Introduction

The Human Immunodeficiency Virus (HIV) represents a persistent viral disease that causes progressive immune system damage through CD4 + T lymphocyte destruction, thus allowing opportunistic infections to develop1. The natural course of HIV infection leads to Acquired Immunodeficiency Syndrome (AIDS) within about ten years unless patients receive sufficient antiretroviral therapy (ART), although the timeline varies based on individual and viral characteristics2,3. The introduction of highly active antiretroviral therapy has led to better survival outcomes and improved quality of life for people who have HIV. The early detection of HIV stands as an essential foundation for successful HIV treatment4. The early detection of HIV enables patients to start ART immediately, which leads to lower death rates and disease complications and prevents additional virus spread5. Early HIV diagnosis means detecting the virus within six months of infection, but late diagnosis occurs when patients have either a CD4 count under 350 cells/µL or an AIDS-defining condition during diagnosis6,7. Early HIV diagnosis, together with prompt antiretroviral therapy implementation, even at advanced disease stages, continues to be a major problem in multiple nations8. Late diagnosis remains a substantial public health challenge in low- and middle-income countries despite worldwide progress in HIV testing and awareness9. The prevalence of late HIV diagnosis in sub-Saharan Africa ranges between 11% and almost 90% because this region carries the majority of the worldwide HIV epidemic10. Research indicates that 68.8% of HIV-positive individuals in Ethiopia receive their diagnosis at a late stage, which demonstrates the critical need to address this problem11. The health outcomes of patients with late HIV diagnosis become worse because they face a tenfold higher mortality rate during the first year of diagnosis and increased medical expenses, as well as extended periods during which they can transmit HIV12. People postpone HIV testing because they lack awareness about the disease, believe they have minimal risk of infection, and because of HIV-related stigma and difficulties accessing healthcare facilities13. The problem becomes worse when people are older, have less education, and lack sufficient knowledge about HIV. The delay in HIV diagnosis occurs because people hold incorrect beliefs about ART treatment and maintain negative views about individuals who have HIV/AIDS4,6,14.

The Ethiopian government runs multiple initiatives for improving HIV prevention, testing, and treatment, which include provider-initiated testing and counseling and public awareness campaigns. The availability of diagnostic tools throughout the country has not prevented the high rate of late HIV diagnosis, which particularly affects underserved regions15. Most of the previous studies had been carried out several years prior and were limited to single hospitals, making it difficult to isolate the determinants. Updated context-specific data are needed. Moreover, recent localized data regarding the factors behind late HIV diagnosis in the West Gojjam Zone of the Amhara Region remain insufficient. Designing targeted interventions for early testing and improved linkage to care requires knowledge of the specific barriers and contributing factors in this context.

Therefore, this study aimed to determine the factors that cause late HIV diagnosis among HIV-positive patients who receive ART services in selected hospitals from West Gojjam Zone, Northwest Ethiopia by addressing the gaps through generating recent evidence-based, multi-hospital, and context-specific data using an unmatched case-control study, specifically, the study sought to answer the following research questions:

  1. 1.

    What socio-demographic factors are associated with late diagnosis among adults attending ART clinics in the study area?

  2. 2.

    What clinical, behavioural, and sociocultural factors are associated with late HIV diagnosis among people living with HIV in the west Gojjam Zone?

The research examines these factors to develop public health approaches that will increase prompt access to HIV care throughout the region.

Method and material

Study area and period

This study was conducted from February to March 2025 in public hospitals located in West Gojjam Zone, Amhara Region, Northwest Ethiopia. West Gojjam is one of the 13 administrative zones of the Amhara Regional State and is geographically bordered by the Abay River to the south, the Awi Zone to the west, North Gojjam to the north, and East Gojjam to the east. Among the four hospitals in the zone, three provide antiretroviral therapy (ART) services: Finote Selam General Hospital, Bure Primary Hospital, and Feresbet Primary Hospital. Dembecha Primary Hospital was excluded because it does not offer ART services. These facilities also deliver provider-initiated testing and counseling (PITC) and prevention of mother-to-child transmission (PMTCT) services. The hospitals are staffed with trained ART professionals: Finote Selam General Hospital has 10, Bure Primary Hospital has 7, and Feresbet Primary Hospital has 2. At the time of the study, a total of 2,522 HIV-positive patients were receiving ART in these hospitals 1,900 at Finote Selam, 530 at Bure, and 92 at Feresbet.

