondents (out of those sampled per SA and per indicator) who have the characteristic of interest (correct responses, eg, received a COVID-19 vaccination) on which the SA is adjudged to have reached average coverage. Any SA whose number of correct responses equals or exceeds the DR is considered to have reached average coverage and thus has acceptable performance in the indicator; otherwise, the opposite is true. An excel spreadsheet and SPSS version 22 were used for the analysis. The Pareto chart was used to identify the common reasons for not vaccinating and those reasons that made up to at least 80% of all reasons were classified as common. However, we removed the “trivial many” reasons that were clustered under the “other” category in the pareto analysis.

Ethics

The Mildmay Uganda Research and Ethics Committee (MUREC) (reference number REC REF 0804–2018) and the Uganda National Council of Science and Technology (UNCST) approved this study (reference number SS639ES). Informed consent was obtained from respondents who signed or thumb printed the informed consent form as proof of acceptance to participate. Participants’ names were not written on any of the data collection tools or mentioned in any report including the manuscript. The study adhered to all Declaration of Helsinki (ethical principles for medical research involving human subjects).¹²

Prior to selection and interviewing minors (those aged below 18 years), written informed consent was obtained from their parents or caregivers were provided with sufficient information about the study objectives, risks and benefits of their children participating in this study, as well as about consent and confidentiality concerns. The parents were also informed of the options for withdrawing their children from the study even after having consented. Following parents’ consent to their children participating in the study, the children were explained the study objectives and their rights. Thereafter, assent was obtained from them as well. For the parents who refused their children to participate in the study, such children were replaced.

Results

Characteristics of Respondents

Majority of the respondents, 22.6% of women 15–49 years and 19.1% of men 15+ years were between the ages of 30 and 34. The majority of respondents, 32.1% of women and 29.3% of men had an incomplete primary education as their highest level of education. Table 2 describes the respondents’ characteristics.

Table 2 Characteristics of the Respondents

COVID-19 Related Knowledge, Practice and Vaccination

We assessed COVID-19 knowledge, adherence to COVID-19 social distancing measures in the previous 24 hours, frequency of handwashing with soap and water or use of a hand sanitizer for COVID-19 prevention, and COVID-19 vaccination among women 15–49 years and men 15+ years. COVID-19 vaccination coverage was calculated among women 15–49 years old and men 18+ years old, as COVID-19 vaccination was only available to people over the age of 17 in Uganda at the time of this study. Table 3 summarizes the overall and district-level coverage (percentage) in all COVID-19-related knowledge, practice, and vaccination indicators from the study, while Table 4 presents the SA-level classification of coverage in selected indicators that are eligible for SA-level classification.

Table 3 Overall and District-Level Coverage in COVID-19 Indicators

Table 4 COVID-19 Indicator Coverage Classified at the SA-Level: Red for Correct Responses < DR (Below Coverage), Green for Correct Responses ≥ DR (Average or Above Coverage)

Knowledge of Ways to Reduce the Risk of Contracting COVID-19

Only 45.4% (95% CI: 41.9–49.0) of women and 48.6% (95% CI: 45.0–52.1) of men could name at least four ways to reduce the risk of contracting COVID-19. Across districts, women generally lagged behind men in understanding COVID-19 risk reduction measures. There were significant gender and district-level disparities in knowledge. For women, the percentage who could name at least four risk reduction methods varied from 23.5% (95% CI: 14.8–32.2) in Nakaseke to 68.0% (95% CI: 58.4–77.6) in Kyankwanzi and Nakasongola. Similarly, among men aged 15+, the lowest proportion was in Nakaseke (16.2%; 95% CI: 8.6–23.7) and the highest in Kyankwanzi (72.8%; 95% CI: 63.7–81.9). Districts with below-average coverage of individuals who knew at least four risk reduction methods were Luwero, Mubende, and Nakaseke for women, and Luwero, Mityana, and Nakaseke for men. Notably, Luwero and Nakaseke districts showed below average coverage for both genders for this indicator.

