Table 1 Unadjusted and adjusted linear models regressing average PROMIS depression symptom T-scores onto average self-reported body temperature.

From: Elevated body temperature is associated with depressive symptoms: results from the TemPredict Study

Model

Model parameter

b

SE

p

95% CI

(LB, UB)

E-value for point estimate (E-value for 95% CI)

Unadjusted

(n = 20,880)

r2 = 0.008

p = 7.1 × 10–38

Intercept

51.488

0.051

 

[51.388, 51.588]

1.60 (1.53)

Self-report body temperature

1.666

0.129

7.1 × 10–38

[1.413, 1.920]

Adjusted

(n = 20,863)

r2 = 0.078

p = 2.3 × 10–128

Intercept

52.940

0.0764

 

[52.790, 53.089]

1.38 (1.30)

Self-report body temperature

0.859

0.126

1.1 × 10–11

[0.612, 1.107]

Age

− 0.103

0.004

9.3 × 10–146

[− 0.111, − 0.096]

Biological sex

− 2.741

0.099

1.8 × 10–165

[− 2.935, − 2.547]

Time of day (B1)

− 0.526

0.117

7.5 × 10–6

[− 0.756, − 0.296]

Time of day (B2)

− 0.774

0.095

4.6 × 10–16

[− 0.961, − 0.588]

  1. Note. Models predict depression symptom T-scores assessed using the Patient-Reported Outcomes Measurement Information System (PROMIS), Adult Health instrument for depression (Form 4a), modified to reflect a 1-month timeframe63. We used survey timestamps to compute the cosine (Time of Day B1) and sine (Time of Day B2) of 2*pi*t for each self-reported body temperature measurement, where “t” was the decimal proportion of the day during which the daily survey was completed (see “Methods”). CI, confidence interval; LB, lower bound; UB, upper bound. Mean self-reported body temperatures were centered around the grand mean of 36.54, such that the intercept represents the mean PROMIS depression T-score for participants with average self-reported body temperatures at the analytic sample mean. Age was centered around the overall analytic sample mean of 46.94.
Back to article page