Table 2.

Effect of health promotion intervention on change in outcome variable in randomised trial of 267 elderly general practice patients. Intention to treat analysis used assuming baseline status unchanged in 34 participants lost to follow up

Outcome variable	Mean effect size (95% CI)	P value
Change in
Walking (mins/day)	8.4 (−0.31 to 17.6)	0.059
Walking (mins/day as a 5 point scale of quintiles)	0.34 (0.1 to 0.58)	0.005
Walking (minutes in previous fortnight)	88 (8 to 168)	0.032
Walking (minutes in previous fortnight as a 3 point scale of tertiles)	0.27 (0.03 to 0.51)	0.025
Total activity (minutes in previous fortnight)	148 (−146 to 442)	0.342
Total activity (minutes in previous fortnight as a 5 point scale of quintiles)	0.23 (−0.17 to 0.62)	0.303
Score for human activities profile (1-92)*	2.10 (−0.94 to 5.1)	0.175
Psychosocial score (1-100)†	1.28 (−0.54 to 3.1)	0.168
Frequency of interactions with family and friends (6 point scale)	0.03 (−0.14 to 0.21)	0.70
Frequency of pleasurable activities (5 point scale)	0.30 (0.16 to 0.43)	<0.001
Self rated health (5 point scale)	−0.09 (−0.14 to 0.33 )	0.44
Self rated health compared with 12 months ago (5 point scale)	0.28 (0.02 to 0.52)	0.029
Total No of drugs taken	−0.08 (−0.36 to 0.53)	0.71
Drugs taken
Non-steroidal anti-inflammatories	−0.01 (−0.08 to 0.07)	0.82
Psychotropics	0.63 (−0.25 to 1.51)	0.16
Analgesics	0.03 (−0.05 to 0.12)	0.45
Influenza vaccination status	−0.63 (−1.45 to 0.20)	0.14

For continuous outcomes change in outcome was dependent variable and intervention group status independent. For dichotomous outcomes follow up status was regressed on baseline status and intervention group status. All analyses were controlled for general practitioner billing status and effect of cluster design. 

^*0 (worst) to 92 (highest function). 

^†Subscale of sickness impact profile: 0 (no disability related to psychosocial health) to 100 (100% disability related to psychosocial health).