Impact of switching from different treatment regimens to a fixed-dose combination pill (polypill) in patients with cardiovascular disease or similarly high risk: The influence of baseline medication on LDL-cholesterol, blood pressure and calculated risk reduction when switching to a cardiovascular polypill Free

Abstract

Aims

Cardiovascular fixed-dose combination pills, or polypills, may help address the widespread lack of access and adherence to proven medicines. Initiation of polypill-based care typically entails switching from current separately taken medications. Given the heterogeneity in usual care, there is interest in the impact of polypill treatment across different patterns of prior medication regimen.

Methods

A total of 2004 participants with established cardiovascular disease or estimated 5-year cardiovascular risk of over 15% were randomised to polypill-based treatment (aspirin 75 mg, simvastatin 40 mg, lisinopril 10 mg and either atenolol 50 mg or hydrochlorothiazide 12.5 mg) or usual care. Baseline medications were classified by potency relative to polypill components. Estimated cardiovascular risk reduction was calculated by combining risk factor changes with results seen in meta-analyses of previous randomised trials.

Results

For cholesterol reduction conferred by polypills, there was a dose response across baseline statin groups, with mean low-density lipoprotein (LDL)-cholesterol differences of 0.37, 0.22, 0.14 and 0.07 mmol/L among patients taking no statin, less potent, equipotent and more potent statin at baseline, respectively. Similarly there were differences in mean systolic BP of 5.4, 6.2, 3.3 and 1.8 mmHg among patients taking 0, 1, 2 or 3 BP-lowering agents. Among patients taking more potent statins at baseline, there was no significant difference in LDL-cholesterol but there were benefits for BP and aspirin adherence. Similar results were seen among patients taking 3 BP-lowering agents at baseline. Switching to a polypill-based strategy resulted in estimated cardiovascular relative risk reductions across a wide range of usual care patterns of antiplatelet, statin and BP-lowering therapy prescribing.

Conclusion

Adherence benefits from switching to a polypill resulted in risk factor changes that were at least as good as usual care across a wide variety of treatment patterns, including equally potent or more potent regimens. The benefits of switching to polypill-based care were greatest among those stepped up from partial treatment or less potent treatment.

Introduction

In patients at high risk of cardiovascular disease there is clear evidence that treatment with statins and blood pressure (BP)-lowering agents is highly effective in reducing cardiovascular morbidity and mortality. In those with established cardiovascular disease the additional use of an antiplatelet drug is recommended to lower the risk of cardiovascular events independent of statin and BP-lowering therapy.¹ Although clearly recommended in clinical guidelines, large treatment gaps between indicated therapy and prescribed medication are observed worldwide, with very low rates of use of effective therapies in middle and low-income countries.² A multitude of barriers may underlie suboptimal prescribing rates and low treatment continuation rates in high-risk patients. While in low-income settings issues of availability and affordability are paramount, the complexity of preventive treatment regimens for both treating physicians and patients confer major difficulties irrespective of income level.³ Dose complexity and the number of pills prescribed are inversely related to patients’ adherence.^4,5

Fixed-dose combination (FDC) pills, ‘polypills’, containing multiple agents addressing various risk factors have been suggested as a novel strategy to improve accessibility to treatment, cost aspects, and patients’ adherence to treatment.⁶ Several studies indicate that FDC pills can produce sizeable risk factor reductions when compared to placebo, with a halving of predicted cardiovascular risk expected if risk factor reductions were maintained long term.^7–9

Evidence was extended by the results from the use of a multidrug pill in reducing cardiovascular events (UMPIRE) trial, showing that a polypill-based strategy compared to usual care improves adherence to guideline-indicated therapy, and lowers both low-density lipoprotein (LDL)-cholesterol and systolic BP in participants with established cardiovascular disease or those at equivalent high risk (an estimated 5-year cardiovascular disease risk of ≥15%) during a median follow-up of 15 months.¹⁰ Allocation to the polypill-based treatment strategy improved the use of indicated medication by one third (relative risk (RR) 1.33; 95% confidence interval (CI) 1.26 to 1.41), and reduced both mean LDL-cholesterol by 0.11 mmol/L (95% CI 0.05 to 0.17) and mean systolic BP by 2.6 mmHg (95% CI 1.1 to 4.0).¹⁰