The findings from Table 4 regarding the classification of supervision areas regarding knowledge of at least four or more ways to reduce COVID-19 risk reveal significant gaps in knowledge about COVID-19 risk reduction measures across various districts. In Kyankwanzi, Kasanda and Mityana, one out of every five “SAs” lacked awareness of at least four recommended ways to reduce COVID-19 risks. Similarly, in Kiboga, two out of every five “SAs” had insufficient knowledge, while in Mubende and Nakaseke, three out of five “SAs” faced the same issue. The majority of “SAs”, specifically four out of five in Luwero district did not meet the DR. Consequently, less than half of the participants residing in these “SAs” were acquainted with adequate COVID-19 risk reduction strategies. However, it is notable that the remaining “SAs” did meet the decision rule (DR), representing at least 50.0% coverage. For men aged 15 and above, the situation was particularly concerning. In Luwero, Mityana, and Mubende districts, one out of four “SAs” failed to achieve the required DR. In Kiboga, it was three out of five while it was four out of the five “SAs” in Nakaseke. In all these “SAs” where the decision rule was not attained, less than 50.0% of men aged 15 and above were knowledgeable about adequate COVID-19 risk reduction measures.

Adherence to COVID-19 Social Distancing Measures During the Last 24 hours

Women aged 15–49 years and men aged 15+ years were asked if they had had direct contact with anyone who was not staying with them in the previous 24 hours (spent more than one minute within two meters of someone or touching, including shaking hands, hugging, kissing, or touching the shoulder). Those who answered “no” were classified as following the COVID-19 social distancing measure. Table 3 shows that 67.2% (95% CI: 63.9–70.6) of women and 66.5% (95% CI: 63.1–69.9) of men reportedly adhered to the COVID-19 social distancing measures in the 24 hours preceding the survey. In Kyankwanzi district, the proportions of women 15–49 years (48.1% (95% CI: 37.9–58.3) and men 15+ years (38.1% (95% CI: 28.2–48.1) who adhered to COVID-19 social distancing measures were (each) lowest. Coverage of women 15–49 years and men 15+ years who adhered to COVID-19 social distancing measures during the 24 hours preceding the survey was lower than the average coverage in the districts of Kiboga (61.2%, 51.5%), Kyankwanzi (48.1%, 38.1%), and Mubende (59.4%, 60.3%). In the Luwero district, social distancing was most frequent among both women (76.6% (95% CI: 67.9–85.4) and men (77.2% (95% CI: 68.6–85.8) (Table 3).

For women aged 15–49 years, Table 4 shows that one out of the five “SAs” in Kasanda and Mityana, two of the SAs in Mubende, Kiboga and in Nakaseke, and four out of the five “SAs” in Kyankwanzi, did not meet the DR of 11, implying that less than 67.2% of women 15–49 years in these SAs reported adhering to COVID-19 social distance standards in the 24 hours preceding the survey. The remaining SAs met the DR and thus had at least 70.0% coverage. Among men aged 15+ years, One out of the five “SAs” in Mityana, two out of five SAs in Nakaseke, Mubende and in Nakasongola, three out of five SAs in Kiboga, and four out of the five “SAs” in Kyankwanzi did not achieve the DR of 11. This implies that less than 70.0% of men 15+ years in these “SAs” reported adhering to COVID-19 social distancing standards. The remaining SAs met the DR and thus had at least 70.0% coverage.

COVID-19 Related Handwashing or Use of Hand Sanitiser

A respondent was considered to have frequently washed hands if s/he reported to have washed hands with water and soap or used a hand sanitiser at least 6 times during the 24 hours preceding the survey. Handwashing frequently was very low generally and in the districts among the women 15–49 years. Only 24.8% of the women (95% CI: 21.7–27.9; range: 14.1% [Mubende] – 31.8% [Luwero]) and 19.0% (95% CI: 16.2–21.8, range: 7.1% [Mubende] – 26.4% [Luwero]) of men frequently washed their hands or used a hand sanitizer during the 24 hours preceding the survey. Overall handwashing frequency was low among women 15–49 years and men 15+ years though some SAs exhibited even a poorer coverage. Whereas all the SAs should be prioritized for improvement, more effort should be put on SAs that did not attain the DR as in Table 4. The poorest of the poor performing SAs regarding handwashing or use of a hand sanitizer among women include; C in Kiboga district, and SAs L and N in Mubende district. Among the men 15+ years, SAs C and E in Kiboga district, and N and O in Mubende district fell short of the DR.

