How do physicians decide to treat: an empirical evaluation of the threshold model

Table 4 Benefit versus harm ratio based on type 1 response*

Variable		n	Mean	Min	Median	Max
PE	B _aut /H _aut	40	4.33	.6	3.00	25.00
Base case	B ₁ /H ₁	40	6.28	0.75	3.18	49.50
Low risk	B ₁ /H ₁	39	12.46	0.66	5.26	100.00
High risk	B ₁ /H ₁	41	1.76	0.05	0.98	18.80
AML	B _aut /H _aut	41	2.29	0.43	2.00	10.00
Base case	B ₁ /H ₁	41	1.55	0.00	1.00	7.07
Low risk	B ₁ /H ₁	39	4.39	0.00	1.94	22.50
High risk	B ₁ /H ₁	40	0.70	0.00	0.50	3.00

Abbreviations: B _aut/H _aut assessment of benefit/harms ratio based on automatic, quick response, B ₁/H ₁-type 1 response driven by regret, PE pulmonary embolism, AML acute myeloid leukemia, low “risk” low threshold, high “risk” high threshold clinical decisions. [*Note that type 2 responses that relied on single values, fixed B₂/H₂ ratios precluding direct statistical comparisons with B_aut/H_aut. However, the values of B₂/H₂ differed considerably from B_aut/H_aut (from 1 to 10 in PE case, and 2 to 0.33 in AML case) consistent with a notion that the B_aut/H_aut estimates did not solely drive the decision-making (see Discussion)].

ISSN: 1472-6947