In areas with high level of malaria transmission, children under five years and pregnant women are the most vulnerable population and the main target of prevention strategies. The present study, carried out at four health care facilities in several areas of Gabon, estimated the evolution P. falciparum infection prevalence between 2005 and 2011. Large differences were observed between the sites, highlighting the heterogeneity of the epidemiological features of malaria in Gabon, even for sites that are geographically close together (e g, Libreville and Melen). Indeed, between, Melen and Libreville, the difference in the proportions of infected patients is lower probably due to the fact that health care facilities in both sites are accessible to all patients. Moreover, the difference between the patients consulting at both structures when considering the socioeconomic level is not so obvious, although, Melen is a sub-urban area of Libreville characterized by a majority of slums with a low income population, compared to the population of Libreville that is more complex in regards to the different living conditions.
As already described in urban cities, the prevalence was lower in coastal urban areas of Libreville and Port-Gentil. It was much higher at Libreville, probably due to a difference in urbanization level, equipment, housing and access to treatment between both cities. These differences do not seem to be related to ACT self medication and bed net coverage as reported elsewhere . In all health care structures, ACT is the main treatment prescribed and frequency of self medication varied among sites, it was not associated with the proportion of MPBS. Moreover, despite bed net coverage variation in 2011, children less than five years are the main bed net users . The confirmed low frequency of malaria infection in Port-Gentil compared to Libreville, although a high EIR and risk of transmission as reported by Mourou et al. could be related to the health care management of the majority of the population in Port-Gentil. Indeed, a great proportion of people benefit for free medical care provided at private health structure of oil companies where more than 70% of the active population work, each family having care free of charge; furthermore, presumptive treatment of fever with anti-malarial drugs is the rule in these centers.
Previous studies reported a decline in malaria prevalence in febrile children over time [4, 14, 18, 19]. In Mlomp, Senegal, an area of moderate malaria transmission, the risk of malaria decreased by about 32 times between 1996 and 2010, including the control strategy implementation period . A reduction of malaria case frequency was observed among inpatients hospitalized in Libreville between 2002 and 2008, accompanied by a high frequency of viral and bacterial infections . The decline of P. falciparum infection rates already observed in Libreville, is also confirmed in other areas of the country .
After the decline in the proportion of malaria cases at all sites between 2005 and 2008, a rebound in infection rates in rural and urban areas appears. In the same time, a slowdown in the prevention activities led by the MNCP and a low ITN coverage may partly explain this phenomenon. As an example, in Malawi, from 2001 to 2005, the proportion of malaria cases decreases of 50% (34.5% to 17.1%), followed by a rebound to 25%. An association with the slowing of interventions from 2001 to 2010 is characterized by a reduction of free ITN distribution, a limited free access to ACT in health care facilities for vulnerable populations, the lack of re-impregnation of used bed nets and indoor residual spraying (IRS) . In Kenya, a higher child mortality attributable to malaria was also associated to a drop in the stock of essential anti-malarial drugs and a disruption of services during civil unrest . In Zanzibar and Zambia where interventions are maintained, there is no rebound in malaria morbidity and mortality [22, 23]. Others factors such as vector resistance to insecticide could also contribute to increase malaria prevalence. In Port-Gentil and Libreville, the main vectors of Plasmodium carry a high proportion of molecular markers of resistance to the commonly used insecticides, with an impact on the effectiveness of the current vector control programs . In Zambia, both Anopheles gambiae s.s. and Anopheles funestus s.s. were controlled effectively with the ITN and IRS programme, maintaining a reduced disease transmission and burden . Environmental factors such as a slower urbanization of Melen and Oyem, as well as a probable high entomological inoculation rate (EIR) could all explain the slow reduction of malaria cases in these sites.
Consistent with other tropical areas, children under five years constitute the majority of patients consulting for fever . The proportion of infected children under five years globally decreased throughout the survey in each site, except at Port-Gentil where it increased probably due to the EIR, and underlying a difference of exposure of patients in both age groups. Between 2005 and 2011, the reduction of malaria cases among children under five years old was of 30% at Melen and almost 50% at Libreville. An obvious shift in the age of infected children towards those aged over five years and who are at higher risk of being infected, confirms previous data obtained from 28,000 children . Malaria risk increased from 0.37 to 5.05 over time among older children during the study period, highest in urban areas, Port-Gentil and Libreville, suggesting a delay in the immunity acquisition as observed in areas with lower malaria transmission [26, 27]. Another explanation could be the density of Anopheles population in both cities [17, 28]. In Thies, an urban area of Senegal, the most affected population was aged from four to 20 years old . In Tanzania, where malaria prevalence usually peaks in younger children (23–25 months), an increased risk of 1.7 of infection was found for children five to 13 years old compared to those of six months to four years . The younger children were also significantly more likely to sleep under ITNs compared to those aged from five to 13 years . Similar trends were found in both rural and urban areas in Uganda .
Together, the rates of malaria prevalence and the shift in the age of at risk children evoke a changing epidemiology of malaria in Gabonese cities, which cannot be all considered as malaria hyperendemic areas. Although the transmission is perennial, it rather becomes meso-endemic in some towns such as Libreville and it is much lower at Port-Gentil. Cohort studies and new entomological surveys from other areas are urgently needed to further characterize the malaria transmission in the country.
This study has some limitations. Patients could not be screened at all the sentinel sites. Otherwise, this study is a health centre survey. Therefore, it does not represent the situation in the whole population but it already provides reliable important data. Furthermore, assessment of bed net effective use and type (i e, insecticide treatment or re-impregnation), and their direct impact on malaria prevalence cannot be established.