COVID-19 Vaccination

COVID-19 vaccination coverage exhibits a notable disparity between initial dose administration and series completion. Among women aged 15–49 years, 83.5% (95% CI: 80.8–86.1) received at least one dose, while men aged 18+ years showed a similar trend at 83.0% (95% CI: 80.0–85.0). However, the proportion of individuals completing the recommended vaccine series (1 dose for Johnson and Johnson, 2 doses each for AstraZenecca, Pfizer, Sputnik V and Moderna) was significantly lower, at 37.5% (95% CI: 34.0–41.0) among women and 41.5% (95% CI: 37.9–45.0) among men. Geographic disparities in vaccination completion were observed, with Kasanda district reporting the lowest coverage estimates at 21.7% (95% CI: 13.3–30.2) among women aged 15–49 years and 28.7% (95% CI: 19.3–38.1) among men aged 18+ years. In contrast, Mityana district achieved the highest coverage, with 56.1% (95% CI: 45.8–66.3) of women aged 15–49 years fully vaccinated. Among men aged 18+ years, Kiboga, Kyankwanzi, and Mityana districts reported completion rates exceeding 50%, at 51.7% (95% CI: 41.4–61.9), 50.7% (95% CI: 40.4–60.9), and 64.8% (95% CI: 54.8–74.8), respectively (Table 3).

Vaccination coverage disparities were observed in various Supervision Areas (SAs) among women aged 15–49 years. In Kiboga, Luwero, and Mubende districts, only three out of five SAs achieved the Decision Rule (DR) of 5, resulting in vaccination coverage of less than 40.0% among women in this age group. In contrast, Nakaseke and Kasanda districts had two and three SAs, respectively, that failed to attain the DR, yielding comparable coverage rates. Conversely, the remaining SAs in these districts achieved the DR, corresponding to vaccination coverage of at least 40.0% (Table 4). Similarly, among men aged 15+ years, vaccination coverage gaps were evident. In Kiboga, Luwero, and Mubende districts, one out of five SAs, and in Kasanda, Nakaseke, and Nakasongola districts, two out of five SAs, failed to reach the DR of 6, resulting in vaccination coverage of less than 45.0% among men in this age group. The remaining SAs in these districts achieved the DR, corresponding to vaccination coverage of at least 45.0% (Table 4).

Reasons for Not Getting Vaccinated

The reasons behind non-vaccination against COVID-19 among women aged 15–49 years and men aged 15+ years who reported never having received a COVID-19 vaccine were investigated. The responses, summarized in Table 5, revealed distinct patterns of reasons for non-vaccination among the men and women. Figures 1 and 2 illustrate the cumulative proportions of the most common barriers to vaccination cited by women and men respectively. Among women, the primary reasons for non-vaccination were: Fear of side effects (27.7%), Confusion regarding COVID-19 vaccine information (13.7%), Perceived ineffectiveness of vaccines (10.9%), Geographic accessibility issues, including long distances to vaccination sites (10.2%) and lengthy travel times (6.6%). In contrast, men cited the following reasons for non-vaccination: Fear of side effects (27.3%), Confusion regarding COVID-19 vaccine information (15.7%), Perceived ineffectiveness of vaccines (14.7%), Time constraints (10.8%), Geographic accessibility issues, including long distances to vaccination sites (10.6%) and lengthy queues at service points (6.4%) and Misconceptions regarding COVID-19 vaccine-related infertility (4.7%).

Table 5 Reasons Why Women 18–49 and Men 18+ Years Have Not Received Any Dose of COVID-19 Vaccine

Figure 1 The common reasons for non-uptake of COVID-19 vaccination among women 18–49 years.

Figure 2 The common reasons for non-uptake of COVID-19 vaccination among men 18+ years.

