Summary
Background
European guidelines recommend that antihypertensive management should be graded as a function of total cardiovascular risk.
Aims
To examine the multilevel (patient- and physician-level) determinants of blood pressure and residual total cardiovascular risk outcomes associated with second-line valsartan therapy.
Methods
The BSCORE study was a prospective, multi-centre, pharmacoepidemiological study of the “real-world” effectiveness of second-line valsartan with or without hydrochlorothiazide.
Results
A total of 3497 patients were recruited by 354 physicians. Mean age was 63.8 ± 12.0 years; 52.3% were male; 20.9% were smokers; 47.7% were dyslipidaemic; and 23.6% had diabetes. On average, reductions in blood pressure and increases in the proportions of patients with controlled blood pressure after 90 days were statistically significant (all P < 0.001). Twenty-one percent of systolic blood pressure and 25.6% of diastolic blood pressure variability at follow-up was attributable to physician-level characteristics. Significant reductions in total cardiovascular risk were observed ( P < 0.001); with 12.5% of the variability in total cardiovascular risk change attributable to physician-level characteristics. Several independent determinants of blood pressure outcomes were identified, many of which are modifiable.
Conclusions
Second-line valsartan therapy improves blood pressure outcomes under variable real-world conditions, and is associated with a decrease in total cardiovascular risk. Optimizing antihypertensive effectiveness, including the reduction of residual cardiovascular risk, involves managing concomitant conditions and risk factors, improving adherence, and identifying physician-level factors amenable to intervention.
Résumé
Justification
Les recommandations européennes indiquent que la prise en charge de l’hypertension artérielle doit être basée sur l’évaluation du risque cardiovasculaire global.
Objectifs
Examiner les déterminants à deux niveaux (patients et médecins) de la pression artérielle ainsi que le risque cardiovasculaire global, associés à un traitement par valsartan de deuxième ligne.
Méthodes
L’étude BSCORE est une étude prospective, multicentrique, pharmaco-épidémiologique d’évaluation de l’efficacité dans le monde réel d’un traitement par valsartan de seconde ligne, avec ou sans hydrochrlorothiazide.
Résultats
Trois mille quatre cent quatre-vingt-dix-sept patients ont été recrutés par 354 médecins. L’âge moyen était de 63,8 ± 12 ans ; 52,3 % des patients étaient des hommes, 20,9 % des fumeurs ; 40,7 % présentaient une dyslipidémie et 23,6 % un diabète. En moyenne, la réduction de la pression artérielle et l’augmentation de la proportion de patients ayant une pression artérielle contrôlée dans les 90 jours étaient statistiquement significatives ( P < 0,001). Vingt-et-un pour cent des chiffres de pression artérielle systolique et 25,6 % de la variabilité des chiffres de pression artérielle diastolique lors du suivi étaient attribuables à l’intervention médicale. Une réduction significative du risque cardiovasculaire global a été observée ( P < 0,001) ; avec une variabilité de 12,5 % du risque cardiovasculaire global lié aux caractéristiques de l’intervention médicale. Plusieurs déterminants indépendants de l’évolution de la pression artérielle ont été identifiés dont plusieurs sont des facteurs de risque modifiables.
Conclusions
Un traitement par valsartan de deuxième ligne améliore le contrôle de la pression artérielle, dans les conditions d’évaluation du monde réel, et est associé à une diminution du risque cardiovasculaire global. Optimiser l’efficacité du traitement hypotenseur en incluant la réduction du risque cardiovasculaire global, implique la prise en charge des pathologies associées concomitantes et des facteurs de risque, contribue à l’amélioration de l’observance, et permet d’identifier des facteurs modifiables par une intervention thérapeutique liée au médecin.
Introduction
Approximately 50% of cardiovascular disease can be attributed to suboptimal BP control . The ESH-ESC Guidelines and the 2009 ESH guideline reappraisal advocate quantifying patients’ TCVR based on BP values and other risk factors or associated organ damage, and grading the intensity of antihypertensive management as a function of TCVR. The SCORE project developed 10-year risk models of fatal cardiovascular disease for various European countries . The Belgian model (B-SCORE), which incorporated gender, age, smoking status, systolic BP, and total cholesterol level, was calibrated based on national mortality statistics and prevalence estimates of major cardiovascular risk factors . The ESH-ESC has recommended that TCVR classification should be part of clinical assessment and treatment planning, but it has not been used as an effectiveness outcome.
