|
Source | Randomisation method | Groups similar at baseline | Loss to follow up | Intention to treat analyses? | Assessors blind? |
|
Exercise | | | | | |
Astengo et al. [16] (2010) | Unclear. Method of randomisation not stated. Authors state randomised “to training group or control.” | Yes Only significant difference was fasting glucose level (higher in intervention (I) group than control (C)). | None. Six patients (9.7%) had major clinical complications and did not complete study. | Unclear | Unclear |
|
Dietary | | | | | |
De Lorgeril et al. [32, 33] (1996, 1999) | Clear. Consecutive patients with first MI randomised. Inclusion based on modified Zelen design. | Yes Except: smokers (more in I group). | Not clear. Rate of withdrawal from follow-up reported: I = 8% C = 7% (De Lorgeril et al. [46], 1994) | Yes | Yes |
|
Psychological | | | | | |
Lewin et al. [29] (2002) | Clear. Eligible patients allocated to I or C group by block randomisation, stratifying for age and sex. | Yes | Not clear. 9 dropouts (6.3%), 1 did not return questionnaires (0.7%), and 2 died (1.4%). | Yes | Yes |
|
Lisspers et al. [26] (1999) | Unclear. Method not stated/described. | Yes; Except C group: significantly higher proportion taking beta blockers. | Data missing from exercise tests (4 I patients, 4 Cs), and questionnaires (numbers not stated). 5-year follow-up (Lisspers et al. [27], 2005): data available from 28 intervention patients (39% loss) and 27 control (34% loss). Further detail not reported. | Unclear | Unclear |
|
Salminen et al. [25] (2005) | Unclear. Method not stated/described. Authors stated that patients identified from longitudinal study data and randomly divided into intervention and control groups. | Yes | None. 12 did not wish to take part in study after randomisation (4.5%); 29 died (10.8%). | Unclear | Unclear |
|
Educational | | | | | |
Carlsson et al. [42, 43] (1997, 1998) | Unclear. Carlsson A and B [42]: patients randomised in groups of 20; no further detail on method given. Carlsson C [43]: patients admitted to coronary care unit who fulfilled inclusion criteria were randomised. | Carlsson A, B, and C: yes | Carlsson A, B, and C: unclear. Carlsson C: 4 (4.9%) of usual care patients lost to follow up; no data given for intervention patients. Authors reported figures for missing data relating to specific variables. | Carlsson A, B, and C: unclear. | Carlsson A, B, and C: unclear. |
|
Cupples and McKnight [22] (1994). | Clear. Patients randomised to one of two groups. A health visitor opened an opaque, sealed, and numbered envelope containing the allocation which had been generated by a computer programme using random permuted blocks. |
Yes | 2 years: I: 25/342 (7.3%). C: 46/346 (13.3%); reasons given 5 years [23]: I: 92/342 (29.9%). C: 109/346 (31.5%); reasons given. |
Yes |
Yes |
|
Heller et al. [44] (1993) | Clear. Patients allocated to I or C groups according to name of GP. General practices stratified by number of doctors and randomly allocated to I or C. | Yes I = 213, C = 237: nos. not similar because randomisation was by GP. | Did not return follow-up questionnaires: I: 45/213 (21.1%) C: 30/237 (12.7%) (no further detail given) |
Unclear |
Unclear |
|
Southard et al. [31] (2003) | Clear. Patients stratified by ethnic group, CR participation, and acute status, then allocated to I or C groups by computer generated random number. |
Yes | 4 (3.84%); reasons stated. |
Yes | Unclear |
|
Multifactorial | | | | | |
Allen et al. [30] (2002) | Clear. Consecutive patients randomised with computerised schema. |
Yes | Unclear. 158 out of 228 completed follow-up (31% loss); reasons for these losses given. |
Yes | Unclear |
|
Campbell et al. [34] (1998) | Clear. Random numbers tables used to centrally randomise patients (by individual after stratification by age, sex, and practice) to I and C groups. |
Yes | Unclear. Available at 1-year analysis: I: 593 (88%) out of original total 673; C: 580 (87%) out of 670; reasons for loss given. |
