Table 3 Causal relationships between glycaemic traits and psychiatric disorders estimated via two-sample Mendelian randomisation using an inverse-variance weighted effect model with multiplicative random effects.

From: Investigation of glycaemic traits in psychiatric disorders using Mendelian randomisation revealed a causal relationship with anorexia nervosa

Trait one

Trait two

Glycaemic → psychiatric (beta)a

Psychiatric → glycaemic (beta)b

ADHD

Fasting insulin

0.21 (0.34)

−0.0004 (0.02)

ADHD

Fasting glucose

0.20 (0.12)

−0.008 (0.02)

ADHD

HbA1c (all)

0.34 (0.20)

−0.02 (0.01)

ADHD

HbA1c (glycaemic)

0.45 (0.37)

N/A

AN

Fasting insulin

−0.72 (0.20)**

−0.0005 (0.02)

AN

Fasting glucose

−0.12 (0.13)

−0.003 (0.03)

AN

HbA1c (all)

0.05 (0.21)

−0.02 (0.01)*

AN

HbA1c (glycaemic)

−0.28 (0.36)

N/A

ASD

Fasting insulin

−0.06 (0.30)

N/A

ASD

Fasting glucose

−0.14 (0.12)

N/A

ASD

HbA1c (all)

−0.16 (0.15)

N/A

ASD

HbA1c (glycaemic)

−0.38 (0.26)

N/A

BP

Fasting insulin

−0.20 (0.52)

0.003 (0.01)

BP

Fasting glucose

−0.15 (0.18)

−0.009 (0.01)

BP

HbA1c (all)

−0.10 (0.15)

0.012 (0.01)

BP

HbA1c (glycaemic)

−0.51 (0.46)

N/A

MDD

Fasting insulin

−0.17 (0.07)*

−0.01 (0.02)

MDD

Fasting glucose

0.02 (0.04)

0.02 (0.02)

MDD

HbA1c (all)

0.09 (0.05)

0.01 (0.02)

MDD

HbA1c (glycaemic)

−0.11 (0.13)

N/A

OCD

Fasting insulin

0.14 (0.65)

N/A

OCD

Fasting glucose

−0.19 (0.23)

N/A

OCD

HbA1c (all)

0.40 (0.43)

N/A

OCD

HbA1c (glycaemic)

0.38 (0.54)

N/A

SZ

Fasting insulin

0.19 (0.29)

0.02 (0.01)*

SZ

Fasting glucose

−0.13 (0.10)

0.008 (0.01)

SZ

HbA1c (all)

0.01 (0.14)

−0.0005 (0.004)

SZ

HbA1c (glycaemic)

−0.36 (0.26)

N/A

TS

Fasting insulin

0.42 (0.42)

N/A

TS

Fasting glucose

0.15 (0.19)

N/A

TS

HbA1c (all)

0.16 (0.33)

N/A

TS

HbA1c (glycaemic)

0.98 (0.56)

N/A

  1. Mendelian randomisation (IVW estimator with multiplicative random effects) was performed in both directions (subject to the availability of IVs), that is, glycaemic traits as the exposure (glycaemic → psychiatric), along with psychiatric disorders as the exposure. The glycaemic traits were as follows: fasting insulin (BMI adjusted, BMI unadjusted estimates available in Supplementary Tables 1 and 2), fasting glucose, glycaeted haemoglobin (all IVs = HbA1c (all), IVs annotated as glycaemic = HbA1c (glycaemic). Bolded beta estimates are statistically significant. N/A represents analyses where less than three overlapping IVs were available.
  2. The psychiatric traits were: ADHD attention deficit hyperactivity disorder, AN anorexia nervosa, ASD autism spectrum disorder, BP bipolar disorder, MDD major depressive disorder, OCD obsessive compulsive disorder, SZ schizophrenia, TS Tourette’s syndrome.
  3. aIVW beta estimates (standard error) of the effect of glycaemic on the risk of psychiatric disorders represent the log odds of the disorder per unit increase of the exposure. The unit of effects were as follows: fasting insulin = natural log transformed pmol/L, fasting glucose = mmol/L, HbA1c = % HbA1c.
  4. bIVW beta estimates (standard error) using the psychiatric disorders as exposures represent the effect on glycaemic traits per 2.72-fold multiplicative increase in the odds of the psychiatric disorder, however, we treat this primarily as a test of the null hypothesis.
  5. *P < 0.05, **P < 0.01.
Back to article page