Paper | Good methodological quality of AE identification | Appropriate tradeoffs between level of detail and scale of the project in economic estimation | Good methodological quality of cost estimations | Detailed description of cost estimation methods | Presentation of cost–benefit ratio | Sensitivity analysis |
Eber et al (2010)32 | Yes, the usage of administrative data was a good option, given the high specificity and positive predictive values of identification of the AEs considered in the records | Yes, cost-to-tariff conversion was appropriate for a gross estimation of nationwide burden of disease | Yes, good criteria were used for matched comparison | Yes | Not applicable | One-way sensitivity analyses were done |
Dietrich et al (2002)31 | Yes, in the prospective series, the usage of validated clinical criteria allows for reliable identification of AEs | Yes, the relatively small sample allowed for detailed micro-costing and analysis | Yes, good example of differences in prospective and retrospective analyses | Yes | Not applicable | Lacking |
Sheng et al (2005)41 | Yes, AEs were identified following validated CDC criteria applied to different infection sites | Yes, micro-costing collection from financial services was appropriate considering the sample size | Yes, good criteria were used for matched comparison | Yes, in particular, the matched comparison analysis is fully described | Not applicable | Lacking |
Fukuda et al (2008)26 | Not applicable | Yes, the results of a pilot study (Fukuda et al25) allowed identification of the costs to be included | Yes, a standardised questionnaire was used for collection throughout HCOs | Yes | Not applicable | Lacking |
Stone et al (2007)45 | Not applicable | No, data collection unduly complete considering the sample size | Yes, cost valuations were standardised for 100 beds, which allowed for comparisons | Yes | Not applicable | Lacking |
Brown and Lilford (2009)48 | No, only published estimates are used in the model | Yes | Yes, although relying on hypotheses and assumptions that are difficult to prove | Yes | Yes, the net cost is presented | Yes |
Calugar et al (2006)46 | Yes, based on clinical and biological diagnosis for cases and usual contact identification and screening techniques | Yes, detailed information on a single outbreak was collected and combined to literature data, which allows for avoiding resource intensive estimations | Yes, costs are estimated conservatively, but results are compared with other similar studies | Yes, clear and fully described | Yes, the cost–effectiveness ratio is presented | One-way sensitivity analyses were done |
Karnon et al (2008)55 | Not applicable (identification technique referenced to another report) | Yes, multiple sources of published data were used and empirical data | Yes, interesting example of usage of decision analytic models for estimating potential benefits of a specific safety practice | Yes, the hypotheses underlying the model are clearly exposed | Yes, the net benefit is presented | Lacking |
Van Rijen and Kluytmans (2009)49 | Yes, based on laboratory data and following systematic screening of ‘suspected’ or ‘at risk’ cases | Yes, very detailed collection was appropriate for the HCO setting | Yes, although it would have been useful to discount values because of the large collection span | Yes | Yes, the cost–effectiveness ratio is presented | Lacking |
Wang et al (2003)28 | No, multiple sources are used and the criteria for defining AEs were not clearly stated | Yes | Yes, very good example of stochastic modelling, with both empirical and literature data | Yes | Yes, the net benefit is presented | Yes, both one-way and five-way sensitive analyses were performed |
AE, adverse event; CDC, Centers for Disease Control; HCO, healthcare organisation.