Table 2:

Difference-in-differences calculation of nonfatal self-injury in the first 5 years after diagnosis

Analysis	Time period	Exposure	N	NFSI frequency	Follow-up time (person-years)	Rate per 10 000 (95% CI)	Difference in rates	Relative rate	Estimates^*
Analysis	Time period	Exposure	N	NFSI frequency	Follow-up time (person-years)	Rate per 10 000 (95% CI)	Difference in rates	Relative rate	DID	Ratio of relative rates^†
Year 0–1 v. 5 year preindex	Pre-index	Cancer	803 740	3704	3 952 538	9.37 (9.07–9.68)	0.73	1.08	–	–
	Pre-index	Controls	1 607 480	6 02	7 870 255	8.64 (8.44–8.85)	0.73	1.08	–	–
	Post-index	Cancer	803 740	716	688 600	10.40 (9.66–11.19)	2.16	1.26	1.43	1.16
	Post-index	Controls	1 607 480	1264	1 534 334	8.24 (7.80–8.71)	2.16	1.26	1.43	1.16
Years 1–5 v. 5 year preindex	Pre-index	Cancer	550 313	2366	2 703 413	8.75 (8.41–9.11)	0.23	1.03	–	–
	Pre-index	Controls	1 100 626	4596	5 394 840	8.52 (8.28–8.77)	0.23	1.03	–	–
	Post-index	Cancer	550 313	1347	1 563 586	8.61 (8.17–9.09)	0.79	1.10	0.56	1.07
	Post-index	Controls	1 100 626	2626	3 356 057	7.82 (7.53–8.13)	0.79	1.10	0.56	1.07
Years 0–5 v. 5 year preindex	Pre-index	Cancer	803 740	3704	3 952 538	9.37 (9.07–9.68)	0.73	1.08	–	–
	Pre-index	Controls	1 607 480	6802	7 870 255	8.64 (8.44–8.85)	0.73	1.08	–	–
	Post-index	Cancer	803 740	2221	2 461 014	9.02 (8.66–9.41)	1.41	1.19	0.68	1.09
	Post-index	Controls	1 607 480	4538	5 960 260	7.61 (7.40–7.84)	1.41	1.19	0.68	1.09

Note: DID = difference in differences (difference in the postindex period – difference in the preindex period), NSFI = nonfatal self-injury.
↵* Unadjusted estimates are presented.
↵† Ratio of relative rates: relative rate in the postindex period / relative rate in the preindex period.