Table 4.

Results from multinomial logistic regression: CR wait time (late CR or CR wait time in days) and change in anxiety and depression category.

	Observed data				Imputed data
Change in HADS category	Change in anxiety category RRR (95% CI)		Change in depression category RRR (95% CI)		Change in anxiety category RRR (95% CI)		Change in depression category RRR (95% CI)
	Late CR	CR wait time	Late CR	CR wait time	Late CR	CR wait time	Late CR	CR wait time
Non-symptomatic to symptomatic	Reference group
Symptomatic to non-symptomatic	0.85 P = 0.04 (0.74, 0.99)	0.99 P = 0.006 (0.994, 0.9991)	0.81 P = 0.01 (0.68, 0.95)	0.99 P = 0.46 (0.997, 1.001)	0.95 P = 0.37 (0.86, 1.05)	0.99 P = 0.16 (0.99, 1.00)	0.93 P = 0.20 (0.83, 1.03)	0.99 P = 0.85 (0.99, 1.00)
No change: symptomatic	1.04 P = 0.58 (0.89, 1.21)	0.99 P = 0.85 (0.997, 1.001)	1.06 P = 0.42 (0.90, 1.24)	1.002 P = 0.004 (1.0008, 1.004)	1.02 P = 0.58 (0.93, 1.13)	0.99 P = 0.96 (0.99, 1.00)	1.03 P = 0.59 (0.90, 1.18)	1.001 P = 0.16 (0.99, 1.00)
No change: non- symptomatic	0.93 P = 0.26 (0.82, 1.05)	0.99 P = 0.03 (0.996, 0.9998)	0.85 P = 0.01 (0.74, 0.97)	0.99 P = 0.09 (0.997, 1.0002)	0.98 P = 0.78 (0.90, 1.07)	0.99 P = 0.29 (0.99, 1.00)	0.94 P = 0.27 (0.85, 1.04)	0.99 P = 0.43 (0.99, 1.00)

RRR: relative risk ratio; CI: confidence interval; CR: cardiac rehabilitation.

Analyses adjusted for age, gender, comorbidity, CR duration, ethnicity, relationship status, employment, history of previous cardiac event, treatment received and year of initiating event.

Data were clustered with CR centres using cluster-robust standard errors.