Yes | Unclear |
Murchie et al. [36] (2003): available at 4-year follow-up analysis: I: 500 (74%); C: 461 (69%); reasons for loss given. |
Delaney et al. [38] (2008): available at 10-year follow-up: I: 385 (57%); C: 365 (54%). These figures include deaths but loss to follow up was reported as 62 patients (4.6%). |
|
Giallauria et al. [18] (2009) | Clear. Patients randomised 1 : 1 using computer generated randomisation. |
Yes |
None |
Unclear |
Unclear |
|
Gianuzzi et al. [21] (2008) | Clear. Patients randomised 1-to-1; randomisation centrally determined by fax at coordinating secretariat using computerised algorithm. |
Yes | 154 (4.7%). |
Yes |
Yes |
|
Hamalainen et al. [45] (1995) | Unclear. Consecutive patients hospitalised for AMI allocated to I or C group by stratified randomisation. No further detail on method given. |
Yes |
Unclear |
Unclear |
Unclear |
|
Murphy et al. [17] (2009) | Clear. Practices: computer generated random numbers; patients: randomly selected at remote site, sent invitation in sequence from lists in random order. | Yes Except: different proportions in control and intervention groups admitted to hospital in previous 12 months. This was adjusted for in analysis. |
None |
Yes |
Yes |
|
Ornish et al. [41] (1998) | Unclear. Method stated; not fully described. After angiography patients randomised using invitational design to minimise crossover, ethical concerns, nocebo effects, and dropout. | Yes Except C group had significantly higher levels of HDL-chol and apolipoprotein A-I. | 20 (71%) intervention patients completed 5-year follow-up; 15 (75%) control patients. |
Yes |
Yes |
|
Redfern et al. [19] (2008) | Clear. Consecutive patients randomised following blinded baseline assessment. Computer generated random allocation sequence was implemented using consecutively numbered envelopes. | Of 15 demographic and clinical characteristics, groups were similar in 9. Statistically different were European and Asian/African origin, employment status, CVD history, and CABG status. | 1 (uncontactable). (4 withdrew, 3 died). |
Yes |
Yes |
|
Vestfold Heartcare Study Group [28] (2003) | Clear. Patients randomised by use of preprepared sealed opaque envelopes containing details on group allocation. Patients opened envelopes so that their group allocation was revealed to them without previous knowledge of study investigators. | Yes | Unclear. Percentages attending follow-up meetings, keeping diaries and adhering to PA levels only given for I group. | Unclear: where there were missing data at 6 months and 2 years, the last recorded value of the variable from previous visit was used. |
Unclear |
|
Wallner et al. [39] (1999) | Unclear. Method of randomisation not stated/described. | Yes Except: statistically significant differences in number with previous MI (I group more), SBP (I group higher), exercise/day (I group more). | Unclear. Patients randomised: I = 28, C = 32. At 12 months, analyses performed in 25 I patients (89%), 13 C (40%) (only patients with complete data for 12 months were included at baseline) because all patients did not agree to complete lengthy questionnaires. Cardiological follow-up was completed in all patients recruited. |
Yes |
Unclear |
|
Organisational | | | | | |
Jolly et al. [40] (1999) | Clear. Practices randomised (before consent sought) to I or C after stratification by size of practice and distance from district general hospital. |
Yes | 10% in both groups; reasons given. |
Yes |
Unclear |
|
Munoz et al. [24] (2007). | Clear. Primary care facilities allocated to I or C by random sequence generated by computer programme. | Yes Except: C group higher proportion had previous hypertension, peripheral vascular disease; higher proportion taking beta blockers, ACE inhibitors; higher BP; I group higher HDL-chol. | 11 (1.1%). Also: Withdrew: 28 (2.8%); Unwilling: 2 (0.2%); Died: 59 (6%). |
Yes |
Unclear |
|