In this study, we found a high prevalence of transfusion-transmissible infections in blood donor candidates of the provincial hospital of Tete, Mozambique, and in particular in replacement donors. Seroprevalence of HIV, HBsAg, and syphilis was 8.5%, 10.6 %, and 1.2% respectively, and no donor was found with confirmed HCV infection. We observed also that the questionnaire had a selective value for candidate donors with HBV infection, but did not discriminate those infected with HIV or syphilis. Finally, the rapid test for HBV screening used by the national program had an unacceptably low specificity, resulting in a high proportion of donations needlessly rejected.
Several limitations are to be mentioned. The study was conducted according to the routine practice of the blood bank, with no additional staff or equipment. Full time supervision (24 hours a day, 7 days a week) by the main investigator was not possible, and errors in performing or reporting may have occurred. Also, rapid tests were compared in parallel and in a non-blinded way, with some risk of "cross-influence" on the results. As mentioned, sophisticated testing could only be performed in Maputo, representing considerable logistical efforts. For this reason, reference testing was not performed on the whole sample set but purposely restricted to the samples with discordant results in Tete, to all samples reactive to HCV rapid test and to the subset of positive and negative samples sent for quality control. Some erroneous results of rapid testing might therefore have been missed, although the very satisfactory results of the quality control suggest that our findings were robust. Finally, the adequacy of clinical indication for transfusion has not been investigated here; also the study was not designed to assess accurately the further uptake of candidate donors diagnosed with any TTI within clinical care programs.
Little is known on the seroprevalence of transfusion-transmissible infections among blood donors in Mozambique. In one such study conducted in Maputo in 2004, the HIV prevalence was higher (13.8%)  than here, reflecting probably the regional differences in "background" HIV endemicity in Mozambique. In fact, our findings are in line with those of the most recent population survey in the province of Tete (7.0% in adults in 2009) . Regarding the prevalence of HBsAg carriage in our study population, it was similar to that observed in blood donors in Maputo (9.3%)  and in neighboring Malawi (8.1%) . For syphilis, little is known among Mozambican blood donors. The prevalence we observed is much lower than that obtained during the last sentinel surveillance round of 2007 among pregnant women in Tete , but this may be partly explained by the differences in diagnostic design: RPR testing is indeed less specific in pregnant women , and no confirmation test was used. Finally, no HCV infection was confirmed in this study, suggesting that HCV is not a major health problem in the province. This is in line with results from Maputo and Malawi where the prevalence of active HCV infection confirmed by molecular methods was below 1% in both sites [20, 23]. Of note, anti-HCV EIA disclosed also about 10% of false positive results that could not be confirmed with the INNO- LIA HCV, like observed elsewhere in Africa [20, 23, 25]. Since this study, HCV screening of blood donation has been implemented in Tete, and since 2010 in the rest of the country as well, in accordance with the 2008 WHO recommendation for universal HCV testing . However cost-effectiveness remains an issue for low-resource settings with very low HCV prevalence [25, 26]. In the blood bank of Tete, only the cost of the rapid HCV screening tests represents an additional amount of at least 7,000 USD annually, for a few prevented infections. Although innovative testing strategies could be explored in Tete , it is likely that HCV screening could become really cost-effective only if the prices of available rapid tests were drastically reduced.
The proportion of young voluntary donors was rather high in comparison with other African countries where it reaches only 20-30% of the blood donors [28, 29]. Not surprisingly, they were less often infected by blood-borne pathogens [28–30]. Although already important, the efforts provided by the Tete blood bank managers to select low-risk groups are for sure still worth being amplified.
In contrast, the selection power of the questionnaire was somehow disappointing. Hepatitis B carriage was rather well discriminated (mainly because of the question 4), but HIV infection and syphilis were as prevalent in the deferred as in the non-deferred group. We did not investigate specifically this issue, but reasons contributing to this low performance may include the rather time-consuming, vague and redundant questionnaire design, favoring a speedy and inadequate question-answer process, the lack of privacy to answer honestly to some delicate questions or the difficulties for both the candidate donors and the blood bank staff in interpreting/translating in local language some questions not immediately obvious or not culturally adapted to rural Africa (questions 3, 5-8, 14). We suggest to both the local and national health authorities to design and implement a shorter, simpler and more contextualized screening questionnaire, and to re-evaluate its discriminative value in similar real-life settings.
This study has also highlighted that pre-donation screening of TTI with rapid tests may be safe in settings where there is no alternative, if all quality procedures are respected. In transfusion medicine, false negative screening results are of course the most harmful, and this was fortunately not observed here with the tests locally used. However, blood is a life-saving product extremely difficult to obtain. The harm of deferring unnecessarily a large number of potential donors and of communicating to them incorrect test results should not be underestimated. In collective donations where blood is often tested subsequently, destroying large amounts of blood bags is a huge waste of time, energy and money in already overstretched settings. Erroneous laboratory results may go undetected in the absence of quality control, and may be due to multiple causes such as intrinsic weaknesses of a test not produced under strict quality regulations, inadequate conservation and manipulation, or inappropriate local validation. Quality assurance and control are therefore critical issues; surprisingly however, it has been difficult during our research to find independent and transparent technical information on laboratory test performances, on appropriate conditions for purchase, transport and conservation, or on the processes of test validation, prequalification, approval or recommendation. There is an urgent need to make this information more accessible for appropriate decision making particularly in resource-constrained blood banks.