Table 4.

Trials in which patients who did not start allocated intervention were excluded from intention to treat analysis

Study	Population	Interventions	Outcome	Exclusions from analysis*
Spruance et al9	2209 patients with history of frequent episodes of herpes simplex labialis	Topical penciclovir or vehicle control cream for recurrence of classic cold sore	Lesion healing	636 patients who did not start treatment
Fazekas et al10	150 patients with relapsing-remitting multiple sclerosis	Monthly intravenous immunoglobulin or placebo	Clinical disability	2 patients in placebo group who withdrew consent between randomisation and start of treatment
CAESAR11	1895 patients infected with HIV-1 and CD4 counts of 25-250×106/l	Addition of lamivudine, or lamivudine and loviride, or placebo to zidovudine based regimens	Progression to new protocol defined AIDS event or death	35 patients who did not start the study treatment. The study was prematurely terminated, but it is not clear whether this was the cause
Landoni et al12	343 women with newly diagnosed stage Ib and IIa cervical cancer	Radical surgery or radiotherapy	Survival	2 patients randomised to surgery: 1 for progression before operation and 1 refused any therapy after randomisation
Rutgeerts et al13	854 patients with bleeding peptic ulcer	Three endoscopic treatments: single injection of polidocanol, or single injection of fibrin glue, or repeated injection of fibrin glue	Endoscopic rebleeding	4 patients in whom, after randomisation, injection treatment turned out to be impossible
Nashan et al14	380 renal allograft recipients	Basiliximab or placebo	Acute rejection	Four patients who received study drug but did no transplant
Jacobson et al15	60 HIV positive patients with oral aphthous ulcers	Thalidomide or placebo	Complete ulcer resolution and change in HIV load	2 patients whose ulcers healed between screening and start of study treatment
Kaplan et al16	198 HIV positive patients with previously untreated, aggressive non-Hodgkin’s lymphoma	Low or standard dose chemotherapy	Survival	1 patient who was never treated due to an acute opportunistic infection and 1 lost to follow up after randomisation but before start of treatment
Englund et al17	839 children with HIV infection	Zidovudin or didanosin, or both	Time to death or progression of HIV disease	7 patients excluded because treatment was refused after randomisation
Tardif et al18	317 patients having elective angioplasty	Combinations of antioxidants or placebo, starting preoperatively	Extent of restenosis	Analysis included “all randomised patients with successful angioplasty” (11 exclusions)
Guilhot et al19	745 previously untreated patients with chronic myelogenous leukaemia	Interferon α-2b and cytarabine or interferon alone	3 year survival	3 patients who declined to participate immediately after randomisation
Daoud et al20	131 patients scheduled for elective heart surgery	Preoperative amiodarone or placebo	Clinical outcome, complications, length of hospital stay, and cost	7 patients in whom surgery was cancelled after randomisation

^*Patients excluded because they did not start the allocated intervention.