Study design

An institution-based unmatched case-control study design was employed to identify determinants of late HIV diagnosis.

Population

Source of population

The source population comprised all HIV-positive patients attending ART clinics in public hospitals of West Gojjam Zone during the study period.

Study population

Cases

HIV-positive individuals who were diagnosed late, defined as having a baseline CD4 count < 350 cells/µL or WHO clinical stage III or IV at the time of first diagnosis, and who were receiving ART follow-up during the data collection period.

Controls

HIV-positive individuals aged 18 years or older who had a baseline CD4 count ≥ 350 cells/µL or WHO clinical stage I or II at the time of first diagnosis, and who were also receiving ART.

Eligibility criteria

Inclusion criteria

Participants were eligible for inclusion if they met the following criteria:

  • HIV positive individuals aged 18 years or older.

  • Currently receiving ART follow-up during the data collection period in public hospitals of West Gojjam Zone.

  • Documented baseline CD4 count or WHO clinical stage at the time of enrolment into the ART clinic.

Exclusion criteria

Participants were excluded from the study if:

  • Baseline CD4 count or WHO clinical stage was not documented at the time of ART enrolment.

  • They were unconscious or unable to provide informed consent during the data collection period.

Sample size determination and sampling technique

Sample size determination

The sample size was calculated using Epi Info version 7.2.5.0, based on the following assumptions: 95% confidence interval (CI), 80% power, case-to-control ratio of 1:1, Proportion of history of chronic illness, as having at least one medically diagnosed long-term non-communicable condition before getting HIV which could be any of the following: diabetes mellitus, hypertension, cardiovascular disease, chronic respiratory disease (like asthma), chronic kidney disease, or cancer among controls and cases was 40.4% and 59.6%, respectively, based on findings from a previous study conducted in Mekele16, a non-response rate of 10% rate. Using these parameters, the final sample size was 255 participants, comprising 127 cases and 128 controls.

Sampling technique

A proportional allocation strategy was employed to distribute the total sample size across the three hospitals providing ART services, based on their respective ART client loads. The formula used for proportional allocation was:

figure a

Where:

  • ni = Sample size allocated to each hospital.

  • Ni = Total number of ART clients in each hospital.

  • N = Total ART client population across all hospitals.

  • n = Overall sample size.

HIV-positive individuals attending ART clinics were classified into two groups based on their baseline CD4 count or WHO clinical HIV staging at the time of initial presentation. A case-based control selection approach was applied, whereby one control was selected for each case using a simple random sampling method. The ART registration book served as the sampling frame for both cases and controls (Table 1).

Table 1 Distribution of ART clients and study participants by hospital in West Gojjam zone, Northwest Ethiopia 2025.
Full size table

Study variables

Dependent variable

  • Late HIV diagnosis.

Independent variables

  • Socio-demographic factors: Age, sex, educational status, place of residence, marital Status, employment status, religious, monthly average income of participants and lliving arrangement.

  • Clinical factors: Decision for HIV testing, presence of symptoms before diagnosis, having chronic health problems and Reason for HIV testing.

  • Sociocultural and Behavioural factors: Contact with commercial sex worker, substance use, perceived stigma, compressive knowledge about HIV/ADS, attitude towards HIV/ADS and knowledge about ART.

Operational definition

Delayed HIV diagnosis

An HIV-positive patient has a lymphocyte CD4 + count < 350 cells/L at the time of diagnosis or, regardless of the lymphocyte count, presentation with an AIDS-defining event at the time of diagnosis14.

Comprehensive knowledge of HIV/AIDS: An individual who correctly answered all of the items was considered as having comprehensive HIV knowledge, but if s/he failed to answer at least 1 of them, s/he was considered as not having it4.

Attitude about HIV: Those scoring less than the mean scores for attitude were classified as having negative attitudes, and those scoring equal to or more than the mean scores were classified as having positive attitudes17.

Perceived HIV Stigma: Study participants who scored greater than or equal to the mean score from 10 stigma assessment questions were classified as having perceived stigma, while a score of less than the mean score was considered as having no perceived stigma18.

Alcohol use: Indicate if they had ever consumed alcohol before being diagnosed with HIV or not. Those who answered “yes” were counted as alcohol users, whereas the other group consisting of “no” respondents was called non-users14.

