Controlled trials examining the effectiveness of using a risk score to aid primary prevention of CVD
| Study and participants | Intervention | Results* | Study quality | ||
|---|---|---|---|---|---|
| Outcome† | Intervention (95% CI/SD) | Control (95% CI/SD) | |||
| *Results for Montgomery et al unadjusted for baseline blood pressure (BP) and practice computer system. | |||||
| †Change in numbers of patients receiving treatment or being referred to dietician. | |||||
| ‡Mantel–Haenszel p<0.02. | |||||
| §Odds ratio (OR) <1 indicates less likely to take >1 class of drug. | |||||
| CDSS, clinical decision support system; CVD, cardiovascular disease; DBP, diastolic blood pressure; GP, general practitioner; MI, myocardial infarction; SBP, systolic blood pressure; TC, total cholesterol. | |||||
| Hall et al32: Hospital outpatients with type 2 diabetes, age 35–75, 52% men | CVD risk score documented at front of patient notes | All patients | n = 162 | n = 161 | Alternate allocation of participants |
| Diabetes treatment | 42% (34 to 50) | 36% (29 to 45) | Doctors were unaware of allocation | ||
| Hypertension treatment | 16% (10 to 22) | 10% (5 to 16) | Length of follow up, losses to follow up unreported | ||
| Lipid-lowering treatment | 12% (7 to 17) | 9% (4 to 14) | |||
| Referral to dietician | 10% (6 to 15) | 13% (7 to 19) | No power calculation | ||
| High-risk patients (>20% 5-year risk) | n = 86 | n = 82 | |||
| Diabetes treatment | 44% (35 to 54) | 35% (24 to 47) | |||
| Hypertension treatment | 23% (15 to 31) | 10% (3 to 17) | |||
| Lipid-lowering treatment | 20% (12 to 27) | 9% (2 to 15) | |||
| Referral to dietician | 10% (5 to 16) | 7% (1 to 17) | |||
| Montgomery et al29: General practice patients with hypertension, age 60–80, 46% men | 1. Computer-based CDSS and risk chart | n = 202 | n = 130 | Cluster randomisation of general practices | |
| Change in mean 5-year risk ⩾10% | 0.65 (0.39) | 0.77 (0.37) | Participants not blinded to study group | ||
| Change in SBP (mm Hg) | −0.04 (1.4) | 0.25 (1.7)‡ | Losses to follow up were 10% at 6 months and 14% at 12 months | ||
| Change in DBP (mm Hg) | 0.36 (0.74) | −1.64 (1.03)‡ | |||
| OR§ for taking 2 classes of drugs v 0 or 1 | 0.5 (0.2 to 0.9) | 0.5 (0.2 to 1.0) | Power calculation using intracluster correlation coefficient from a published study | ||
| OR for taking ⩾3 classes of drugs v 0 or 1 | 0.3 (0.1 to 0.6) | 0.3 (0.2 to 0.7) | |||
| 2. Chart only, interventions by GP or practice nurse. 6 and 12 month follow up | n = 199 | ||||
| Change in mean 5-year risk ⩾10% | −0.48 (0.35) | ||||
| Change in SBP (mm Hg) | 2.66 (1.4) | ||||
| Change in DBP (mm Hg) | −1.1 (0.78) | ||||
| OR§ for taking 2 classes of drugs v 0 or 1 | 1.0 | ||||
| OR for taking ⩾3 classes of drugs v 0 or 1 | 1.0 | ||||
| Hanon et al31” Patients with hypertension, age 19–74, 54% men | 10-year CVD risk communicated to GP as <15%, 15–20%, 20–30%, >30%. Controls had CVD risk estimated by physician. 8 week follow up | n = 556 | n = 712 | ||
| Framingham Anderson 1991 10-year CVD risk | 26 (12) | 25 (12) | Individual randomisation | ||
| Physicians not blind to intervention | |||||
| Change in SBP (mm Hg) | −27 | −26 | No information on assessor blinding | ||
| Change in DBP (mm Hg) | −15 | −16 | 17% lost to follow up | ||
| Patients with BP <140/90 mm Hg | 64% | 62% | No power calculation | ||
| Patients taking 2 hypertension drugs rather than 1 | 41% | 46% | No confidence intervals reported | ||
| Hetlevik et al30: General practice patients with hypertension, age range not reported, 42% men | Computer-based CDSS incorporating Westlund MI risk score | Change in SBP (mm Hg) | −2.3 (n = 816) | 0.8 (n = 1023) | Cluster randomisation of health centres |
| Change in DBP (mm Hg) | −1.8 (n = 816) | −1.2 (n = 1023) | 56/213 (26%) of invited GPs were randomly assigned | ||
| Change in serum TC (mmol/l) | 0.04 (n = 581) | −0.13 (n = 768) | Unblinded | ||
| Uptake of computerised CDSS was only 12% in intervention group | |||||
| Some risk score variables were missing for >91% of participant | |||||
| No power calculation | |||||
| No confidence intervals reported | |||||