Correction to: Nature Communications https://doi.org/10.1038/s41467-018-07617-9, published online 11 December 2018

The original version of the Supplementary information associated with this Article contained an error in Supplementary Table 7 in which the power calculations were incorrect as a result of an error in the standard deviation of the exposure variable when performing the calculations. The correct version of Supplementary Table 7 is:

Supplementary Table 7: What is the power in our replication to detect the most important results in the Överkalix studies?

All-cause mortality result (Kaati et al., 2007)

Males

 Food access

Exposed ancestor

Hazard Ratio

p

Deaths_Överkalix

Deaths/Total N_replication

POWER

 good

father

1.70

0.01

146

3419/3820

>0.99

 good

paternal grandfather

1.45

0.05

164

339/3224

0.48

 poor

paternal grandfather

0.60

0.01

164

339/3224

0.62

Females

 Food access

Exposed ancestor

Hazard Ratio

p

Deaths_ Överkalix

Deaths/Total N, replication

POWER

 good

paternal grandmother

1.75

0.01

139

222/3051

0.50

 poor

paternal grandmother

0.71

0.01

135

222/3051

0.30

Diabetes and cardiovascular mortality results (Kaati et al., 2002)

Diabetes, males and females combined

 Food access

Exposed ancestor

Odds Ratio

p

Deaths/Total N_ Överkalix

Deaths/Total N_ replication

POWER

 good

father

0.14

0.06

19/239

544/7280

>0.99

 good

paternal grandfather

2.34

0.09

19/239

26/6275

0.49

 poor

paternal grandfather

0.35

0.09

19/239

26/6275

0.06

 poor

maternal grandmother

2.73

0.06

19/239

41/5891

0.54

CVD, males and females combined

 Food access

Exposed ancestor

Odds Ratio

p

Deaths/Total N_ Överkalix

Deaths/Total N_ replication

POWER

 poor

father

0.42

0.05

128/239

3846/7280

>0.99

Notes: Hazard ratios and odds ratios as reported in Kaati et al. (2002) and Kaati et al. (2007). The power analyses for all-cause mortality (one-sided test) were computed using Schoenfeld’s sample size-formula for the proportional hazards regression model. For Diabetes and CVD mortality the power was computed using a two-sample proportions test (one-sided test, only decreased individuals in UBCoS were used to determine the sample size).

which replaces the previous incorrect version:

Supplementary Table 7: What is the power in our replication to detect the most important results in the Överkalix studies?

All-cause mortality results (Kaati et al., 2007)

Males

 Food access

Exposed ancestor

Hazard Ratio

p

Deaths_Överkalix

Deaths/Total N_replication

POWER

 good

father

1.70

0.01

146

3419/3820

>0.99

 good

paternal grandfather

1.45

0.05

164

339/3224

0.93

 poor

paternal grandfather

0.60

0.01

164

339/3224

0.99

Females

 Food access

Exposed ancestor

Hazard Ratio

p

Deaths_ Överkalix

Deaths/Total N, replication

POWER

 good

paternal grandmother

1.75

0.01

139

222/3051

0.99

 poor

paternal grandmother

0.71

0.01

135

222/3051

0.72

Diabetes and cardiovascular mortality results (Kaati et al., 2002)

Diabetes, males and females combined

 Food access

Exposed ancestor

Odds Ratio

p

Deaths/Total N_ Överkalix

Deaths/Total N_ replication

POWER

 good

father

0.14

0.06

19/239

544/7280

>0.99

 good

paternal grandfather

2.34

0.09

19/239

26/6275

0.76

 poor

paternal grandfather

0.35

0.09

19/239

26/6275

0.87

 poor

maternal grandmother

2.73

0.06

19/239

41/5891

0.90

CVD, males and females combined

 Food access

Exposed ancestor

Odds Ratio

p

Deaths/Total N_ Överkalix

Deaths/Total N_ replication

POWER

 poor

father

0.42

0.05

128/239

3846/7280

>0.99

Notes: Relative probabilities were reconstructed from sample sizes and odds ratios reported in (Kaati et al., 2002). The power analyses for all-cause mortality were computed using Schoenfeld’s sample-size formula for the proportional-hazards regression model. For Diabetes and CVD mortality the power was computed using a two-sample proportions test (only deceased individuals in UBCoS were used to determine the sample size).

Furthermore, the original version of this Article contained an error in the second paragraph of the ‘Results’ section, which incorrectly read ‘In analyses of G2 mortality this power varied from 72 to 99%.’ The correct version reads ‘In analyses of G2 mortality this power was modest’. These have been corrected in both the PDF and HTML versions of the Article and Supplementary information file. The HTML has also been updated to include a corrected version of the Supplementary information.