Data collection tools and procedures

The data collection tool was adapted from different literature13,15,17,18. A pre-tested structured and interviewer-administered questionnaire was used to collect the data. The questionnaire was divided into five sections, such as Socio-demographic characteristics, clinical and behavioral characteristics, comprehensive knowledge of respondents toward HIV, attitudes about HIV, and perceived stigma. The Socio-demographic characteristic section contains nine questions. The clinical and behavioral characteristics of respondents section contain sixteen questions13,15. The comprehensive knowledge of respondents toward HIV was assessed based on five items which is a standard and validated tool in the Ethiopia and were used in Ethiopian Demographic and Health Survey 201619. The attitudes toward HIV were assessed by five items with yes/no responses17, and perceived HIV stigma was assessed by a 10-item perceived HIV stigma scale assessment tool which contains 10 stigma assessment questions with Likert scale18.

Data quality assurance

The data collection tool was prepared first in English, then translated into the local Amharic language, and then translated back to English to check the consistency by language experts. One-day training about the objectives and process of the data collection was given to data collectors and supervisors. A trained supervisor was supervising the data collectors daily for the completeness and consistency of the filled questionnaires. Before the actual data collection, a pretest of the questionnaire was conducted on 5% of the sample size in a hospital out of the study area. The questionnaire was validated for its consistency by using Cronbach’s alpha test. In addition, the data was thoroughly cleaned and carefully entered into a computer for analysis.

Data processing and analysis

The data was checked, coded, cleaned, and entered into Epi Data version 4.6 before being exported to the Statistical Package for Social Sciences (SPSS) 25 for analysis. Categorical variables were computed by using frequencies and percentages, whereas for continuous variables with normal distribution, the mean and standard deviation were used. Frequencies were used to summarize descriptive statistics. Binary logistic regressions were used to determine the association between the outcome variable and explanatory variables. Adjusted odds ratios with their corresponding 95% CI were computed to determine the presence and strength of the association. All variables in the Bi-variable logistic regression model whose p-value is less than or equal to 0.25 were included in the multivariable analysis model and analyzed by using the backward likelihood ratio method. A p-value of < 0.05 was taken to declare the presence of statistical significance. Model fitness was tested with Hosmer and Lemeshow’s Goodness of Fit test and fit with a p-value of 0.801. Multi-collinearity was checked between independent variables by using the variance inflation factor, and there was no multi-collinearity problem with a variance inflation factor < 4 for all variables. The analyzed data was presented using text, figures, and tables.

Results

Socio-demographic characteristics of the study participants

Out of the 255 ART clients recruited, 251 of the participants (125 cases and 126 controls) completed the study, with a response rate of 98.2%. Among these respondents, 62.9% were female. The mean age of the cases and controls was 37.68 and 38.35 years, with standard deviations of 9.68 and 9.64, respectively. About 90 cases (72%) and 94 controls (74.6%) were urban residents. In terms of educational status, the highest proportion of cases (46 individuals, 36.8%) and controls (34 individuals, 27%) were unable to read and write (Table 2).

Table 2 Socio-Demographic characteristics of study participants in public hospitals of West Gojjam Zone, Amhara Region, Northwest Ethiopia, 2025.
Full size table

Clinical and behavioral characteristics of PLWHA

Among the study participants, 84 cases (67.2%) presented with WHO clinical stage III, whereas 87 controls (69%) were classified as WHO clinical stage I at the time of diagnosis. Regarding the decision to undergo HIV testing, 83 cases (66.4%) and 56 controls (44.4%) were tested through provider-initiated testing and counseling (PITC). Additionally, 69 cases (55.2%) and 24 controls (19%) reported that they sought HIV testing due to feeling ill. Prior to their HIV-positive diagnosis, 77 cases (61.6%) and 96 controls (76.2%) had heard about antiretroviral therapy (ART). Furthermore, 44 cases (35.2%) and 121 controls (96%) were aware that early initiation of ART improves survival outcomes (Table 3).

