Cardiovascular risk management requires a combination of cardiovascular preventive pharmacotherapy and non-pharmacological interventions 

We thank Professor Hoffman for his editorial¹ related to our article ‘Cardiovascular preventive pharmacotherapy stratified by predicted cardiovascular risk: a national data linkage study’.² We support his viewpoint that our results are highly plausible and highlight the complexities of pharmacotherapy uptake among those at highest cardiovascular disease (CVD) risk. We also agree that a unique strength of our study is the inclusion of an entire national population, which renders our results ‘truly representative’ and minimally affected by selection bias. However, we disagree with the two ‘caveats’ that Professor Hoffman suggests should be considered in relation to our findings.

Firstly, Professor Hoffman asserts that our methodology has the ‘severe flaw’ of risk stratification based solely on routinely collected parameters available in administrative data that include baseline dispensing of CVD pharmacotherapy, since dispensing of CVD medications are an outcome of interest for the study. As stated in our article, baseline blood pressure lowering (BPL) and lipid-lowering (LL) medication dispensing were two of the variables used to assign CVD risk whereas our outcomes were BPL and LL dispensing stratified by risk group from our follow-up period (1/04/2013-31/03/2016). We consider the issue that Professor Hoffman refers to as a minor limitation because there was strict temporal separation between the dispensing data used to assign baseline risk and the dispensing data from the follow-up period used for determining outcomes.

In addition, when developing the administrative health data-based equations³ that were used for CVD risk stratification in the current study, likelihood ratio χ² analyses indicated that the most influential predictor of CVD risk was age, with sex-specific values 13–21 times higher than baseline BPL dispensing (the third-ranked predictor). The influence of baseline LL dispensing on the model fit for either sex was negligible. Nevertheless, the patterns of follow-up dispensing according to the risk observed in our study were similar for LL and BPL dispensing, which supports a lack of positive association between use of baseline dispensing of BPL and LL dispensing as predictor variables and our outcomes of interest.

Furthermore, a previous study that we undertook in a clinical cohort⁴ examined BPL and LL pharmacotherapy dispensed from 2006 to 2009 according to baseline risk assigned using a Framingham-based equation. Although BPL and LL medication dispensing were not predictor variables in the Framingham equation, we observed very similar dispensing trends according to CVD risk as in our current study, including slightly greater dispensing of BPL than LL medications. This similarity between the prior analysis and the current study also adds reassurance to our methodology.

The second caveat that Professor Hoffman raises in relation to our findings is that the use of antiplatelet medication was not evaluated as an outcome and could have led to underestimation of beneficial pharmacotherapy. As stated in the article, dispensing of aspirin and other antiplatelet agents was not examined given the evolving scientific literature over the last 10–15 years in relation to use of these medications for primary prevention of CVD.^5,6 Patterns of antiplatelet dispensing during the study period are, therefore, not as straightforward to interpret as BPL and LL therapy and are better examined in a separate study. However, unpublished analyses in our baseline cohort indicate that antiplatelet therapy was uncommon among individuals who were untreated with BPL and/or LL therapy. These unpublished findings are supported by similar findings in separate unpublished analyses undertaken by our research programme in a New Zealand primary care clinical cohort.

As Professor Hoffman notes, non-pharmacological interventions have an important role in the primary prevention of CVD for multiple reasons, including improving quality of life, stress reduction, and mitigation of other long-term conditions. As such, they are incorporated in most CVD risk management guidelines for all individuals irrespective of their CVD risk.⁷ As with CVD preventive pharmacotherapy, though, behavioural modifications are underutilised, with well-documented complexities in their initiation and maintenance.⁸ The relative risk reductions associated with behavioural modifications are also difficult to quantify given the variation in type, intensity and duration of available non-pharmacological interventions and evidence regarding the magnitude of benefits has been slow to emerge.⁷ The relative risk reductions associated with CVD preventive pharmacotherapy targeted to those with elevated absolute CVD risk, on the other hand, are well-established. Published randomized controlled trials indicate that every 1 mmol/L reduction in LDL-C with statin therapy can achieve ∼25% relative risk reduction in CVD events over a 5-year period.⁵ Each 10 mmHg reduction in systolic blood pressure associated with BPL therapy results in ∼20% relative CVD risk reduction over 5 years.⁵ The absolute risk reduction associated with preventive pharmacotherapy is largely dependent on the pre-treatment predicted absolute risk and not the presence of specific modifiable risk factors. Indeed, randomized controlled trials have shown that the absolute risk reduction associated with CVD preventive pharmacotherapy is observed even among normotensive individuals⁹ and those with normal lipid levels.¹⁰ For both pharmacotherapy and non-pharmacological interventions, though, the greatest absolute benefits of interventions occur among those with the highest absolute CVD risk.

Our findings of increasing CVD preventive pharmacotherapy with increased CVD risk group together with a large treatment gap among those at highest CVD risk reflect both current clinical behaviour in New Zealand and patient behaviour. We agree with Professor Hoffman that access to risk management information is lacking for a large proportion of the population and is often accompanied by a reluctance to take medications while asymptomatic. These observations underscore one of the key messages of our article that system-wide measures are required to incentivize clinical risk communication. Conversations about CVD risk and risk management, including with individuals who are currently asymptomatic, require time but are hard to accomplish in short clinical interactions where CVD risk is often not on the patient’s agenda. Our findings emphasize the importance of supporting health provider/patient discussions regarding risk management through system-wide measures such as nationally consistent indicators and targets, as well as funding allocated for these discussions. Promoting opportunities for risk communication can encourage the initiation and/or optimization of both preventive medications among those who meet guideline thresholds and non-pharmacological behavioural modifications. Supporting clinical risk communication can, therefore, ultimately improve CVD outcomes across the risk spectrum.

References

Author notes