The seroprevalence of T. gondii
infection of 5.1% in the rural area of the Kyrgyz Republic was low compared to international numbers and to data from surrounding countries, whilst the seroprevalence in the Bishkek population (16.4%) is more in line with international data 
. Several factors might contribute to the low seroprevalence in the rural study-population. Although consumption of undercooked meat and consumption of unwashed fruit or vegetables were associated with a higher infection risk in different studies 
, these are consumed only infrequently in rural Kyrgyzstan. Shashlyk (explained in the “Methods”-section) is the only meat dish eaten occasionally without thorough cooking, but this was not found to be a significant risk-factor. In the urban population of Bishkek, the type of food consumed is more likely to be a source of infection, especially as it frequently includes shashlyk and other undercooked meats, as well as salads, raw vegetables, and fruit which could be contaminated by oocysts. The risk for infection by oocysts is highlighted by several reported outbreaks of toxoplasmosis caused by contamination of surface water or public water supplies in various parts of the world 
. Contaminated public water supplies may be a potential source of infection. However, in Bishkek the water supply is obtained mainly from underground sources and is of good quality with little treatment needed to make it potable. Likewise, in the rural study population, most of the water supplies (78%) came from pipes outside the house which are mainly fed by ground-water, where a low probability of contamination with oocysts of T. gondii
can be assumed. To a minor degree these pipes are fed by reservoirs containing only spring water, where contamination with oocysts cannot be ruled out. Only a minority of the population (15%) takes drinking water from rivers. This was found to be a significant risk factor for toxoplasmosis in a similar region in Kazakhstan 
As there are no significant deviations in the linear increase of the odds ratio of seroprevalence with age, there is no evidence from the present data that the socioeconomic changes which began some 20 years ago have had an effect on the incidence of T. gondii infection.
Age, low social status and a low number of sheep owned were the only significant risk-factors in the rural population. Indeed, social status in this area is linked to a low number of sheep owned. On the other hand, people with high social status, income or education (e.g. medical personnel, teachers) in our study group own no or only few sheep which suggests that social status as well as low number of sheep are independent risk factors. In this respect, the results of our study are similar to those of a study performed in Kazakhstan, where a low number of cattle per family was also found to be a significant risk factor 
. Different eating habits, especially consumption of vegetables, or differences in personal and food hygiene (e.g. hand-washing, washing of leaf vegetables) may be responsible for the higher prevalence in people with lower social status. A low education level, commonly linked to lower social status, has been found being a risk-factor for T. gondii
. Home-slaughtering of sheep including direct contact with raw meat 
or consumption of unpasteurized milk 
were not significant risk factors. However, home-slaughtering is practised in nearly all households (91%), thus this factor was not discriminatory between affected and non-affected individuals. Other previously described, possible risk-factors are contact with soil 
, and poor personal and/or food hygiene 
. Since T. gondii
oocysts are likely spread all over the settlements as cats roam freely, direct faecal-oral infection without food or water as an intermediate may be an important infection route. However, this is hard to document and evaluate, as almost all of the screened people work in agriculture and/or have regular contact with soil. The seroprevalence increased with age of participants consistent with other studies performed in various parts of the world (e.g. Europe, USA, Brazil, Kazakhstan 
The present results of our estimate on congenital toxoplasmosis indicate the possible extent of the problem. There are of course many assumptions in these calculations, the most important are that infection pressure has not changed over time, it is constant throughout the urban and rural regions of Kyrgyzstan and the sampled population is representative. The logistic regression approach to estimate the seroprevalence at each age allows for a higher infection pressure in children (indirectly through having an intercept) but otherwise would assume a more constant infection pressure. There is clearly uncertainty about this. Nevertheless, the seroprevalence in the sampled rural population is low compared with virtually all other countries 
and is lower than found in a similar study in rural Kazakhstan 
. Therefore, it can be argued that the incidence of congenital toxoplasmosis is unlikely to be substantially lower, but could indeed be higher than the calculated incidence. There is an important corollary to this. If the infection pressure in the population is higher, a greater proportion of girls will seroconvert before they reach reproductive age and hence a smaller proportion of women will be at a higher risk of seroconverting during pregnancy. Thus, there will be non-linear effects of increasing population exposure to T. gondii