Motivators for COVID-19 Vaccine Uptake Among the Unvaccinated Respondents

We inquired with respondents who had not received any COVID-19 vaccine dose about their motivations for vaccination. Among vaccine-hesitant women aged 15–49, 19.0% cited trust in health workers’ recommendations. Motivational factors varied by district. In Kiboga, many women expressed willingness to vaccinate if assured of the vaccines’ safety based on global usage. Conversely, in Kyankwanzi, 36.4% preferred vaccines manufactured domestically. In Mityana, the majority relied on Ministry of Health (MoH) recommendations. In Luwero, 25.1% emphasized the importance of easy access to vaccines at local health facilities. For men aged 18+, 20.2% were swayed by health worker recommendations. The sight of earlier recipients without side effects influenced decisions significantly, particularly in Kiboga (28.3%), Mityana (46.7%), and Nakaseke (48.2%). Additionally, MoH endorsement held weight in Luwero (12.7%) and Mubende (39.0%). These insights underscore the localized nature of vaccine hesitancy and the need for tailored approaches to address it (Table 5).

Top motivators for women aged 15–49 to get vaccinated include health worker recommendations (19.0%), easy accessibility (16.0%), MoH endorsement (13.6%), observing side-effect-free users (11.7%), and shortened vaccination site distance (8.4%). For men 18+, motivators are health worker recommendations (20.2%), observing side-effect-free users (19.5%), MoH endorsement (12.6%), accessibility (11.0%), and concern over vaccination requirements for public places or travel (8.4%) (Table 6).

Table 6 Overall and District-Level Factors/Issues That Would Motivate Respondents Who Have Not Had Any COVID-19 Vaccination to Get Vaccinated

People Who Would Influence Defaulters to Take Up COVID-19 Vaccination

Respondents who had not received a COVID-19 vaccine were asked about influential figures in their decision to vaccinate. Among women aged 15–49, health workers or family doctors (31.3%), followed by village health team members (19.5%), and local leaders (19.4%) held the most sway. Family members or relatives (15.0%) and friends (9.3%) also played roles. Among men aged 18 and above, local leaders (27.8%) were most influential, followed by health workers or family doctors (24.6%), family members or relatives (14.3%), mass media information (12.1%), and village health team members (10.7%) (Table 7).

Table 7 Overall and District-Level Proportion of Different Categories of People Who Would Influence Defaulters to Take Up COVID-19 Vaccination

Discussion

This study demonstrates the importance of utilising localized and timely data-driven strategies for public health response management. To target interventions more effectively, healthcare managers and leaders at mid- and lower levels can use the LQAS methodology to identify areas of low public health response measure adoption or poor adherence to pandemic, epidemic, or outbreak prevention interventions.

The findings revealed that despite widespread awareness about COVID-19, knowledge of prevention measures was low among both men and women. Less than half of the respondents demonstrated knowledge of at least four ways to reduce the risk of COVID-19 contraction. Adherence to social distancing standards was also inadequate in many areas, with 17 supervision areas (SAs) for women and 14 SAs for men falling short. While first-dose vaccination coverage was high (83.5% for men and 83.0% for women), full vaccination coverage remained low (37.5% for women and 41.5% for men). Additionally, handwashing and sanitizing habits were poor, with only 24.8% of women and 19.0% of men reporting frequent hand hygiene practices in the previous 24 hours. With a significant decrease in COVID-19 cases at the time of the survey, complacency may have set in, leading to a disregard for standard operating procedures (SOPs). Additionally, a large proportion of the population had received their first vaccine dose, potentially created a false sense of protection and increased disregard for SOPs like social distancing and handwashing. The Omicron variant, which was less fatal than the previous Delta variant, may have also contributed to a sense of security. Furthermore, Uganda was nearing the end of the Omicron pandemic wave, leading to fatigue in adherence to COVID-19 prevention guidelines, as seen in other studies.^13–16 Adherence to COVID-19 standards in Uganda, had been strictly enforced by security forces. The relaxation of strict enforcement by security forces at the study time may have also played a role.¹³

