Assessing the reliability of self-reported weight for the management of heart failure: application of fraud detection methods to a randomised trial of telemonitoring

Table 5 Simulation analysis – effect of alternative assumptions for number of weight readings submitted

	Assignment of EDP patients (%)		Assignment of NEDP patients (%)		Proportion of people in A1-A4 with NEDP (%)
	A0	A1-A4	A0	A1-A4	Proportion of people in A1-A4 with NEDP (%)
10	42.6	1.9	3.5	12.9	97.5
20	96.0	4.7	8.4	34.1	97.6
40	99.8	0.6	4.4	17.4	99.4
60	100.0	0.3	4.5	16.9	99.7
80	100.0	0.3	6.7	25.9	99.8
162 (from Table 1)	100.0	0.0	5.9	23.5	100.0
180	100.0	0.0	6.0	22.6	100.0

Note: Assumes 15% of patients have EDP, 30% of the time, with weights that are uniformly distributed on {0, 1, 2, 3, and 4} modulo 5. Based on 10,000 replications of the simulation experiment

ISSN: 1472-6947