Table 4.

—[Section 2.3] Summary of Findings: Should Anticoagulant Prophylaxis (LMWH, UFH, Fondaparinux) vs Placebo/No Treatment Be Used in Hospitalized Medical Patients?^24-26

Outcomes	Anticipated Absolute Effects		Relative Effect (95% CI)	No. of Participants (Studies) Follow-up	Quality of the Evidence (GRADE)
	Baseline Riska	Risk Difference With Anticoagulant Prophylaxis (95% CI)
Symptomatic DVT	Low risk		RR, 0.47 (0.22-1)	5,206 (4 RCTs) 1-14 d	Moderate due to imprecisionb
	2 per 1,000	0 fewer per 1,000 (from 1 fewer to 1 more)
	High risk
	67 per 1,000	34 fewer per 1,000 (from 51 fewer to 0 more)
Nonfatal pulmonary embolism	Low risk		RR, 0.61 (0.23-1.67)	5,206 (6 RCTs) 1-22 d	Moderate due to imprecisionb
	2 per 1,000	1 fewer per 1,000 (from 1 fewer to 1 more)
	High risk
	39 per 1,000	15 fewer per 1,000 (from 30 fewer to 26 more)
Major bleeding	4 per 1,000	1 more per 1,000 (from 1 fewer to 6 more)	OR, 1.32 (0.73-2.37)	8,605 (8 RCTs) 10-110 d	Moderate due to imprecisionb
Overall mortality	45 per 1,000	1 fewer per 1,000 (from 9 fewer to 8 more)	OR, 0.97 (0.79-1.19)	7,355 (5 RCTs) 1-22 d	Moderate due to imprecisionb
Thrombocytopenia	13 per 1,000	1 fewer per 1,000 (from 6 fewer to 7 more)	OR, 0.91 (0.54-1.53)	4,624 (3 RCTs) 6-21 d	Low due to risk of bias and imprecisionb

GRADE = Grades of Recommendations, Assessment, Development, and Evaluation; RR = risk ratio; UFH = unfractionated heparin. See Table 1 legend for expansion of other abbreviations.

^aBaseline risk for DVT and PE in low-risk population were derived from the RAM by Barbar et al.⁹ Baseline risk for mortality and bleeding is derived from the control arm of medical patients in a meta-analysis (Dentali et al).²⁴

^bWe will consider the presence of serious imprecision when there are <300 events in total (events in treatment and control patients) or when CIs include appreciable harms and benefits.