Our findings show that 11.58% of clients of the Manhiça Hospital VCT in southern Mozambique may be considered to have a recent HIV infection. There was a higher prevalence in women than in men. Assessment of HIV viral load setpoint in recently infected individuals revealed great heterogeneity of HIV-RNA viral load from which two patterns emerged. Individuals with HIV viral load above the median at enrollment (>4.98 log10 copies/mL) sustained a high HIV-RNA level and a setpoint of 5.22 log10 copies/mL. This contrasted with individuals with an initial HIV viral load below the median and a viral load setpoint of 4.15 log10 copies/mL.
The prevalence of recently infected HIV patients (11.58%) among HIV-seropositive VCT clients in Manhiça was similar to that of other African countries 
but lower than that observed in European countries such as Switzerland where a prevalence of 37% of recent HIV infections was reported in VCT clients as determined by the BED-CEIA 
. A low prevalence of recent infections suggests that a large proportion of the Manhiça population attends the VCT at later stages of HIV/AIDS. Thus, recently infected individuals, who are more likely to transmit, may have lower VCT uptake. This does not include pregnant women who undergo HIV testing in the Manhiça antenatal clinic (ANC). Clients of the ANC are younger with a median age of 24 years and may have a higher prevalence of recent HIV infections than those attending the general VCT 
. Late diagnosis of HIV has been observed in other African countries, and it is likely due to a poor access to health care in these settings 
. Late presentation is associated with a higher mortality after ART initiation 
and this has led to piloting new HIV testing approaches to increase uptake, including mobile units, provider-based and home-based counselling and testing 
The gender and age distribution of recent HIV infection observed in our study is likely to be a reflection of the HIV epidemic and trends in VCT uptake in the area. We observed that the prevalence of recent HIV infections was significantly higher in women as compared to men, corroborating findings of increased HIV prevalence in women in sub-Saharan Africa 
. In our study, recently infected men were significantly older, almost 10 years when comparing with recently infected women. This may reflect migration trends for young healthy men from Manhiça working in South Africa 
. This would most probably lead to underrepresentation of young men in the recently infected population, and overrepresentation in more advanced stages of disease when HIV progression may impede seeking employment in South Africa. These age and gender distributions suggest that other approaches to HIV testing are needed to increase earlier diagnosis of HIV in all age groups with particular attention to VCT uptake in men.
In terms of clinical signs and symptoms that could potentially differentiate recent from long-standing HIV infections, we did not observe any non-specific self-reported signs associated with either. It has been shown that nonspecific flu-like syndromes may help to identify patients during the acute phase of the infection which lasts up to 6 weeks 
. However, although patients may have high HIV viral loads over a longer period of time, nonspecific patient-reported symptoms do not appear to distinguish recent from longstanding HIV infected patients in our population. This study did not include clinical exploration for opportunistic infections.
HIV viral load setpoint has been loosely defined as a stable viral load level established within the first year after seroconversion 
. However, in the literature the establishment time of this HIV viral load setpoint has been defined at various time points ranging from 4 months to 24 months post-infection 
. We estimated the HIV viral load setpoint over the 1st
year of HIV infection. In our study population, HIV-RNA levels of patients classified as recently infected remained elevated throughout the follow up. In addition, those with HIV-RNA levels greater than the median (4.98 log10
copies/mL) at diagnosis maintained HIV-RNA levels above 4.98 log10
copies/mL during the first year after infection with an estimated viral load setpoint of 5.22 log10
copies/mL. These individuals may be more rapid progressors and could have an increased risk of transmitting HIV. There is indeed agreement that higher HIV-RNA setpoints are associated with faster progression to AIDS and increased risk of HIV transmission 
. Individuals with HIV viral load above 4.91 log10
copies/mL have been suggested to progress to AIDS as quickly as 3 years after infection 
, and those patients with HIV-RNA levels above 4.7 log10
(50000) copies/mL may have the highest HIV transmission rates 
It has recently been suggested that individuals maintaining HIV-RNA viral loads greater than 4.7 log10
(50000) copies/mL could be a target for antiretroviral treatment for prevention to reduce HIV transmission 
. The percentage of individuals with HIV viral loads above 4.7 log10
copies/mL in Botswana was suggested to be approximately 25%–30% both in the general population and in antenatal clinics tested by rapid serology tests 
. According to this approach, up to 50% of recently infected individuals in the present study could be candidates for antiretroviral treatment for prevention and suggests that screening for recent HIV infections may identify a greater proportion of individuals with elevated viral load.