We recently applied a novel analytical framework for observational “real world” effectiveness studies to examine the determinants and outcomes of second-line antihypertensive treatment with valsartan (the PREVIEW study) . This framework assumes that patients seen by the same physician share a commonality defined by that physician’s knowledge, experience, and expertise (physician class effect). As such, statistical independence of patients “nested” under physicians cannot be assumed, and hierarchical (or multilevel) analytical approaches are warranted . From a sample of 3194 Belgian patients cared for by 504 GPs, we found that about a quarter of the variability in BP values at 90 days was attributable to the compliance of these GPs with the ESH-ESC recommended triggers for treatment initiation and intensification, their practice patterns, experience in treating hypertension, and years in practice . The remainder of the variance in BP was accounted for by patient-level variables, many of which were modifiable and manageable by evidence-based medical care and effective patient self-care. Similar factors were identified in an analysis of why certain patients did not achieve BP control.
PREVIEW was the first observational study of its kind in hypertension, and we believe it is in need of replication. The B-SCORE risk classification model enabled us to not only replicate the PREVIEW study, but also to assess physician- and patient-level variables influencing changes in TCVR and to identify determining factors reducing TCVR after second-line treatment with valsartan with or without hydrochlorothiazide (hereafter referred to as valsartan).
Methods
The observational BSCORE study (so named to emphasize its link to the B-SCORE risk model) was designed as a prospective, open-label, multicentre, multilevel (patients nested under physicians) pharmacoepidemiological study of predictors of, and changes in, BP and TCVR in patients with hypertension after 90 days of treatment with valsartan. Being partly a replication of PREVIEW, the methodology was virtually identical to the PREVIEW study , except for the addition of TCVR calculation according to the B-SCORE model. Only key elements of the methodology are summarized below. Further details are available from the corresponding author on: recruitment, screening and enrolment procedures; sample size calculations; schedule of assessment; complete data model; details on variables and measurement; procedures for data collection and management; and statistical analysis methods.
To be eligible, patients had to be consenting adults (male or female) who did not tolerate and/or who did not benefit sufficiently from prior antihypertensive treatment, as evidenced by systolic BP (SBP) ≥ 140 mmHg and/or diastolic BP (DBP) ≥ 90 mmHg (except for patients with diabetes, in which case investigators were asked to consider the 2007 ESH-ESC recommended levels of SBP ≥ 130 mmHg and DBP ≥ 80 mmHg); and who were started de novo on one of the following commercially available valsartan formulations: 80 mg, 160 mg, 80 mg/12.5 mg hydrochlorothiazide, 160 mg/12.5 mg hydrochlorothiazide, or 160 mg/25 mg hydrochlorothiazide. Patients could be on either valsartan mono- or combination therapy. Patients who did not tolerate prior antihypertensive treatment were eligible, hence the presence of 2% of patients with baseline normal and high normal BP in the sample. The evaluable sample, defined primarily as patients with baseline and follow-up BP values, included 3497 patients contributed by 354 from all over Belgium.
Patient baseline data included: demographics; anthropometrics; hypertension and cardiovascular history; comorbidities; lifestyle; prior antihypertensive medications; BP; TCVR; clinical status; starting valsartan dose; concomitant antihypertensive and other relevant medications; and adherence. Data recorded at 90-day follow-up were: BP; residual TCVR; clinical status; changes in valsartan dose since previous visit; concomitant medication(s) taken or changed since previous visit; self-reported adherence within the past 4 weeks; and side-effects over the past 90 days. Only patient data collected routinely in clinical practice were recorded; no additional tests or exams were ordered.
Separately, and using a specially developed questionnaire, we also assessed several physician variables that might influence BP or TCVR outcomes: practice type, location/setting, patient mix; demographics; sources of information and knowledge related to hypertension; hypertension management practices; prescription patterns; management of adverse effects; SBP/DBP thresholds for treatment initiation and intensification; perceptions of patient adherence; and knowledge of practice guidelines.
