Re: Use of dietary linoleic acid for secondary prevention of coronary heart disease and death: evaluation of recovered data from the Sydney Diet Heart Study and updated meta-analysis
We thank Hornstra et al. for their interest in our manuscript . However, their letter includes important errors and misrepresentations, which distract from—but do not alter—the central finding of the Sydney Diet Heart Study. The intervention group, which substituted linoleic acid (n-6 LA) rich liquid safflower oil and safflower polyunsaturated margarine in place of animal fats and common margarines, had increased risk of death from all-causes, cardiovascular disease, and coronary heart disease, despite a significant reduction in blood cholesterol.
Below, we address the most salient errors in their letter. Further, we invite any reader seeking the broader context that is needed to evaluate the results and the limitations of the Sydney Diet Heart Study to read our detailed manuscript , web appendices , and a previous author reply .
Incorrect representation of the numbers of deaths
Hornstra et al. incorrectly assert that there were “63 all cause mortality cases and just 7 additional deaths in the intervention as compared to the control group”. In fact, there were 67 total deaths, 39 of 221 randomized to the n-6 LA intervention group versus 28 of 237 randomized to the control group (a difference of 11). These are the same numbers reported in the 1978 publication . Using standard survival analysis methodology to compare mortality rates between groups on an intention-to-treat basis, we found that the intervention group had increased all cause mortality (17.6% v 11.8%; hazard ratio 1.62 (95% confidence interval 1.00 to 2.64); P=0.05).
Importantly, we were also able to recover and analyze critical outcomes (mortality from cardiovascular disease and coronary heart disease), which were previously unpublished. The intervention group had increased cardiovascular mortality (17.2% v 11.0%; 1.70 (1.03 to 2.80); P=0.04), and mortality from coronary heart disease (16.3% v 10.1%; 1.74 (1.04 to 2.92); P=0.04). These adverse outcomes, which show that subjects randomized to the LA intervention fared worse than subjects randomized to the control group, are the main findings of the trial.
Mischaracterizing limitations of a randomized controlled trial
The Sydney Diet Heart Study was a randomized controlled trial (RCT), not an observational study. All non-dietary components were designed to be equivalent in both groups, allowing for assessment of the effects of the dietary intervention. As noted in our manuscript, both groups made healthy lifestyle modifications (e.g. smoking reduction), probably from the personal reassessment that accompanies a coronary event. This is consistent with other secondary prevention coronary heart disease trials. Since both groups were well balanced at baseline, this RCT design evaluates the effects of the dietary intervention.
As Hornstra et al. noted, a limitation of this RCT is that the intervention group received more intensive treatment than the control group (which had no food provision or dietary counseling). However, a resulting placebo effect would be expected improve outcomes in the intervention group (5). Therefore, the finding of higher death rates in the intervention group—opposite of the expected benefit—may actually underestimate the adverse effects of the intervention.
Mischaracterization as a ‘recalculation study’
Hornstra et al. could have benefited by reading our manuscript  and detailed web appendices  more carefully before mischaracterizing our secondary analysis as a ‘re-calculation’ study. While not all trial records could be recovered, only variables that exactly matched published data were included in our analysis. All matching variables were further verified by Dr. B. Leelarthaepin, an original study investigator and co-author on the present manuscript, in order to ensure accuracy.
The original data included baseline and longitudinal intakes of target nutrients (PUFA, SFA and PUFA to SFA ratio) calculated by the original study investigators. These nutrient variables exactly matched the baseline and follow-up nutrient intake data reported in Tables 2 and 3, respectively, in the 1978 publication , affirming the accuracy of these recovered original data. Some of these nutrient variables were not normally distributed. We therefore appropriately expressed the nutrients as medians (and interquartile ranges) rather than as means (and standard deviations) reported in the 1978 publication. The median baseline and composite follow-up dietary data are reported in Table 3 of our manuscript, and more detailed follow-up dietary data (for 5 years) are presented in Web Appendix Table 2. Together, our presentation of these recovered (not re-calculated) original data provide a more complete depiction of the study diets than was presented in the 1978 publication.
In addition to correcting the erroneous ‘recalculation’ claim, we would like to clarify that the original study investigators did not evaluate the association between increases in n-6 LA and mortality. At the time of the Sydney Diet Heart Study, PUFAs were regarded as a uniform molecular category with one relevant biological mechanism—the reduction in blood cholesterol. Because the original study investigators did not appreciate that individual PUFA species may have different biochemical and clinical effects, they did not attempt to estimate the effects of n-6 LA on mortality outcomes.
Hornstra et al. suggest that a ‘golden rule of Good Clinical Practice’ precludes not only evaluation of missing clinical trial data, but also re-evaluation of completed trials in light of scientific progress. We would like to direct Hornstra et al. to the BMJ theme emphasizing the importance of recovering missing clinical trial data to correct distortions in the evidence-base [6,7]. Our manuscript presents critical—but previously missing—outcomes from a randomized controlled trial, and further evaluates trial data in the context of current understanding of polyunsaturated fatty acids.
Recovery and analysis of these missing data has filled a critical gap in the published literature archive. For the first time, scientists and policymakers have access to cardiovascular and coronary heart disease outcomes from all known randomized controlled dietary trials testing the cardiovascular effects of replacing saturated fats with n-6 linoleic acid, or a mixture of n-6 and n-3 polyunsaturated fatty acids.
1. Ramsden CE, Zamora D, Leelarthaepin B, Majchrzak-Hong SF, Faurot KR, Suchindran CM, et al. Use of dietary linoleic acid for secondary prevention of coronary heart disease and death: evaluation of recovered data from the Sydney Diet Heart Study and updated meta-analysis. BMJ 2013;346:e8707.
2. Ramsden CE, Zamora D, Leelarthaepin B, Majchrzak-Hong SF, Faurot KR, Suchindran CM, et al. BMJ 2013;346:e8707. Web appendices/Supplementary material, pages 1-20.
3. Ramsden CE, Zamora D, Leelarthaepin B, Majchrzak-Hong SF, Faurot KR, Suchindran CM, et al.
Author reply: Can vegetable oil processing help explain the increased cardiovascular mortality in the Sydney Diet Heart Study? (http://www.bmj.com/content/346/bmj.e8707/rr/631590)
4. Woodhill JM, Palmer AJ, Leelarthaepin B, McGilchrist C, Blacket RB. Low fat, low cholesterol diet in secondary prevention of coronary heart disease. Adv Exp Med Biol 1978;109:317-30.
5. Bienenfeld L, Frishman W, Glasser SP. The placebo effect in cardiovascular disease.
Am Heart J. 1996; 132:1207-21
6. Lehman R, Loder E. Missing clinical trial data. BMJ 2012; 344 BMJ 2012;344:d8158
7. Loder E. A theme issue in 2011 on unpublished evidence. BMJ 2011;342:d2627.
Competing interests: No competing interests