The main purpose of public health surveillance is to provide timely, scientifically sound evidence to stakeholders and decision makers, in order to make decisions for improving public health in their jurisdiction . Completeness of reporting is an important attribute to achieve this objective. The surveillance system should be appropriately inclusive and the reported cases should also represent the complete list of eligible persons. In practice, it is a very challenging task to capture all cases of a disease in the population; thus, evaluation of completeness is required on a regular basis . Quantification of underreporting allows the establishment of a baseline for data quality and the identification of areas for improvement [5, 23]. It, also, allows comparisons between countries, since diversity of the health care and surveillance systems handicaps data comparability.
The estimated reporting rates of salmonellosis and shigellosis in this study were similar to previous estimations in the country at a local level  and to the estimations of reporting rates in Spain . Other European countries have much higher completeness that has been reported to reach almost 100% when an electronic reporting system is in place; 95.8% in Sweden , 99.0% in Germany  and 82.0% in Italy .
The estimations of the C-RC study were similar to the results of the hospital registry review. Authors believe that the three of the four assumptions of C-RC studies were not seriously violated; the population was a closed, well-defined “cohort”, cases of the surveillance systems were identified and matched, and all cases had the same probability to be reported to each of the systems (equal catchability) [16, 17]. On the other hand, there might have been positive dependence between the data leading to an underestimation of underreporting. In better staffed hospitals, for example, cases that were reported to MNS might have been more likely to also be found in the NRLSS dataset [16, 17]. However, positive dependence in this case was quite small as indicated by the comparison with the results of the registry review.
The results of this study and of similar studies can be used for better understanding the causes of underreporting of the diseases to MNS; for example, reporting from larger university hospitals may be impaired by the increased work load of personnel or by the lack of supervision and training on reporting processes. Geographical and seasonal differences, also, need further study. Due to the small number of reported cases, these parameters could not be assessed for shigellosis. Low internal completeness of the variable “serotype” for Salmonella and the variable “species” for Shigella of the MNS dataset did not allow the estimation of underreporting by serotype/species. The factors that mainly affect the completeness of surveillance systems and should be further studied are related to the health care system (lack of personnel, of technical support, etc.), the data providers (lack of interest or training, etc.) or the surveillance system (long or complicated reporting forms, lack of electronic reporting systems, etc.) [26, 27]. Results of this study were disseminated to the hospitals accompanied with a request to all clinical doctors to systematically notify salmonellosis and shigellosis cases and to report possible problems of the reporting process.
Estimation of the diseases’ community incidence using the hospitalisation rate of the diseases reported by other European countries may not have been totally accurate; however this approach was decided since a nation-wide study on under-ascertainment of the two diseases is not available. Published estimations of the foodborne diseases’ community incidence in Greece are based on travelers [28, 29] or data from other countries [30–32] and are subject to several biases. In addition, the available estimations refer to past data and may be outdated and no longer applicable . Thus, we believe that the extrapolation of the hospitalisation rate based on recent data from other European countries with quite similar morbidity patterns with Greece -though not the optimum method- leads to a more accurate estimation than the already available.
Another limitation was that private hospitals were not included in the estimation. Based on the literature, admissions to private hospitals in Greece account for 16% of total admissions, but a specific estimation for salmonellosis and shigellosis cases is not available . The range of the estimated community incidence of salmonellosis, though wide, was compatible to estimations from other European countries , higher than in Canada , Australia , and Netherlands  and lower than Poland .