Intended for healthcare professionals

Letters Calcium supplementation

Results may not be generalisable

BMJ 2008; 336 doi: (Published 21 February 2008) Cite this as: BMJ 2008;336:403
  1. Joan M Lappe, professor of nursing and medicine,
  2. Robert P Heaney, John A Creighton university professor, professor of medicine
  1. 1Creighton University, Omaha, NE 68178, USA
  1. jmlappe{at}

Bolland et al showed an increased risk of vascular events, particularly myocardial infarction, in healthy postmenopausal women receiving calcium supplements.1 2 We recently completed a similar study with a primary skeletal end point.3 4 We queried our database about the principal adverse events reported by Bolland et al.

We followed up 1179 women for four years, similar to the 1471 women followed up by Bolland et al. The calcium supplement dosage was 1400-1500 mg/day, nearly 50% higher than in Bolland et al, and principally with the same product. We recorded adverse events at six monthly intervals, verified them with the participant’s personal physician, and entered in the database using the corresponding ICD-9 code.

Our study had three treatment arms: a double placebo, calcium plus vitamin D, and calcium and a vitamin D placebo. The total number treated with calcium (with or without additional vitamin D) was 892, and the person years of exposure to calcium was 3568, almost identical to the figure in Bolland et al (3660).

The vascular event rate across all groups was 5.3/1000 persons/year, less than half the rate of Bolland et al (12.2/1000). This difference may be because the participants in the study of Bolland et al were on average seven years older. We found no excess occurrence of myocardial infarction or other vascular events in either calcium treatment group compared with placebo, despite the higher calcium supplement dose. When the two calcium treatment groups were combined the vascular event rate is 4.76/1000 persons/year, the rate in the placebo group being 6.94 events/1000 persons/year. This difference, while not reaching significance, is in a direction opposite to that found by Bolland et al.

We conclude that the excess of events found by Bolland et al was either a chance occurrence or at least not generalisable to other populations.


  • Competing interests: None declared.


View Abstract