Table 3 Clinical and behavioral characteristics of study participants in public hospitals of West Gojjam Zone, Amhara Region, Northwest Ethiopia, 2025.
Full size table

Respondents’ Knowledge, attitude about HIV/AIDS, and their internalized stigma towards their HIV positive status

Among the participants, 20 cases (16%) and 61 controls (48.4%) demonstrated comprehensive knowledge about HIV. In terms of attitude, the majority of cases, 76 individuals (60.8%), and 49 controls (38.9%) exhibited a negative attitude toward people living with HIV. Additionally, 54 cases (43.2%) and 17 controls (13.5%) reported experiencing perceived stigma related to their HIV-positive status (Fig. 1).

Fig. 1
Fig. 1
Full size image

Knowledge, Attitudes Toward HIV/AIDS, and internalized stigma among study participants in public hospitals of west gojjam zone, amhara region, northwest Ethiopia, 2025.

Determinants of late HIV diagnosis

Binary logistic regression was used to identify determinants of late presentation of HIV diagnosis. Both bivariable and multivariable logistic regression analyses were used. From the bivariable logistic regression analysis, educational status, decision for testing HIV, hearing about ART, higher survival while taking ART, having contact with CSW, alcohol use, chat chewing, cigarette smoking, comprehensive knowledge about HIV/AIDS, attitude toward HIV/AIDS, and perceived HIV stigma were included in the multivariable logistic regression analysis to determine statistical significance with P-values < 0.05 combined with adjusted odds ratios (AOR) and their 95% confidence intervals (CIs).

The multivariable analysis revealed that decision-making for HIV testing, alcohol use, comprehensive knowledge about HIV/AIDS, and perceived HIV stigma were significantly associated with late HIV diagnosis. After adjusting for potential confounders, individuals whose HIV testing was initiated by friends, family, or social contacts were 3.54 times more likely to be diagnosed late compared to those who initiated testing themselves (AOR = 3.54; 95% CI: 1.34–9.36; p = 0.011). Similarly, individuals tested through provider-initiated testing and counseling (PITC) were 3.49 times more likely to be diagnosed late than those who sought testing voluntarily (AOR = 3.49; 95% CI: 1.70–7.20; p = 0.001). Alcohol users were approximately 3.96 times more likely to present late for HIV diagnosis compared to non-users (AOR = 3.96; 95% CI: 2.13–7.38; p < 0.001). Participants lacking comprehensive knowledge about HIV/AIDS were 3.24 times more likely to be diagnosed late than those who were knowledgeable (AOR = 3.24; 95% CI: 1.64–6.40; p = 0.001). Furthermore, individuals who reported perceived HIV-related stigma were 4.83 times more likely to experience late diagnosis compared to those without perceived stigma (AOR = 4.83; 95% CI: 2.39–9.74; p = 0.001) (Table 4).

Table 4 Bivariable and multivariable logistic regression analysis of determinants of late HIV diagnosis in public hospitals of West Gojjam Zone, Amhara Region, Northwest Ethiopia, 2025.
Full size table