BP was measured three times at 1–2-minute intervals, in a sitting position after 5 minutes of rest. The mean was recorded as the mean sitting SBP and mean DBP. BP control was defined as per the 2007 ESH-ESC guidelines as SBP < 140 mmHg and DBP < 90 mmHg; except for patients with diabetes mellitus and/or high or very high TCVR, in which case targets were SBP < 130 mmHg and DBP < 80 mmHg . Of note, the 2009 update advocates a 140/90 mmHg cut-off for all populations, but our study was conducted under the 2007 guidelines .
TCVR was computed based on the cross-classification of BP by risk factors (e.g. smoking and dyslipidaemia), metabolic syndrome, organ damage, diabetes, and established cardiovascular or renal disease (myocardial infarction, coronary blood vessel disease, heart failure, cerebrovascular conditions, peripheral vascular conditions, and renal conditions [defined as serum creatinine > 1.5 mg/dL]) . Possible classifications included: average risk, low added risk, moderate added risk, high-added risk, and very high-added risk. We computed “change in TCVR” by subtracting TCVR at baseline from TCVR at the end of treatment, yielding possible scores from −4 (greatest TCVR improvement possible; from very high added to average risk) to + 4 (greatest TCVR worsening; from average to very high added risk). Patients with established cardiovascular or renal disease were not included in the TCVR calculations and analyses as they cannot improve according to the ESH-ESC TCVR classification. We also classified patients dichotomously as having achieved, or not, a reduction of at least one level of TCVR at 90 days (excluding patients who were in the average risk category at baseline and those with established cardiovascular or renal disease).
The ESH-ESC guidelines , recommend different BP thresholds for the general population and for patients with diabetes and/or high/very high risk. Hence, using baseline data, we classified patients in a binary manner as average to moderate added risk (LOW risk) or diabetes and/or high added/very high-added risk (HIGH risk). This classification was used in the evaluation of BP values and BP control over the 90-day study period.
Descriptive statistics of frequency, central tendency, and dispersion were applied under consideration of levels of measurement. The t -test for dependent samples was used to measure BP changes over time; the McNemar test was used for changes in proportions in BP control and TCVR reduction of at least one level. We used conditional hierarchical/multilevel linear modelling to identify patient- and physician-level determinants of changes in BP and TCVR; unconditional hierarchical/multilevel linear modelling to calculate the proportion of variability in these variables attributable to physicians (ICC); and logistic regression modelling to identify independent predictors of uncontrolled BP and improvement in TCVR.
Results
Patients
Among 3497 patients, the mean age was 63.8 ± 12.0 years (range 19–98 years); 13.0% were < 50 years old, 52.4% were aged 50–69 years, and 34.6% were ≥ 70 years old. Around half were male (52.3%), 20.9% were smokers, 23.6% had diabetes, and 47.7% were dyslipidaemic. With regard to cardiovascular conditions, 11.0% had coronary blood vessel disease, 8.9% peripheral vascular conditions, 8.3% had a history of myocardial infarction, 7.6% had cerebrovascular conditions, and 5.5% had heart failure. Only 3.6% had renal disease, and 19.7% had no pre-existing conditions.
For the 3475 patients (99.4%) for whom binary risk status at baseline could be calculated, 1359 patients (39.1%) were in the LOW-risk and 2116 patients (60.9%) were in the HIGH-risk category at baseline. TCVR could be calculated for 2344 patients (67.0%) who had data available at both baseline and follow-up. Note that 2% of patients had normal or high normal BP at baseline. These were patients who, despite therapeutic benefits, did not tolerate prior BP-lowering treatment and were converted to a valsartan regimen.
Effectiveness outcomes
Blood pressure
Reductions in mean SBP and DBP, and increases in the proportions of patients with controlled BP after 90 days on valsartan ( Table 1 ), were statistically significant in the entire sample and in the LOW-risk and HIGH-risk groups (all P < 0.0001). At baseline, the HIGH-risk group had significantly higher mean SBP and DBP levels than the LOW-risk group (both P < 0.001) However, this was not the case at follow-up, indicating a proportionately higher reduction in BP among HIGH-risk patients.