The survey found high first-dose vaccination coverage rates: 83.5% (95% CI; 80.8–86.1) for women and 83.0% (95% CI; 80.0–85.0) for men. At the time of the study, Uganda’s national coverage on March 14, 2022 was 64.4%, with 8,014,082 (36.5%) of the target population fully vaccinated.¹⁷ As of mid-March In the study area, full vaccination coverage was 37.5% (95% CI; 34.0–41.0) for women and 41.5% (95% CI; 37.9–45.0) for men, with men’s coverage significantly higher than the national average.¹⁸ Women’s coverage was slightly higher than the national average, but not statistically significant. The higher coverage in the study area may be due to its location in central Uganda, with better access to COVID-19 services, proximity to the central vaccines store, and a well-developed road network. As the COVID-19 vaccination program began in this region, community members may have been early adopters of the vaccine, contributing to higher coverage rates.

This study found that men had higher COVID-19 full vaccination rates than women. This is contrary to women’s typical higher use of routine health services compared to men.¹⁹ This trend is seen in other countries, where women are more hesitant to get vaccinated due to various myths, including the false belief that COVID-19 vaccines cause infertility.^7,20,21 In Uganda, 0.9% of unvaccinated women and 4.7% of unvaccinated men cited this myth as a reason for not getting vaccinated.²² Similar gender gaps in vaccine acceptance exist elsewhere in Africa, with women showing higher rates of resistance and hesitance.^23,24 The infertility myth may lead men to discourage their wives from getting vaccinated.^7,20 Besides, early vaccine scarcity may have favored men who could travel to access vaccines, contributing to the observed gender disparity.

Despite the low full vaccination coverage observed, Mityana district stands out among all other districts for having significantly higher full COVID-19 vaccination coverage, whereas Kasanda district has the lowest coverage for women and men. The COVID-19 vaccination coverage observed in this study could also be explained by logistical, structural, and other contextual factors, as has been the case throughout Africa. Such issues have included vaccine distribution challenges, particularly in rural areas, as well as vaccine storage challenges, particularly due to poor cold chain due to a lack of electricity in rural communities.²⁵ Uganda has used a variety of vaccines, the supply of which has been inconsistent.²⁶ This resulted in situations such as preferred vaccines not being available at vaccination centers, as reported by 1.6% of non-COVID-19-vaccinated individuals, or the absence of eligible second dose vaccines for those seeking a second dose.²⁷ Addressing such logistical and structural issues may aid in improving vaccine access, uptake, and adherence.

The findings reveal that while there were shared concerns and barriers to COVID-19 vaccination among women and men, distinct differences also existed. Women mentioned that their motivation for COVID-19 was majorly influenced by the convenience and accessibility to vaccination sites to their residences and work places. Conversely, men would be persuaded to get vaccinated due to the requirements of COVID-19 vaccination to travel or to access their work places, vaccination status influencing access to public places and events as well as due to peer pressures, social norms. Similar gender differences in motivations have been observed in various countries, including the United States, Europe, and Australia.^28,29 As women and men have different motivations for COVID-19 vaccination uptake, this demonstrates that gender-sensitive communication strategies are needed in public health campaigns especially in disease outbreak responses. Thus, adapting messaging and responses to these gender differences such as emphasizing convenience for women and social influences for men can increase public health response uptake more so if they involve vaccinations.

The results further revealed that health workers’ recommendations for vaccine uptake are a stronger motivator for COVID-19 vaccination for both women and men. This observation has been mentioned elsewhere as a motivating factor COVID-19 vaccination among both men and women.³⁰ Women in this study were also found to have greater trust in community health workers regarding COVID-19 vaccination information. This could be explained by the strong social relationships, a need for individualized communication, and a sense of empathy and understanding that the women may be benefiting from the community health workers also known as the Village Health Teams (VHTs) as pointed out in previous research.³¹ On the other hand, men preferred more formal and authoritative sources like formal health workers and community leaders as trusted sources of information. This is contrary to the findings in another study where health workers were a less reliable source of information and trust especially on COVID-19 vaccination given their mistrust of the vaccines deriving from the negative information from unreliable sources such as social media.³²