Discussion

This case-control study investigated the determinants of late HIV diagnosis among adults attending public hospitals in West Gojjam Zone, Amhara Region, Northwest Ethiopia. The analysis identified several key factors significantly associated with late presentation for HIV diagnosis. The major determinants included the decision-making process for HIV testing, alcohol use, comprehensive knowledge about HIV/AIDS, and perceived HIV-related stigma. The study found that individuals whose HIV testing was decided or initiated by friends, family members, or socials were more likely to be diagnosed late compared to those who sought testing independently. Similarly, individuals who underwent HIV testing through provider-initiated testing and counseling were more likely to present late than those who initiated testing on their own. This finding was consistent with a previous study conducted at Bahir Dar Felege Hiwot Referral Hospital in Northern Ethiopia15. A possible explanation is that individuals whose testing is externally initiated may have lower perceived personal risk or less intrinsic motivation to seek care. Consequently, they may only undergo testing when external pressure becomes significant20. This study also revealed that alcohol users were significantly more likely to present late for HIV diagnosis compared to their counterparts. The result is consistent with findings from studies conducted in the West Arsi Zone14 and in Jima Southwest Ethiopia13. The possible explanation may be that alcohol consumption is often linked to reduced health-seeking behavior and poor risk perception. Individuals who frequently consume alcohol may underestimate their vulnerability to HIV, delay recognizing symptoms, or avoid health facilities due to stigma or fear of a positive diagnosis21. Additionally, alcohol use can impair judgment and decision-making, leading to postponement or neglect of important health actions, including timely HIV testing and diagnosis22. Furthermore, the study showed that individuals lacking comprehensive knowledge about HIV/AIDS were more likely to be diagnosed late compared to those who were knowledgeable. The finding is consistent with a study conducted in Debre-Markos and Finote-Selam Hospitals, in Northwest Ethiopia4, as well as in Venezuela23. This may be because a lack of knowledge can lead to underestimation of personal risk, misconceptions about HIV transmission, and fear of stigma, all of which can contribute to late health-seeking behavior24. The other variable found to be a significant determinant for late HIV diagnosis was having perceived HIV stigma. This finding is in line with studies conducted in Zimbabwe25, Bahir Dar Felege Hiwot Referral Hospital15, and West Arsi zone14. This may be because perceived stigma may discourage individuals from seeking timely testing and care due to fear of discrimination, social rejection, or breach of confidentiality26. As a result, individuals may postpone HIV testing until they experience symptoms or reach advanced stages of the disease. These findings have significant implications for HIV prevention and care strategies. The link between late HIV diagnosis and provider- or socially-initiated testing, alcohol use, poor HIV knowledge, and perceived stigma highlight the need to strengthen voluntary testing strategies, integrate substance use counseling into HIV services, and increase community-based education to improve comprehensive HIV knowledge. Furthermore, tackling HIV-related stigma through public awareness campaigns and health professional training is critical for encouraging timely testing. These findings lend support to the creation of targeted, stigma-sensitive treatments and policies that encourage early diagnosis, enhance access to care, and, ultimately, reduce HIV-related morbidity, mortality, and transmission in the community.

Conclusions

The significant determinants of late HIV diagnosis among individuals attending ART clinics in public hospitals of West Gojjam Zone, Northwest Ethiopia included that HIV testing initiated by friends, family, or social contacts; provider-initiated testing and counseling; alcohol use; limited knowledge about HIV/AIDS; and perceived HIV-related stigma were significantly associated with late presentation for diagnosis. These findings highlight the complex interplay between behavioral, social, and informational factors that contribute to delayed HIV diagnosis.

Recommendations

To reduce the burden of late HIV diagnosis, targeted public health interventions are needed. Efforts should focus on promoting self-initiated voluntary HIV testing and reducing reliance on externally driven testing decisions. Community-based education campaigns should be strengthened to improve comprehensive knowledge about HIV/AIDS and the benefits of early ART initiation. Integrating substance use counseling into HIV services may help mitigate the impact of alcohol-related delays in health-seeking behavior. Additionally, addressing HIV-related stigma through public awareness initiatives and healthcare provider training is critical to fostering a supportive environment for early testing and care. These strategies can contribute to improved linkage to care, reduced transmission, and better clinical outcomes for individuals living with HIV.

Strength of the study

To begin with, the application of an unmatched case–control design made it possible to conduct a very efficient search for the precursors of late HIV diagnosis. Moreover, the fact that the research was conducted in several public hospitals and the sample was divided up in proportion to the ART clients in the different hospitals further increased the representativeness and generalizability of the results to the entire West Gojjam Zone. Last but not least, the study measured key variables such as HIV knowledge and perceived stigma using standardized and previously validated instruments, which contributed to the reliability of the measurements.

Limitations of the study

Despite the strengths, this study has some limitations. There might be recall and social desirability bias from the participants because the participants’ responses depended on recalling retrospectively what occurred over a long period and in a socially acceptable manner, instead of providing socially acceptable answers. It is impossible to the complete ruled out of selection bias since the cases and controls were obtained from health institutions, which could result in restriction of generalization to the overall population. With the case control design, it is not possible to establish the time sequence or causation between the exposure and the outcome.

Public health implications

The results of this particular study carry significant implications for public health and health programs respectively. By recognizing provider-initiated and socially initiated testing, alcohol use, lack of knowledge about HIV, and perceived stigma as contributors to late diagnosis of HIV, interventions are required which will lead to the acceptance of voluntary HIV testing, community education and awareness, and cutting down on stigma associated with HIV. Moreover, the incorporation of substance use screening and counseling into HIV testing and care services can also be a good way to eliminate delayed diagnosis. These findings will help the formation of strategies by policymakers, program planners, and healthcare providers that are targeted, context-specific, and thus more effective in improving early HIV diagnosis and timely linkage to care in the respective regions.