in terms of numbers of congenitally infected children. A mathematical model has suggested that an infection pressure representing approximately 4% seroconversions per annum amongst sero-negative individuals would result in the highest risks of congenital toxoplasmosis, with approximately 67% of women being seropositive at age 27. This would result in approximately 1% of pregnancies being affected by toxoplasmosis 
. Higher infection pressure resulting in exposure of girls before reproductive age and lower infection pressures with a lower risk of exposure both result in fewer congenitally infected infants. The necessity of secondary prophylaxis by serological screening for and treatment of prenatal infection has currently been challenged in Europe within the last years as for example Denmark has effectively stopped its national prenatal screening program 
. In contrast, primary prophylaxis by means of hygiene and adaptation of eating habits are still believed to be of importance in preventing the disease, although hard scientific data is lacking 
According to our estimate approximately 350–970 HIV-patients are currently co-infected with T. gondii
in Kyrgyzstan, and between 125–360 AIDS patients are at risk for becoming infected within the next 10 years 
. Thus, AIDS-related toxoplasmosis may cause considerable mortality, estimated to range between 0–17% (mean: 13%) in resource-poor settings even under treatment 
. In addition there is disease morbidity, AIDS-related orphans and economic costs. Due to major differences in HIV-prevalence in official reports and in UNAIDS estimates we calculated two separate estimates with both these numbers. For this we assumed that the T. gondii
infection seroprevalence levels were representative for the rural population, and the population of the two major Kyrgyz cities of Bishkek and Osh.
We estimate 66 clinically relevant cases of congenital infection and 125–360 cases of AIDS-related toxoplasmosis annually in the Kyrgyz Republic. Therefore, measures for primary prevention of T. gondii infection during pregnancy, education of the medical personnel and provision of urgently needed adequate material for diagnosis and treatment are essential for tackling toxoplasmosis in the future.
Limitations of the study
Although we believe that the present study includes a representative sample of a mainly rural community of pastoralists and the urban population in Bishkek, there are important agrarian populations of Kyrgyz, Uzbek and Tajik ethnicity in the Fergana valley area of Southern Kyrgyzstan, which may have a different seroprevalence. Thus, the seroprevalence found in our study may not be representative for the entire country. Thus, the case numbers of clinically relevant toxoplasmosis have to be seen as an estimate representing mainly pastoralist communities and larger cities. Nevertheless, the magnitude of future putative clinical cases of toxoplasmosis can be derived from these estimates.
The sampling strategy was also not random as there were some departures from the normal population profile in Kyrgyzstan. In particular there is an over representation of women in our sample. This might produce some bias with regards to the factors associated with toxoplasmosis seropositivity in the rural population. However, the sampling strategy should neither affect the estimates of the incidence of congenital toxoplasmosis nor the incidence of complications of HIV infection as these were calculated from the age specific prevalences. This study design in which people were invited to participate clearly resulted in a disproportionate number of women entering the study and this may be because many men were unavailable due to work. In the rural population there were similar numbers of boys and girls in the study under 10 years of age which would be consistent with this hypothesis. The urban samples were supplied by diagnostic laboratories and it is not clear why these samples were over represented by women.
An additional limitation is the assumption that infection pressure or exposure has not changed with time. For example older people may have had greater exposure when they were younger and this could inflate the increase in age as a risk and overestimate the numbers of infants born with congenital infections. However changes in infection pressure would be expected to give non-linear increases in the age stratified sero prevalence. We tested this hypothesis by using generalised additive models which should have detected any significant deviations in the linear increase in the log of the odds ratio. Whilst this does not prove that infection pressure has not changed over time it does provide additional evidence that our assumptions are valid. On the contrary, the migration of individuals from rural areas with a low infection pressure to the cities with a higher infection pressure might mean that that this cross sectional study has underestimated the infection pressure and hence under estimated the number of cases of congenital toxoplasmosis.
The diagnostic test we developed and used in this study may have resulted in a small underestimate of the prevalence of toxoplasmosis. The Platelia Toxo IgG assay against which we evaluated our test system has a reported sensitivity and specificity approaching 100% 
. Our assay was positive in 49 of 50 samples positive with the commercial test. This would indicate a sensitivity of 98% (CIs 89.35–99.95%) assuming that the Platelia Toxo IgG assay is indeed such a perfect gold standard. Therefore it is possible we have underestimated the prevalence of toxoplasmosis in these populations by approximately 2%.