In this patient-level model, each hypothetical patient was assigned an age, gender, and disease state (i.e. ACS or no ACS) based on the population distribution derived from a cross-sectional diagnostic study [12, 13]. For ACS patients in primary care, the probabilities of having either UAP, NSTEMI or STEMI were based on a prospective study [14].

Of all chest pain patients presenting to the GP, 14.7% was immediately referred to the hospital without diagnostic testing, because of a high perceived probability of ACS by the GP (Table 1) [15]. All patients that were referred to the hospital were assumed to be transported by ambulance. Among the patients that were not immediately referred, all GPs who had an ECG in their practice (45%) [16], were assumed to use it in those patients. In case the patient had a STEMI, there was a 50% probability that this was diagnosed in case an ECG was performed [17]. Those patients were also immediately referred to the hospital. In current practice, the GP decides on referral of the remaining patients without POC troponin. The sensitivity and specificity of this decision (i.e. 88.3% and 72.2%, respectively) were derived from Nilsson et al. [18]. For the POC strategy, the time from symptom onset to presentation was estimated for each patient separately [19], and then categorized according to time since symptom onset as a) <4.0 h, b) 4.0–5.5 h, c) 5.5–7.0 h, d) 7.0–10.0 h, and e) >10.0 h. As it is known that the performance of the POC troponin test is less in patients presenting shortly after symptom onset [7], it was assumed that this test was not performed in patients presenting with <4.0 h symptoms [7, 19]. In those patients, the referral decision was based on the judgment of the GP as in current practice [18]. In the remaining patients, it was assumed that the POC troponin test was performed, and the sensitivity and specificity for time categories b), c), d) and e) were derived from Diercks et al. [7]. In case the result of the POC troponin test did not correspond with what would have been the (initial) referral decision of the GP in current practice (i.e. without POC troponin), a probability of the GP revising the initial referral decision was incorporated in the model [15]. In both strategies, all patients who were not immediately referred to the hospital were assumed to be sent home without receiving medical care.

Mortality and heart failure rates were dependent on whether patients had ACS, and if present, whether ACS was correctly diagnosed. In addition, as the POC troponin test was assumed to delay time to treatment with 10 min, using this test would increase the mortality rate in AMI patients with 0.2% [20]. This 10 min duration of the POC troponin test was chosen as previous research has shown that GPs consider this the maximum acceptable duration of the POC troponin test [15]. Productivity losses incorporated lost working time due to GP visits, emergency department visits, and due to UAP, STEMI, NSTEMI, heart failure, and mortality [21‐23].

The quality-adjusted life expectancy (QALE, based on combining expected quality of life with life expectancy) were obtained directly from a systematic review and economic model for STEMI, NSTEMI and UAP patients as well as for healthy patients [24‐26]. Separate QALE estimates were used for categories of age and gender [24‐26]. A missed diagnosis of ACS indirectly lowered quality of life by increasing the risk of heart failure and mortality. For heart failure patients, the QALEs were estimated by multiplying the reported utility of heart failure patients with their reported average life expectancy [27, 28].

Costs of both a GP consultation and ambulance transportation to the emergency department were derived from the Dutch cost manual [23], whereas costs of performing an ECG were derived from the Dutch Healthcare Authority [29]. As no tariff currently exists for the POC troponin test, and because a troponin test performed in a regular laboratory costs approximately €8 [29], the costs of a POC troponin test were conservatively estimated at €15. Average diagnosis and treatment costs per (suspected) ACS patient were derived from the Dutch Healthcare Authority [3]. Healthcare costs for the remaining lifetime of STEMI, NSTEMI, and UAP patients were derived from Goodacre et al. [24‐26]. Lifetime healthcare costs per patient with heart failure were based on Dutch data concerning the point prevalence of heart failure [30], the annual healthcare costs of those patients [31], and combined with the average remaining life expectancy [27]. All costs were converted to 2015 using the consumer price indices as reported by the Dutch Central Bureau of Statistics, and converted to Euros if necessary (exchange rate 1.271 GBP = €1.000, 31 March 2016) [32, 33]. All costs and effects were discounted at an annual rate of 4.0% and 1.5%, respectively, in accordance with Dutch costing guidelines [23]. An overview of all cost parameters, including costs of productivity losses, is shown in Additional file 1.