Practices and patients
In six collaborating general practices with 10 GPs, in Katwijk (an urbanised rural town in the Netherlands), serving a total of 17 200 patients (2500–3000 per practice), patients were selected who had no known target organ damage but were receiving medication for hypertension and/or hypercholesterolaemia. In the Netherlands, all patients are registered at a general practice, which also keeps the electronic patient record with information on diagnoses and medication.
How this fits in
According to the new Dutch guideline for cardiovascular risk management, patients with a low risk of cardiovascular mortality may have insufficient benefit to warrant medication. In this study, the feasibility and consequences of a re-evaluation programme for patients without target organ damage treated for hypertension and/or hypercholesterolaemia in general practice was explored. The results suggest that stopping medication in patients with a low risk for cardiovascular mortality is safe. GPs’ and nurse practitioners’ views seem to play a role in advising to stop or to restart medication.
From the electronic patient records, the study selected all patients aged 25–75 years who were prescribed drugs in the past 12 months with the following cardiovascular-related Anatomical Therapeutic Chemical (ATC) codes: C02 (antihypertensives), C03 (diuretics), C07 (beta-blocking agents), C08 (calcium-channel blockers), C09 (agents acting on the renin–angiotensin system), and C10 (lipid-modifying agents).11 All patients who were coded as being diagnosed with cardiovascular disease were excluded, that is, codes K74, K75, K76, K89, K90, K91, K92, and T90, according to the International Classification of Primary Care, version 1 (ICPC-1).12 The selected patients were then manually screened and excluded in the case of non-coded cardiovascular disease, or when taking drugs for indications other than hypertension or hypercholesterolaemia (for example, for migraine). Patients on the final list were invited by mail to re-evaluate their risk of cardiovascular mortality and were asked for written informed consent to participate in the present study.
Re-evaluation
In the period January 2008 to November 2009 (23 months), patients were invited for a re-evaluation by one of the nurse practitioners at the general practice. At intake the nurse practitioners registered comorbidity (asthma/chronic obstructive pulmonary disease [COPD], digestive tract, urogenital tract, psychosocial, and/or musculoskeletal problems), number of oral drugs taken (0–2/>2), number of oral cardiovascular drugs taken (0–2/>2), prevalence of cardiovascular disease in first-degree relatives (no/yes), and smoking behaviour (no/yes).
The nurse practitioners then calculated the patient’s risk of cardiovascular mortality in the next 10 years, according to the new Dutch guideline. For patients treated for hypertension or hypercholesterolaemia, the medical history was searched for the pretreatment levels of either blood pressure or cholesterol. If unknown, a systolic blood pressure of 180 mmHg was assumed, and, in the case of medical treatment for hypercholesterolaemia, a HDL-cholesterol to cholesterol ratio of 8.0 was assumed. These levels were chosen because this strategy would minimise the number of underestimated levels (as a sort of worst-case scenario). For patients aged between 66 and 75 years, the calculated risk for patients aged 65 years was used, which is the highest ranked age reported in the guideline.
The new guideline advises medication in cases of a high calculated risk: a 10-year risk of cardiovascular mortality of 10% or more or, in the case of an additional risk factor (for example, obesity or family history), of 5% or more. In cases of a lower risk, patients were advised to visit their GP to discuss continuation of medication. If patients decided to stop medication, the GP handed them a personalised medication-reduction schedule which, in complicated cases (such as polypharmacy), had been discussed with a pharmacist (for example, Box 1). Then, based on previous reports of long-term diuretic medication,13 and a relapse of high blood pressure in the case of withdrawal,8–10,14 an individual follow-up took place, which involved measuring blood pressure, serum lipid levels, and possible adverse events (for example, oedema, headache, high blood pressure, high cholesterol, feeling unwell, other problems). In cases of stopping diuretics, follow-up took place after 1, 2, 4, and 12 weeks, for stopping other antihypertensive medication after 4 and 12 weeks, and after stopping only statins after 12 weeks.
Box 1 Example of a medication-reduction scheme
Patient Awith a calculated 3% risk of cardiovascular mortality
Medication: hydrochlorthiazide 1 daily dose 25 mg, metoprolol 1 daily dose 50 mg
Reduction scheme:
Week 1: hydroclorthiazide 1 daily dose 25 mg and metoprolol 1 daily dose 25 mg
Week 2: hydroclorthiazide 1 daily dose 25 mg and metoprolol 1 daily dose 12.5 mg
Week 3: follow-up control; hydroclorthiazide 1 daily dose 25 mg and metoprolol stop
Week 4: hydroclorthiazide stop
Weeks 5, 7, 9, and 16: follow-up control
For patients who decided to stop medication, it was checked whether they restarted cardiovascular drugs within 6 months and, for those who did, the most important reason(s) for doing so were registered. Whatever the decision, all patients were offered a follow-up.
Registration and analysis
The outcome measures were: (1) attendance for re-evaluation, (2) calculated cardiovascular risk, (3) advice to stop medication, (4) patients’ decision to stop medication, and (5) patients’ compliance with the discontinuation at 6 months.
To determine indicators for successful stopping of medication, first, univariate analyses were performed; Student’s t-test in the case of a normal distribution and the Wilcoxon test in other cases. In the multivariate backward logistic regression analyses, variables with P<0.20 in the univariate analysis were included (P entry 0.05, P removal 0.10). To assess the prognostic value of determinants, a prognostic model was built by backward stepwise logistic regression, and the area under the receiver operating characteristic curve (AUC) was calculated. In addition, differences between practices were analysed by one-way analysis of variance and the Bonferroni post hoc test. Data analysis was performed with the Statistical Package for Social Sciences for Windows (SPPS 17.0).