Concordance of phenprocoumon dosage in married couples

BMJ 1997; 314 doi: (Published 10 May 1997) Cite this as: BMJ 1997;314:1386
  1. Timon W van Haeften, internista,
  2. Johanna de Vries, internist directora,
  3. Jan J Sixma, professorb
  1. a Thrombosis Service Utrecht Netherlands
  2. b Department of Haematology Utrecht University Hospital Utrecht Netherlands
  1. Correspondence to: Dr T W van Haeften Department of Internal Medicine G 02.228 Utrecht University Hospital PO Box 85500 NL 3508 GA Utrecht Netherlands.
  • Accepted 24 January 1997


Coumarin derivatives are used in large numbers of patients. However, patients vary considerably in the dosage required to achieve a given level of anticoagulation.1 Whether this variation is due to differences in diet (for example, intake of vitamin K) or other factors (for example, liver metabolism) is unclear. To determine the influence of life circumstances such as diet on anticoagulant dosage we studied couples in whom both spouses used phenprocoumon. In addition, we compared the dosages used in these patients with those in other subjects, matched for age and sex.

Patients, methods, and results

Our thrombosis service treats around 10 000 patients a year with coumarin derivatives, prothrombin times being monitored by a computerised system. Three target levels of anticoagulation are aimed at depending on the diagnosis. We identified all couples (n=33; 66 patients) in whom both spouses used phenprocoumon for anticoagulation. Median ages were 72 years (range 61-87) for men and 70 years (62-84) for women. Twenty five couples shared the same target levels of anticoagulation. We then sought 66 controls matched for age, sex, and anticoagulation target level. Thirty three “matched couples” were formed with mean ages of 72 years (range 62-84) for “matched husbands” and 69 years (56-87) for “matched wives.”

Individual mean dosages and international normalised ratios2 between October 1993 and May 1994 were analysed. Only one batch of samples had been used for prothrombin time determinations. Data are given as medians and ranges. Paired Wilcoxon tests were used. Differences in phenprocoumon dosage between spouses and between “matched spouses” were analysed with calculation of 95% confidence intervals for the difference between differences in dosage.3

Median anticoagulation levels in the 33 couples did not differ significantly between spouses (international normalised ratio 3.3 (range 2.3-4.0) in men v 3.2 (1.8-4.7) in women; P=0.77) or between the matched spouses (3.4 (2.5-4.1) v 3.4 (2.2-4.1); P=0.74). Median phenprocoumon dosage did not differ between husbands and wives (1.98 (range 0.87-3.29) v 2.03 (0.71-3.10) mg/day respectively; P=0.88), and the dosage in matched husbands (2.27 (0.98-4.65) mg/day) was not significantly higher than the dosage in matched wives (1.98 (0.99-3.19) mg/day; P=0.094). Similarly, couples with identical target levels of anticoagulation showed no differences in dosage. There was a significant linear correlation of phenprocoumon dosage between spouses (r=0.57 (n=33); P=0.0008) (fig 1), which was not found in matched couples. After excluding the eight couples with different target levels of anticoagulation the correlation was not altered (r=0.59 (n=25); P=0.0022) and again not significant in the matched couples.

Fig 1
Fig 1

Linear correlation of phenprocoumon dosage in 33 husbands and their wives

The dosage difference within couples was smaller than the dosage difference within matched couples (mean difference of dosage differences -0.49 (95% confidence interval -0.78 to -0.20) mg/day). Similarly, the dosage differences within the 25 couples with identical target levels of anticoagulation were smaller than within the 25 matches (mean difference -0.44 (-0.77 to -0.07) mg/day). In none of the four groups was phenprocoumon dosage significantly correlated with age (all P>0.30).


Coumarin derivatives act by inhibiting vitamin K dependent synthesis of several coagulation factors.4 We found a significant relation in phenprocoumon dosage between spouses, which we assume was due to diet. A diet rich in vitamin K (broccoli, spinach, Brussels sprouts, or lettuce) leads to increased coagulation activity whereas a diet poor in vitamin K leads to stable anticoagulation.5

Our findings suggest that diet (and dietary vitamin K) is more important for individual coumarin requirements than is generally assumed.


Funding: None.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract