Systematic review, meta-analysis and economic modelling of ...

Assessment of cost-effectiveness evidenceThe following strategies were tested in the main analysis:1. Cheapest and least effective Discharge all patients home immediately without testing or treatment.2. Most effective and expensive Measure troponin level after 10 hours has elapsed from the worstsymptoms, admit to hospital and treat if troponin assay is positive, discharge home without treatmentis troponin assay is negative.3. Troponin testing on arrival Measure troponin level on arrival, manage according to strategy 2 ifpositive (i.e. measure troponin level again after 10 hours from worst symptoms), discharge homewithout treatment if negative. This strategy was tested using different initial troponin assays andthresholds for positivity.In each strategy we assumed that there was a 2-hour delay from the time at which sampling could beperformed to the time at which results became available and a decision made. If the results were availablewithin 4 hours of patient presentation to hospital we assumed that the patient was still in the ED anda decision could be made immediately. If not, we assumed that they had moved to another location(a ward or clinical decision unit) and managed according to one of the three scenarios outlined below.We also assumed that there was a 1-hour delay between arrival at hospital and biomarker assessmentcommencing. This effectively meant that only decisions made on presentation biomarkers could be actedon in the ED.With regard to patient management after the ED, we tested the model in three different scenarios:1. The ‘doctor on demand’ scenario, in which medical staff were available 24 hours a day to make adisposition decision within 1 hour of the results being available.2. The twice-daily ward round scenario, in which medical staff were only available at twice-daily wardrounds (9 am and 6 pm) to make disposition decisions.3. The once-daily ward round scenario, in which medical staff were only available at one daily wardround (2 pm) to make disposition decisions.We took this approach because it was possible that different strategies may have different levels of costeffectivenessin different settings. For example, early discharge strategies may be less cost-effective if theLoS associated with delayed testing strategies is controlled by efficient patient review. Users of the resultsare thus able to decide which scenario best reflects their local practice.We also undertook a secondary analysis that involved adding other biomarkers to troponin at presentationto determine whether adding an alternative biomarker was cost-effective compared with troponin alone atpresentation or a 10-hour troponin test. This analysis was undertaken using data from primary studies thatcompared the sensitivity and specificity of troponin alone to troponin with the biomarker (with elevationof either biomarker being considered positive). We assumed that the additional biomarker would incuran additional cost, but otherwise the model would follow the main analysis. For each study the modelcompared the following strategies:1. discharge without testing or treatment2. presentation troponin alone3. presentation troponin in combination with the other biomarkers4. 10-hour troponin test.Diagnostic parameters of each strategyEach strategy specified how the biomarker(s) should be interpreted and what decision would be made onthe basis of each biomarker result. The options were:1. MI ruled out: discharge with no further testing84NIHR Journals Library