Objective: To determine whether there are significant circadian, weekl
y or lunar variations in self-poisoning presentations and whether pati
ents' names or dates of birth have an influence on the likelihood of s
elf-poisoning by analysing biorhythms, numerology and star sign. Setti
ng: Hunter Valley, Australia. Subjects: Consecutive adult patients adm
itted with self-poisoning between January 1987 and June 1993. Results:
There were 2215 patients admitted. There was a marked circadian varia
tion. Over 6% of all admissions occurred in each of the hours between
6 p.m. and 1 a.m. compared with less than 2% per hour between 5 a.m. a
nd 9 a.m. This pattern was not different for patients with a diagnosis
of depression. Numerology, biorhythms and star signs had no significa
nt correlations with self-poisoning, nor was there a significant weekl
y or yearly variation in presentations. There was a small but statisti
cally significant sex difference in presentations analysed by lunar ph
ases. At the new moon 60% of self-poisonings were in women, compared w
ith 45% when the moon was full. The odds ratios (OR) for women to be a
dmitted at full moon and at new moon were 1.27 (95% confidence interva
l [CI], 0.92-1.66; P value not significant) and 0.73 (95% CI, 0.57-0.9
2; P = 0.009) respectively. The mean illumination of the moon at the t
ime of overdose was 50.63% +/- 0.91% for men, compared with 47.45% +/-
0.85% for women (P= 0.014). Conclusion: The circadian cycle (but not
weekly, yearly or mystical cycles) should be taken into account when d
etermining staffing levels for poison information and casualty service
s. The full moon is protective for women.