The traditional gender roles of valuing authority and expertise tend to lead men to seek health information from formal sources such as health workers and community leaders.³¹ It is possible that this perception is based on the belief that health workers undergo extensive training and therefore possess expertise, experience, objective, credible and reliable information.³³ In addition, the belief that local leaders are trustable sources of information may have driven male to prefer seeking information from local leaders who predominantly are male as seen in other studies.³⁴ Thus, gender-specific influencers and communication channels should be considered when selecting media and people to air out or carry out health education aimed at disseminating public health responses information. Empowering the trusted information sources like the COVID-19 ambassadors in this study to deliver public health response information could go a long way in achieving desired results.³⁵

Observing no side effects experienced by those who have been vaccinated was equally alluded to by both men and women as a key motivating factor that increases willingness to get vaccinated against COVID-19. Consequently, testimonies and positive livid experiences given by those who have been vaccinated, can be used to demystify false beliefs, myths, and negative perceptions against COVID-19 vaccination, and help those who are unwilling to vaccinate to change their beliefs about vaccination.²⁴ Nevertheless, demystifying such beliefs may be difficult because it may necessitate countering myths with evidence-based messages rather than traditional health education and directing-based approaches.³⁶ Besides, a study has revealed that side effects of COVID-19 vaccination can be helpful in preventing severe disease among the vaccinated.³⁷ Hence, it is also crucial to alleviate fears, build trust and encourage vaccine uptake by emphasizing that side effects are normal, beneficial and protective. Altogether, public health campaigns can increase vaccine acceptance and uptake by leveraging healthcare professionals’ recommendations and social proof, while addressing gender-specific concerns and barriers.

Comparison of the reasons given by women and men for not getting vaccinated reveals both similarities and differences. The most common reason for non-vaccination acceptance among both women (27.7%) and men (27.3%) was the fear of side effects of the vaccine. Besides, confusion regarding COVID-19 vaccine information is also a significant concern for both groups (13.7% among women and 15.7% among men). The perceived ineffectiveness of vaccines is another shared reason (10.9% among women and 14.7% among men).

The divergent cumbrances to COVID-19 uptake were lengthy queues leading to long waiting times at service points were constraints more concerning to men (10.8%) than women (6.4%) to deter them to go and receive vaccination. In addition, misconceptions regarding COVID-19 vaccine-related infertility was unique COVID-19 vaccination deterring factor more pronounced among men. On the other hand, lengthy travel distances and times is a more significant concern for women (6.6%) than men. Hence, improving COVID-19 vaccination uptake as well as any future pandemic, epidemic or outbreak vaccination related responses will require addressing any gender-specific concerns and barriers such as those highlighted in this study. For instance, alleviation of confusion and misconceptions about vaccine safety and effectiveness may be achieved through targeted health education campaigns to address any gendered misconception about vaccination and any other barriers. Additionally, mobile vaccination units or extended service hours, can help efforts aimed at addressing accessibility to vaccines brought about by geographic and time-related barriers. Promoting more inclusive vaccination strategies could benefit from addressing infertility misconception, a gender-specific concerns among men.

Conclusions and Recommendations

Despite the awareness of the pandemic in the study area located in Central Uganda, understanding of COVID-19 prevention measures was low, leading to poor adherence. While many had received at least one COVID-19 vaccine dose, completion rates were low, with disparities across districts and supervision areas. Fear of side effects, misinformation, and accessibility issues contribute to non-uptake. Targeted messages and ambassadors such as health workers, community leaders, and family members can help dispel myths and encourage vaccination. Interventions should prioritize poor-performing areas and indicators to improve coverage and uptake. By addressing these gaps, COVID-19 vaccination programs can increase effectiveness and reach more people.

Study Limitation

This cross-sectional study’s findings are specific to the time period and may not be generalizable due to the evolving global COVID-19 situation. Another study conducted at a different time may yield different results.

Acknowledgments

The authors declare that this study is part of a multi-indicator survey that includes non-COVID-19 data. As a result, other research may be published with the same study subjects and methodology but with different objectives.

Funding

This research was funded by the Centre for Disease Control, Kampala Office via Cooperative Agreement GH002046.

Disclosure

The authors report no conflicts of interest in this work.

