Table 2 |.

Best-fit ARIMA models for interrupted time series analysis on the effect of multicomponent intervention on smoking and urine cotinine trajectories, by quitting status

	Mean daily no of cigarettes					Mean urine cotinine level (range, 0–6)
	Best-fit ARIMA model		Postintervention change			Best-fit ARIMA model		Postintervention change
	Model (p, d, q)a	p for residual autocorrelationb	Mean change	SE	p	Model (p, d, q) a	p for residual autocorrelationb	Mean change	SE	p
All (n = 30)
Early intervention (n = 9)	(1,0,0)	.732	–5.88	0.36	<.001	(1,0,0)	.552	–1.91	0.28	<.001
Delayed intervention (n = 14)	(1,0,0)	.357	–5.22	0.25	<.001	(1,0,0)	.486	–2.08	0.35	<.001
Late intervention (n = 7)	(1,0,1)	.113	–4.04	0.59	<.001	(1,0,0)	.831	–2.73	0.30	<.001
Quitters (n = 21)
Early intervention (n = 7)	(1,0,0)	.585	–6.52	0.16	<.001	(1,0,1)	.450	–2.07	0.53	<.001
Delayed intervention (n = 10)	(1,0,1)	.498	–5.34	0.08	<.001	(1,0,1)	.435	–2.76	0.37	<.001
Late intervention (n = 4)	(1,0,2)	.238	–4.67	0.28	<.001	(1,0,1)	.798	–3.75	0.23	<.001
Nonquitters (n = 9)
Early intervention (n = 2)	(1,0,0)	.504	–4.52	0.76	<.001	(0,0,1)	.988	–0.02	0.06	.715
Delayed intervention (n = 4)	(1,0,0)	.697	–6.07	0.31	<.001	(0,0,1)	.999	–0.04	0.07	.562
Late intervention (n = 3)	(1,0,0)	.920	–8.50	1.73	<.001	N/Ac

ARIMA, autoregressive integrated moving average; SE, standard error.

^aARIMA model terms: p is the order of the autoregressive part; d is the order of the differencing; q is the order of the moving-average process.

^b p Value for autocorrelation of residuals to lag 6.

^cARIMA model not applicable because of the constant mean urine cotinine Level 6 throughout the 8-week intervention period.