Intense Focus of Alveolar Echinococcosis, South Kyrgyzstan

Human alveolar echinococcosis (AE) is a highly pathogenic zoonotic parasitic disease caused by Echinococcus multilocularis. An ultrasound study in southern Kyrgyzstan during 2012 revealed a prevalence of 4.2% probable or confirmed AE and an additional 2.2% possible AE, representing an emerging situation. The risk for probable or confirmed AE was significantly higher in dog owners.

interpretation in terms of significant variable as relative risk of infection associated with that variable (as opposed to the logit, which gives the odds ratio). All computations were undertaken in R (https://www.r-project.org). A backward selection method was used with all variables initially included; with each iteration, the least significant variable was removed until only variables with p<0.15 remained in the model. Variables with p<0.05 in the final model were reported as significantly associated with the presence of an AE lesion diagnosed by ultrasound.
The adequacy of the model was addressed by a sensitivity analysis using R. Briefly the data were repeatedly and randomly split in subsets to address the predictive value of the model. Thus, parameters were derived from part of the data and used to predict the accuracy of the model on the remainder of the data. From these, an estimate of the sensitivity and specificity of the model can be made, as well as the overall accuracy (area under the curve in a Receiver Operating Characteristic Curve plot). The Area Under the Curve was 0.64 for the binomial model, which indicates the model had a modest predictive power, despite the significant parameters (Technical Appendix Figure 2). In addition the pseudoR 2 was 0.046 indicating that the final model parameters: dog ownership, sex, age and slaughter of livestock could only predict 4.6% of the variation in infection with AE. Variable importance analysis indicated that dog ownership (3.02) had the greatest influence, followed by age of participant (2.68), sex (2.30), and home slaughter (2.02).

Patient Details
Of the 1,617 study participants examined by ultrasound, suspected AE was diagnosed in 106; in 1, cystic echinococcosis (CE) was concomitantly diagnosed. CE was diagnosed in 3 additional participants. Of these, 53 patients with AE had the lesions measured. The mean age of those with measured lesions was 28 years, and 49% were male; mean of those with unmeasured lesions was 20 years and 61% were male. Of the 37 patients who had follow up and lesions confirmed as AE by histopathology, 13 had been measured at the initial ultrasound scan. These had a mean size of 40.1 mm (range 7-133 mm). Of the remaining 70 who were lost to follow up, 40 had their lesions measured during the original ultrasound scan and had a mean lesion size of 25.5 mm (range 5-197 mm). There was some statistical evidence that the group with follow up had larger lesions (Wilcoxen test, p = 0.02).
Page 3 of 6 CE was diagnosed in only 3 participants (a prevalence of just 0.2%), which is too small a sample size to undertake any analysis for this disease. It is also a somewhat lower prevalence than found in similar studies elsewhere in Kyrgyzstan (2) or in a neighboring region of Kazakhstan (3). The reasons for this are not known, but ongoing research mapping all cases of echinococcosis and analyzing the geographic distribution may give clues to this.

Accuracy of the Diagnostic Procedures
Diagnostic accuracy of the serology and ultrasound examination can be assessed only against proven AE infections. We have examined 3 possible scenarios, all with the assumption that ultrasound has a diagnostic sensitivity of 100% (Technical Appendix Table 3 The prevalence of infection, assuming this lower specificity, would be 37/1,617 = 2.3%. These all assume that the sensitivity of ultrasound is near 100%. This is reasonable because a case definition is a visible lesion consistent with AE on ultrasound. It is possible that there may have been cases of AE without a primary lesion in the liver, which would have escaped detection by ultrasound, but such cases are rare. Just 9 of 387 AE cases in France had primary extrahepatic lesions (4).
3. Alternatively 49 study participants (of 95 with serology) were serologically positive, and so probably had AE. In addition, there were 18 confirmed cases of AE (seronegative or no serum sample, but confirmed by histology), which would suggest 67 cases as confirmed or probable AE. This gives a prevalence of 4.1% and a specificity of the ultrasound as 97.4% (39 false positives in 1,550 not infected) The sensitivity of the serology varies little between scenario 1 and 2, and we cannot estimate it in scenario 3. The positive predictive values (i.e., the probability of having AE given a positive diagnosis with ultrasound) varies from 34.9% to 94.8% depending on which scenario is adopted. The prevalence also varies, but even with the most conservative estimate, assuming all cases lost to follow up were AE negative, it still gives a high prevalence of 2.3%.