References

1. Amodan BO, Bulage L, Katana E, et al. Level and determinants of adherence to COVID-19 preventive measures in the first stage of the outbreak in Uganda. Int J Environ Res Public Health. 2020;17(8810). doi:10.3390/ijerph17238810

2. Arce JSS, Warren SS, Meriggi NF, et al. COVID-19 vaccine acceptance and hesitancy in low- and middle-income countries. Nat Med. 2021;27:1385–1394. doi:10.1038/s41591-021-01454-y

3. Dubé E, Gagnon D, Nickels E, Jeram S, Schuster M. Mapping vaccine hesitancy — country-specific characteristics of a global phenomenon. Vaccine. 2014;32(49):6649–6654. doi:10.1016/j.vaccine.2014.09.039

4. Forman R, Shah S, Jeurissen P, Jit M, Mossialos E. COVID-19 vaccine challenges: what have we learned so far and what remains to be done? Health Policy. 2021;125(5):553–567. doi:10.1016/j.healthpol.2021.03.013

5. WHO Africa. Africa needs to ramp up COVID-19 vaccination six-fold. 2022. Available from:. https://www.afro.who.int/news/africa-needs-ramp-covid-19-vaccination-six-fold. Accessed May 17, 2022.

6. Idris IO, Ayeni GO, Adebisi YA. Why many African countries may not achieve the 2022 COVID-19 vaccination coverage target. tropical Medicine and Health. 2022;50(1). doi:10.1186/s41182-022-00407-6

7. Bono SA, de M VEF, Siau CS, et al. Factors affecting COVID-19 vaccine acceptance: an international survey among low-and middle-income countries. Vaccines. 2021;9(5):1–19. doi:10.3390/vaccines9050515

8. Harding E, Beckworth C, Fesselet JF, Lenglet A, Lako R, Valadez JJ. Using lot quality assurance sampling to assess access to water, sanitation and hygiene services in a refugee camp setting in South Sudan: a feasibility study. BMC Public Health. 2017;17(1):1–11. doi:10.1186/s12889-017-4656-2

9. Odaga J, Henriksson DK, Nkolo C, et al. Empowering districts to target priorities for improving child health service in Uganda using change management and rapid assessment methods. Glob Health Action. 2016;9(1):30983. doi:10.3402/gha9.30983

10. Shittu E, Adewumi F, Ene N, Keluo-Udeke SC, Wonodi C. Examining psychosocial factors and community mitigation practices to limit the spread of COVID-19: evidence from Nigeria. Healthcare. 2022;10(3):585. doi:10.3390/healthcare10030585

11. Valadez JJ, Bamberger M. Monitoring and Evaluating Social Programs in Developing Countries: A Handbook for Policymakers, Managers, and Researchers. Valadez J, Bamberger M eds.. The World Bank; 1994.

12. World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. J Am Med Assoc. 2013;310(20):2191–2194

13. Storer E, Jones T. Key considerations: adherence to COVID-19 preventive measures in greater Kampala, Uganda. Med Anthropology. 2022;41:387–403. doi:10.1080/01459740.2022.2047675

14. Machingaidze S, Wiysonge CS. Understanding COVID-19 vaccine hesitancy. Nat Med. 2021;27(8):1338–1339. doi:10.1038/s41591-021-01459-7

15. Bhopal S, Nielsen M. Vaccine hesitancy in low- and middle-income countries: potential implications for the COVID-19 response. Arch Dis Child. 2021;106(2):113–114. doi:10.1136/archdischild-2020-318988

16. Sallam M. COVID-19 vaccine hesitancy worldwide: a concise systematic review of vaccine acceptance rates. Vaccines. 2021;9(160):160. doi:10.3390/vaccines9020160

17. The Republic of Uganda. COVID-19 Vaccination Progress 15th. March 2022.

18. The Republic of Uganda. Uganda receives 864,000 doses of COVID-19 vaccines

19. Accorsi S, Fabiani M, Nattabi B, et al. Differences in hospital admissions for males and females in northern Uganda in the period 1992 — 2004: a consideration of gender and sex differences in health care use. Trans R Soc Trop Med Hyg. 2007;101(9):929–938. doi:10.1016/j.trstmh.2007.03.019

