. 2021 Feb 22;24(4):131–136. doi: 10.1136/ebmental-2020-300232

Table 2.

The output of the Laupacis et al’s NNT estimators for a simulated ‘panss_survival.csv’ dataset

	NNT	CI TR L	CI TR U	CI DL L	CI DL U	CI NBS L	CI NBS U	CI PBS L	CI PBS U
NNT-KM(7)	4.05	2.32	15.89	1.03	7.07	2.92	6.65	NA	NA
NNT-COX(7)	4.21	1.84	Inf	1.00	9.63	3.46	5.39	3.09	5.32
NNT(7\|100)	2.30	1.00	Inf	1.00	24.14	1.90	2.89	1.70	2.90

NNT: point estimate of the NNT.

The output of the Laupacis et al’s NNT estimates for a simulated ‘panss_survival.csv’ dataset. The chosen fixed timepoint for adjustment is seven (ie, 7 months after the treatment admission), and the adjusted value of the baseline PANSS is 100. The table presents the point estimates of the unadjusted Kaplan-Meier NNT (NNT-KM(7)), marginal Cox NNT (NNT-COX(7)) and adjusted (NNT(7|100)) Laupacis et al’s NNT for a survival time until the discontinuation of the treatment. Additionally, the table presents the corresponding transformation-based (TR), delta-method-based (DL), non-parametric-bootstrap-based (NBS) and parametric-bootstrap-based (PBS) 95% CIs. Inf, infinity. For the unadjusted NNT point estimator, the PBS-based CI limits are not available (NA) since the NNT-KM is a non-parametric estimator.

CI DL L, the lower limit of the 95% level delta-method-based CI; CI DL U, the upper limit of the 95% level delta-method-based CI; CI NBS L, lower limit of the 95% level non-parametric-bootstrap-based CI; CI NBS U, the upper limit of the 95% level non-parametric-bootstrap-based CI; CI PBS L, the lower limit of the 95% level parametric-bootstrap-based CI; CI PBS U, the upper limit of the 95% level parametric-bootstrap-based CI; CI TR L, the lower limit of the 95% level transformation-based CI; CI TR U, the upper limit of the 95% level transformation-based CI; Inf, Infinity; NA, Not available; NNT, number needed to treat; NNT(7|100), Cox regression-based estimator of the adjusted NNT for timepoint 7 and covariate value of 100; NNT-COX(7), Cox regression-based estimator of the marginal NNT for timepoint 7; NNT-KM(7), the Kaplan-Meier-based non-parametric maximum likelihood estimator of the unadjusted NNT for timepoint 7; PANSS, Positive and Negative Syndrome Scale.