Table 3

Dyspnoea results

Study	No of patients	Outcome measure¹	Mean on opioid	Mean on placebo	Pooled SD	Mean difference	SE² crude/corrected	SMD (95% CI)
¹Change from baseline or post-treatment measurement.
² “Crude” standard errors ignore crossover designs (or assume a correlation of 1); “corrected” standard errors are from paired analyses.
*Standard errors imputed using correlations observed in other studies.
†Standard errors based on assumptions of zero correlation coefficient.
Oral and parenteral opioid studies
Bruera⁹	10	Change	–14	4	23.52	–18	10.5/10.5†	–0.77 (–1.65 to 0.11)
Chua¹⁰	12	Post	2.91	3.60	0.82	–0.69	0.33/0.18	–0.85 (–1.28 to –0.42)
Eiser¹¹ (a)	10	Post	7.0	6.5	2.3	0.5	1.03/0.58*	0.22 (–0.27 to 0.71)
Eiser¹¹ (b)	8	Post	6.2	6.1	1.98	0.1	0.99/0.71	0.05 (–0.66 to 0.76)
Johnson¹²	18	Post	6.7	7.6	2.11	–0.9	0.7/0.23	–0.43 (–0.62 to –0.23)
Light¹³	7	Change	–0.14	0.14	0.49	–0.28	0.26/0.26†	–0.57 (–1.61 to 0.47)
Poole¹⁴	14	Change	–2.50	–0.44	4.38	–2.06	1.66/1.35	–0.47 (–1.06 to 0.12)
Woodcock¹⁵	12	Change	–1.39	–0.16	1.61	–1.23	0.66/0.66†	–0.77 (–1.57 to 0.03)
Woodcock¹⁶	11	Post	–45.1	–42.5	9.25	–2.6	3.95/2.23*	–0.28 (–0.75 to 0.19)
Nebulised opioid studies
Davis¹⁹	70	Change	–9.58	–7.66	20.89	–1.91	3.53/3.3	–0.09 (–0.41 to 0.22)
Leung²²	10	Change	–0.72	–0.45	1.05	–0.28	0.47/0.53	–0.26 (–1.26 to 0.74)
Noseda²⁴	14	Post	37.2	40.3	27.13	–3.1	10.3/4.07	–0.11 (–0.40 to 0.18)

Meta-analysis	No of studies	Test for heterogeneity			Test for overall effect			SMD (95% CI)
All studies	12	Q=19.4 (p=0.05)			p=0.0008			–0.31 (–0.50 to –0.13)
Non-nebulised studies	9	Q=13.86 (p=0.09)			p=0.0006			–0.40 (–0.63 to –0.17)
Nebulised studies	3	Q=0.1(p=0.95)			p=0.31			–0.11 (–0.32 to 0.10)
COPD studies	9	Q=10.47(p=0.2)			p=0.004			–0.26 (–0.44 to 0.08)