20. Zintel S, Flock C, Arbogast AL, Forster A, von Wagner C, Sieverding M. Gender differences in the intention to get vaccinated against COVID-19: a systematic review and meta-analysis. J Public Health. 2022;1–25. doi:10.1007/s10389-021-01677-w

21. Patwary MM, Alam MA, Bardhan M, et al. COVID-19 vaccine acceptance among low- and lower-middle-income countries: a rapid systematic review and meta-analysis. Vaccines. 2022;10(3):427. doi:10.3390/vaccines10030427

22. Kabagenyi A, Wasswa R, Nannyonga BK, et al. Factors associated with COVID-19 vaccine hesitancy in Uganda: a population-based cross-sectional survey. Int J Gen Med. 2022;15:6837–6847. doi:10.2147/IJGM.S372386

23. Nalubega P, Karafillakis E, Atuhaire L, et al. Maternal vaccination in Uganda: exploring pregnant women, community leaders and healthcare workers’ perceptions. Vaccines. 2021;9;1–10.

24. Kigozi A, Greener C. Access to COVID-19 vaccines for refugees in Uganda. Oxfam International. 2022. doi:10.21201/2022.6849.Oxfam

25. Lugada E, Komakech H, Ochola I, Mwebaze S, Oteba MO, Ladwar DO. Health supply chain system in Uganda: current issues, structure, performance, and implications for systems strengthening. J Pharm Policy Pract. 2022;15(14):1–11. doi:10.1186/s40545-022-00412-4

26. USAID. Accelerating success: US government support enables dramatic vaccination gains in Uganda. 2022.

27. Drivers of the COVID-19 vaccination process in Ugandan communities. 2021.

28. Jayawardana S, Esquivel M, Orešković T, Mossialos E. Gender differences in COVID-19 preventative measures and vaccination rates in the United States: a longitudinal survey analysis. Vaccine. 2024;42:126044. doi:10.1016/j.vaccine.2024.06.012

29. Zhang R, Qiao S, McKeever BW, Olatosi B, Li X. Listening to voices from African American communities in the Southern States about COVID-19 vaccine information and communication: a qualitative study. Vaccines. 2022;10(7). doi:10.3390/vaccines10071046

30. Adams J, MacKenzie MJ, Amegah AK, et al. The conundrum of low covid-19 mortality burden in sub-saharan Africa: myth or reality? Glob Health Sci Pract. 2021;9(3):433–443. doi:10.9745/GHSP-D-21-00172

31. de Vries DH, Bruggeman J, Benoni TE, et al. Social networks for health communication in rural Uganda: a mixed-method analysis of dekabusa trading centre, Luwero County. Glob Public Health. 2020;15(11):1674–1688. doi:10.1080/17441692.2020.1775870

32. Osuagwu UL, Mashige KP, Ovenseri-Ogbomo G, et al. The impact of information sources on COVID-19 vaccine hesitancy and resistance in sub-Saharan Africa. BMC Public Health. 2023;23(1):1–16. doi:10.1186/s12889-022-14972-2

33. Mphepo KYG, Muula AS, Suzi J, Phuka F, Mfutso-Bengo J. Exploring culturally-preferred communication approaches for increased uptake of voluntary medical male circumcision (VMMC) services in rural Malawi. BMC Public Health. 2023;23(1):590–606. doi:10.1186/s12889-023-15363-x

34. Tran BX, Dang AK, Thai PK, et al. Coverage of health information by different sources in communities: implication for COVID-19 epidemic response. Int J Environ Res Public Health. 2020;17(10):2–12. doi:10.3390/ijerph17103577

35. U.S. Department of Health and Human Services Centers for Disease Control and Prevention. COVID-19 vaccination field guide: 12 Strategies for Your community.

36. Lewandowsky S, Cook J, Schmid P, et al. The COVID-19 vaccine communication handbook. Practical Guide Improving Vaccine Communication Fighting Misinformation. 2021.

37. Drury RE, Camara S, Chelysheva I, et al. Multi-omics analysis reveals COVID-19 vaccine induced attenuation of inflammatory responses during breakthrough disease. Nat Commun. 2024;15(1). doi:10.1038/s41467-024-47463-6