Our results showed that patients with elevated viral load had higher CD8 T-cell counts and lower CD4
CD8 T-cell ratios at diagnosis. Other studies have suggested that within HIV-1 subtype C infected patients, those individuals who maintain a higher viral load setpoint display lower levels of CD4 cells 
. We did not observe these low levels of CD4 counts in the group maintaining higher HIV-RNA viral load, however these patients showed decreased CD4
CD8 ratios over the year of follow-up.
This study had several limitations. Although the BED-CEIA assay is widely used tool, it has not been validated for subtype C. It has been suggested that the BED-CEIA period of recency may be wider for subtype C than for subtype B. The recency window for subtype B is 162 days and could be up to 203 days for subtype C 
. Nevertheless, several studies have used BED-CEIA for measuring recent infections and estimating incidence in countries where HIV subtype C is the major subtype 
. We considered a broader recency window of up to 8 months. However it is important to emphasize that the present study did not have the aim of measuring incidence but only a prevalence of recent infections in the area.
Another limitation is that BED-CEIA is known to overestimate the proportion of recently infected individuals which may lead to misclassification of advanced AIDS as recent infections 
. To avoid this misclassifications a false positive rate (FPR) can be calculated for each population studied. However, the calculation requires comparison with longitudinal incidence studies, which were not available for the Manhiça population. We thus used CD4 counts and HIV-RNA levels as surrogates of FPR. Hence, our results were based on the assumption that BED<0.8n O.D., CD4>200 cells/µl and VL>400 copies/mL identifies recent infection. We, as other authors, considered those individuals reading as recent HIV infections by BED-CEIA but with CD4<200 cells/µl or VL<400 copies/mL as false recent infections 
. This is based on the assumption that the production of HIV-specific IgG decreases when viral replication is suppressed or in the presence of severe immunosuppression. This in turn can lead to a low proportion of specific HIV-IgG relative to total IgG giving a false recent infection by the BED-CEIA 
. Nevertheless, we cannot exclude that a small number of advanced AIDS patients with CD4>200 cells/µl or VL>400 copies/mL were misclassified as recent infections which could lead to an overestimation of both the prevalence of recent infected patients and median HIV-RNA levels. On the other hand, the potential exclusion of recently infected patients with CD4<200 cells/µl or VL<400 copies/mL could lead to underestimating the prevalence of recent infected patients. However this is less likely. Finally, the study suffered a large loss to follow-up during the 10 months of the study leading to a low sample size at 10 months of followup. However, the loss was equal in both high HIV-RNA patients and those with lower HIV-RNA levels.
In addition to the use of BED-CEIA and FPR, other approaches such as antibody avidity tests or other detuned ELISA have been used to estimate recent infections alone or in combination 
. The use of a second avidity test to identify recent infections diminishes the likelihood of misclassifications 
, however it may require specific technology which can limit its use in a developing country. The approach of using the BED-CEIA assay and CD4<200 cells/µl and VL<400 copies/mL as surrogate markers for FPR may facilitate the identification of recent infections in situ using the HIV monitoring tools already available or being scaled up in many low income countries.
In summary, the present study provides data on recent HIV infections and viral load setpoint from an area with little previous information. Our results suggest that in a rural area of southern Mozambique, a low proportion of individuals seek HIV testing at early phases of HIV infection and points to a need for new approaches to HIV testing. Our results also identify a group of patients in early phases of HIV infection with an elevated HIV-RNA setpoint greater than 5.0 log10 copies/mL which could be target for ART for prevention strategies.