Table 2.

GRADE evidence table

Quality assessmentSummary of findings
No. of participants (studies)Risk biasInconsistencyIndirectnessImprecisionPublication biasOverall quality of evidenceStudy event rates (%)RE (95% CI)Anticipated absolute effects
Temporary cessation of VKA aloneTemporary cessation of VKA + vitamin K administrationRisk with temporary cessation of VKA aloneRisk difference with temporary cessation of VKA + vitamin K administration (range)
Mortality (follow-up: 30-90 d; assessed with all-cause mortality)
 860 (3 RCTs)NSNSNSS*None⊕⊕⊕○ Moderate13/439 (3.0)16/421 (3.8)RR = 1.24 (0.62-2.47)30 per 10007 more per 1000 (11 fewer-44 more)
Major bleeding (follow-up: mean, 90 d; assessed with fatal bleeding or bleeding requiring blood transfusion or admission)
 801 (2 RCTs)NSNSNSS*None⊕⊕⊕○ Moderate4/409 (1.0)10/392 (2.6)RR = 2.43 (0.81-7.27)10 per 100014 more per 1000 (2 fewer-61 more)
Any thromboembolism (follow-up: mean, 90 d; assessed with venous or arterial thromboembolism)
 801 (2 RCTs)NSNSNSS*None⊕⊕⊕○ Moderate4/409 (1.0)5/392 (1.3)RR = 1.29 (0.35-4.78)10 per 10003 more per 1000 (6 fewer-37 more)
Proportion reaching goal INR (follow-up: mean, 1 d; assessed with INR goal ranges: INR, 1.8-3.2; INR, 2.3-4.5; and INR, 2.0-4.0)
 1025 (5 RCTs)SSNSS*None⊕○○○ Very low90/518 (17.4)218/507 (43.0)RR = 1.95 (0.88-4.33)174 per 1000165 more per 1000 (21 fewer-579 more)
  • Overall quality of evidence measured according to GRADE criteria.

  • RE, relative effect; NS, not serious; S, serious.

  • * Lower and upper bounds of 95% CI may lead to different recommendations.

  • Four of the 5 studies did not blind patients and personnel, or outcome assessors.

  • I2 = 93%.