Intended for healthcare professionals

CCBYNC Open access

Rapid response to:

Research

Mortality from different causes associated with meat, heme iron, nitrates, and nitrites in the NIH-AARP Diet and Health Study: population based cohort study

BMJ 2017; 357 doi: https://doi.org/10.1136/bmj.j1957 (Published 09 May 2017) Cite this as: BMJ 2017;357:j1957

Rapid Response:

Re: Mortality from different causes associated with meat, heme iron, nitrates, and nitrites in the NIH-AARP Diet and Health Study: population based cohort study

Etemadi and colleagues reported that red meat was associated with increased risk of all-cause mortality (1).

I calculated 33 outcomes adjusted with 20 covariates just for red meat in their Figure 1. One can specify around 35 million Cox regression models with the combination of 33 outcomes and 20 covariates (33 x 2^20 = 34603008).

In addition, I calculated, for example, 264 P-values for different linear trends in supplementary tables A-D.

Perhaps significance threshold 0.05 is simply meaningless within this framework.

But let's put these calculations aside.

Recently, Schwingshackl et al reported that high versus low consumption of red meat was associated with an average risk ratio (RR) of 1.10 (95% CI 1.00 to 1.22) for all-cause mortality (supplementary figure 25) (2).

Statistical heterogeneity was evident in their meta-analysis (I^2 93%, tau^2 0.02) and of 12 comparisons, four showed risk ratios below 1 and eight showed risk ratios above 1 (2). One of the included study in their meta-analysis was an earlier publication of the NIH-AARP Diet and Health Study (3).

I replicated their meta-analysis with the use of R software and meta package (4). However, I replaced the aforementioned earlier publication of the NIH-AARP Diet and Health Study (3) by Etemadi et al (1). Furthermore, I calculated prediction interval (PI) which provides expected range for the true association in a new analogous study.

Average RR (95% CI) was 1.10 (1.01 to 1.20) for the association between high versus low consumption of red meat and all-cause mortality in replicated meta-analysis (I^2 92.1%, tau^2 0.0185 with DerSimonian-Laird estimator).

On the other hand, 95% PI was 0.80 to 1.51 so therefore in some future circumstances association between high versus low consumption of red meat and all-cause mortality could be protective.

In sensitivity analysis, average RR was 1.10 (95% CI 0.99 to 1.21, tau^2 0.0255, 95% PI 0.75 to 1.59) with the use of method by Paule and Mandel which has been recommended for estimation of between-study variance (5).

PIs provide more meaningful interpretation of heterogeneous random-effects meta-analyses (6). As Higgins et al stated:
“…appropriate consideration must be given to the whole distribution to avoid misleading generalizations about effects across studies…” (6).

Red meat might be harmful, on average, but less so if one appreciates the distribution of effects depicted by predictive interval.

References

1. Etemadi A, Sinha R, Ward MH, et al. Mortality from different causes associated with meat, heme iron, nitrates, and nitrites in the NIH-AARP Diet and Health Study: population based cohort study. BMJ 2017;357:j1957.

2. Schwingshackl L, Schwedhelm C, Hoffmann G, et al. Food groups and risk of all-cause mortality: a systematic review and meta-analysis of prospective studies. Am J Clin Nutr 2017 Apr 26 pii: ajcn153148.

3. Sinha R, Cross AJ, Graubard BI, Leitzmann MF, Schatzkin A. Meat intake and mortality: a prospective study of over half a million people. Arch Intern Med 2009;169:562-71.

4. Schwarzer Guido. meta: An R package for meta-analysis. R News 2007;7(3):40-45.

5. Langan D, Higgins JP, Simmonds M. Comparative performance of heterogeneity variance estimators in meta-analysis: a review of simulation studies. Res Synth Methods 2016 Apr 6. doi: 10.1002/jrsm.1198

6. Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-analysis. J R Stat Soc Ser A Stat Soc 2009;172:137-159.

Email: jesper.m.kivela@helsinki.fi

Competing interests: I like statistics.

01 June 2017
Jesper M. Kivelä
MD, PhD student (pediatrics)
Institute of Clinical Medicine, University of Helsinki
Helsinki, Finland