Table 1.
Probabilities of annual state transition as well as underlying assumptions and sources.
| Serial number | Model item | Assumptions | Sources and description |
| 1 | Prevalence of AFa at baseline | Based on the CHA2DS2-VAScb score: 0.01, 0.015, 0.034, 0.067, 0.118, 0.182, 0.255, 0.302, 0.403, 0.492 | Derived from the study of Saliba et al [7]. The prevalence was used to simulate the initial proportion of patients with AF. |
| 2 | Incidence of AF in the general population (per 100 person-years) | Based on the CHA2DS2-VASc score: 0.17, 0.21, 0.49, 0.94, 1.65, 2.31, 2.75, 3.39, 4.09, 6.71 | Derived from the study of Saliba et al [7]. The incidence was used to estimate the number of new cases of AF each year. |
| 3 | Sensitivity of mHealthc devices | 93%d | Derived from the study of Bonomi et al [8]. Sensitivity of PPG compared to 24/48-hour Holter electrocardiogram readings in outpatient settings; 93 out of 100 patients with AF receive a true-positive diagnosis. |
| 4 | False-positive AF detection rate (mHealth device) | 0.2%d | Bonomi et al [8] described the false-positive detection rate as lower than 0.2%; 0.2% of subjects without AF receive a false-positive diagnosis. |
| 5 | Confirmation of the mHealth diagnosis (by a physician using ECGe) | 100%, 75%, and 50% | Because of the nonpersistent forms of AF, the disease cannot always be confirmed through ECG follow-up. Nevertheless, in the first step, the assumption was made that a true-positive mHealth diagnosis of AF can always be confirmed by a physician. In subsequent simulations, the proportion was altered. |
| 6 | Clarification of a wrong mHealth diagnosis (by a physician using ECG) | 100% | Assumption that in patients with no AF, the attending physician will not find artefacts of arrhythmia in the electrocardiogram. |
| 7 | Proportion of AF detected without a device | 36.09% | Steinhubl et al [9] investigated the detection rate of AF in active home-based monitored individuals. They found newly diagnosed AF in 6.7 per 100 person-years in the monitored individuals and 2.6 per 100 person-years in unmonitored individuals. The proportion of AF detected with the aid of wearables was multiplied with the AF ratio between unmonitored and monitored individuals. Yearly, 36.09% of AF cases can be detected without the use of mHealth devices. |
| 8 | Stroke incidence in untreated patients with AF (per 100 person-years) | Based on the CHA2DS2-VASc score: 0.2, 0.6, 2.5, 3.7, 5.5, 8.4, 11.4, 13.1, 12.6, 14.44 | Derived from the study of Friberg et al [10]. The stroke incidence yields the probability of experiencing a stroke. |
| 9 | Stroke incidence in patients with no AF (per 100 person-years) | Based on the CHA2DS2-VASc score: 0.0826, 0.2479, 1.0331, 1.5289, 2.2727, 3.4711, 4.7107, 5.4132, 5.2066, 5.9669 | According to Odutayo et al [2], patients with AF have a 2.42-fold increased risk for stroke compared to patients with no AF. The stroke incidence in untreated patients with AF was divided by 2.42. |
| 10 | Stroke incidence in patients with AF receiving NOACf (per 100 person-years) | Based on the CHA2DS2-VASc score: 0.068, 0.204, 0.85, 1.258, 1.87, 2.856, 3.876, 4.454, 4.284, 4.9096 | VKAg reduces the risk of stroke by two-third (66%) [1]. Rivaroxaban is noninferior to warfarin [11]. Thus, the risk reduction through NOAC should be at least as high as the one from VKA. |
| 11 | Stroke mortality in patients with no AF | 34% | Derived from the study of Reinhold et al [6]. If a patient does not have AF but experiences a stroke, there is a 34% probability that the stroke is fatal. |
| 12 | Stroke mortality in untreated patients with AF | 63% | Derived from the study of Reinhold et al [6]. If a patient has AF and does not receive medication, there is a 63% probability that the stroke is fatal. |
| 13 | Stroke mortality in patients with AF receiving NOAC | 42% | Derived from the study of Reinhold et al [6]. If a patient has AF and receives medication, the probability that an occurring stroke is fatal is 42%. |
| 14 | Mortality in patients with no AF, no stroke | 6% | Derived from the study of Reinhold et al [6]. Probability that an individual who does not have AF dies due to reasons other than stroke. |
| 15 | Mortality in untreated patients with AF, no stroke | 11.1% | Derived from the study of Reinhold et al [6]. The probability that an untreated patient with AF dies due to reasons other than stroke. |
aAF: atrial fibrillation.
bCHA2DS2-VASc: Congestive heart failure, Hypertension, Age≥75 years, Diabetes mellitus, Stroke, Vascular disease, Age 65-74 years, Sex category (female).
cmHealth: mobile health.
dvalues were changed in sensitivity analysis.
eECG: electrocardiography.
fNOAC: non–vitamin K antagonist.
gVKA: vitamin K antagonist.