The study was conducted at the National Laboratory of Clinical Biology and Public Health in Bangui, Central African Republic, the main reference center for HIV diagnosis in the country. The laboratory is equipped to perform serological testing, fourth-generation ELISA, rapid diagnostic tests, and HIV-1 PCR, enabling full implementation of the diagnostic algorithm under study.

This was a descriptive and analytical cross-sectional study conducted over nine months, from December 2024 to August 2025, aimed at evaluating the performance of an optimized HIV screening algorithm.

The target population included individuals aged 18 months and older presenting for HIV testing, with or without clinical symptoms suggestive of infection.

Participants (or parents/legal guardians for minors) who provided informed consent, agreed to blood sample collection, and were not receiving antiretroviral therapy were included.

Individuals who refused participation, those with insufficient biological samples, and patients already diagnosed with HIV and receiving antiretroviral therapy were excluded.

An exhaustive sampling strategy was used, including all eligible participants presenting during the study period, to obtain a comprehensive assessment of algorithm performance.

Biological samples consisted of whole blood collected in EDTA and plain tubes. Serological testing included Determine^TM and Uni-Gold^TM rapid tests and a fourth-generation ELISA. HIV-1 PCR was performed for discordant or indeterminate cases. All tests were conducted according to standardized laboratory protocols and WHO recommendations.

Ethical Approval and Consent to Participate

The present study was reviewed and approved by the Doctoral School of Science and Technology of the University of Bangui (EDST).

All procedures were conducted in accordance with the applicable national ethical standards as well as the international principles outlined in the Declaration of Helsinki of the World Medical Association (2013 revision).

The study adhered to the fundamental principles of biomedical research ethics, including respect for human dignity, participant autonomy, beneficence, non-maleficence, and justice.

Consent to Participate

Written informed consent was obtained from all participants prior to their inclusion in the study.

Before signing the consent form, participants received clear and detailed information regarding:

the objectives of the study;the procedures involved;the potential benefits and risks;the confidentiality of the collected information;their right to refuse participation or to withdraw from the study at any time without any consequences for their medical care.

For minors or legally incapacitated participants, consent was obtained from their legal representatives in accordance with applicable ethical requirements.

Confidentiality and Data Protection

The collected data were anonymized using unique identification codes. No directly identifiable participant information was included in the analytical databases.

Access to the data was strictly limited to authorized members of the research team. Electronic data were password-protected, and physical documents were stored in a secure location.

Data were collected using standardized forms. Statistical analysis was performed using EPI Info version 7. Results were presented as frequencies, percentages, means, and standard deviations. Algorithm performance was evaluated using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), with statistical significance set at p < 0.05.

During the study period, a total of 500 individuals underwent HIV screening at the National Laboratory of Clinical Biology and Public Health in Bangui. Of these, 120 participants (24.0%) tested reactive using the fourth-generation ELISA, while 380 (76.0%) were non-reactive.

All ELISA-reactive individuals were included in the diagnostic accuracy analysis. Among ELISA-negative participants, 90 individuals (23.7%) reported recent HIV exposure within the previous six weeks and were scheduled for follow-up testing according to the study protocol. However, follow-up results were not available at the time of analysis and were therefore not included in the final evaluation.

The flow of participants and distribution of ELISA results are summarized in Table 1 and Table 2.

Diagnostic accuracy was assessed using a composite reference standard, defined as follows:

concordant positive rapid diagnostic test results (DetermineTM and Uni-Gold^TM), orHIV-1 PCR positivity in cases with discordant serological results

Individuals with negative PCR results in discordant cases were classified as uninfected.

The study population was predominantly female (58.0%), with the largest age group being 25 - 34 years (37.0%). A recent risk exposure (<6 weeks) was reported in 36.0% of participants, indicating a substantial proportion at risk of early infection.

Sociodemographic and clinical characteristics of participants are summarized in Table 1.

The population was predominantly female (58%), with a median age in the 25 - 34-year group. Recent risk exposure was reported by more than one-third of participants (36%), justifying the use of diagnostic tools capable of detecting very early infections.

Table 1.Sociodemographic and clinical characteristics of the study population (n = 500).

Among the 500 participants, 120 (24.0%) were ELISA reactive, while 380 (76.0%) were non-reactive (Table 2). Among ELISA-negative individuals, 90 (23.7%) reported recent exposure and were scheduled for repeat testing.

Fourth-generation HIV ELISA results

The results of the initial fourth-generation ELISA screening are presented in Table 2.

Table 2.Initial screening results by fourth-generation HIV ELISA.

Among the 380 ELISA-negative patients, 90 (23.7%) reported recent exposure and were scheduled for retesting after 2 - 4 weeks.

Serological confirmation using rapid diagnostic tests

Among the 120 ELISA-reactive individuals, 95 (79.2%) had concordant positive results on Determine^TM and Uni-Gold^TM, confirming HIV infection. However, 25 cases (20.8%) showed discordant or indeterminate results (Table 3).

Among the 25 discordant cases, HIV-1 RNA was detected in 18 individuals (72.0%), while 7 (28.0%) had undetectable viral RNA (Table 4).

These PCR-positive cases were classified as confirmed HIV infections, likely corresponding to early-stage infections not fully detectable by serological assays.

The serological confirmation results obtained using rapid diagnostic tests are shown in Table 3.

Table 3.Serological confirmation results.

These findings show that nearly one in five patients had serological discordance, highlighting the limitations of serological confirmation alone.

The distribution of HIV-1 PCR results among discordant cases is detailed in Table 4.

Table 4.HIV-1 PCR results among discordant patients (n = 25).

Seventy-two percent of discordant cases corresponded to very early HIV infections detectable only by PCR.

After applying the complete diagnostic algorithm, a total of 113 individuals (22.6%) were classified as HIV-positive, while 387 (77.4%) were considered HIV-negative (Table 5).

Among confirmed infections, 18 cases (15.9%) were identified exclusively through PCR among discordant samples and would have been missed using a purely serological algorithm.

Table 5.Final HIV status after full application of the algorithm.

Among confirmed cases, 18 (15.9%) were very early infections that would have been missed by a strictly serological algorithm.

Compared to the standard algorithm based solely on rapid diagnostic tests, the optimized approach detected more HIV cases (113 vs. 95) and substantially reduced the number of false-negative results (2 vs. 18).

The estimated median diagnostic delay was reduced from 28 days to 14 days (Table 6). Overall, this corresponds to a relative increase in case detection of 18.9%.

Table 6.Comparison of diagnostic performance.

The optimized algorithm yielded a relative detection increase of 18.9%.

The diagnostic performance of the optimized algorithm was assessed using a contingency table (Table 7).

True positives: 111False positives: 3True negatives: 384False negatives: 2

These results indicate a high diagnostic performance of the optimized algorithm.

Table 7.Performance indicators of the optimized algorithm.

Although the optimized algorithm demonstrated high sensitivity and specificity, these findings should be interpreted with caution due to the use of a composite reference standard and the absence of systematic PCR testing in all participants.

This study has several limitations. Its single-centre design may limit generalisability. PCR was performed only in discordant cases, introducing potential verification bias. Missing follow-up data in ELISA-negative participants may underestimate early infections. The composite reference standard may lead to misclassification, and no cost-effectiveness or feasibility analysis was conducted.