Table 1.

The impact of introducing a new TB diagnostic test to laboratories on unfavourable outcomes among TB patients, using different analysis methods. Mixed‐effects models used observation‐level data; within‐period methods used cluster‐level data

Analysis Method	Intervention Effect	P‐Value
	Odds ratio
Non‐parametric within‐period method	0.78 (0.61, 0.96)	0.02
Parametric within‐period method	0.85 (0.74, 0.98)	0.02
Cluster mixed‐effect model	0.83 (0.67, 1.03)	0.10
Cluster‐period mixed effect model	0.83 (0.67, 1.03)	0.10
	Risk difference
Non‐parametric within‐period method	−4.8% (−10.0%, −0.3%)	0.04
Parametric within‐period method	−4.2% (−7.2%, −1.1%)	0.007