While overall the UMPIRE trial suggested the adherence advantages for the polypill outweighed the disadvantages from reduced tailoring of treatment, important questions remain about the generalisability of this finding across patient groups. Are benefits restricted to patients switched from less potent treatment to polypill-based care? Is there evidence of harm in patients switched from separate pill regimens with more potent agents compared to polypill-based care? The UMPIRE trial provides a unique opportunity to assess these questions, given the heterogeneity of usual care at baseline. This analysis aims to provide details of baseline medication regimens in the UMPIRE trial and to evaluate the effects on cardiovascular risk factors of switching from different treatment regimens to polypill-based care.

Methods

Study design

The present report is a post-hoc analysis of the UMPIRE trial. The UMPIRE trial was a randomised, open label, blinded endpoint clinical trial comparing a polypill-based treatment strategy with usual care in participants with established cardiovascular disease or those at a similarly high risk (estimated 5-year cardiovascular risk of ≥15%) in India and in three European countries (England, Ireland and the Netherlands). Participants were randomly assigned to continuation with usual care or to a polypill-based treatment strategy with follow-up planned for 12 to 24 months. The primary objective of the UMPIRE trial was to assess whether provision of a polypill compared to usual medications improved adherence to indicated therapy and, as a consequence, whether it improved the LDL-cholesterol and systolic BP. The rationale, design, conduct and the main outcomes have been described in detail elsewhere and these details are also available at www.spacecollaboration.org.^10,11 The UMPIRE trial is registered with the European Clinical Trials database (https://eudract.ema.europa.eu/index.html): EudraCT number 2009-016278-34 and with the Clinical Trials Registry, India (http://www.ctri.in/Clinicaltrials/index.jsp): number CTRI/2010/091/000250. Ethics approval was granted by the relevant committees in each participating country.

Study population

Men and women aged 18 years or older at a high cardiovascular risk (defined as either established cardiovascular disease with a history of coronary heart disease, ischaemic cerebrovascular disease, or peripheral vascular disease or an estimated 5-year cardiovascular risk of ≥15%) were eligible. There had to be clear indications and no contraindications for the components of at least one of two FDC formulations.

Participants were excluded if the treating doctor considered that changing medications was clinically inappropriate, if there was a known situation in which medication might be altered for a prolonged interval, or if the participant was unlikely to complete trial procedures. Participants in Europe were recruited via research databases, hospital clinics and general practice registries to three trial centres in London, UK, in Dublin, Ireland and in Utrecht, the Netherlands. Indian participants were recruited via hospital specialist clinics in 28 centres across the country.

Treatment and randomisation

Randomisation was conducted 1:1, polypill treatment:usual care ratio. Participants randomly assigned to usual care continued to be treated at the discretion of their routine doctor consistent with current local guidelines. Participants randomly assigned to the polypill strategy were prescribed one of two FDC formulations at the discretion of the treating physician: polypill version 1 (containing aspirin 75 mg, simvastatin 40 mg, lisinopril 10 mg and atenolol 50 mg) or version 2 (containing aspirin 75 mg, simvastatin 40 mg, lisinopril 10 mg and hydrochlorothiazide 12.5 mg). Physicians had the discretion to add additional medications, stop the FDC pill and begin treatment with separate medications, or switch version of the FDC pill.

Trial procedures

The minimum scheduled follow-up was 1 year. The end-of-study visit for all participants was scheduled to take place around 12 months after random assignment of the last participant and all participants were followed up for as long as possible within that interval. The end-of-study visits took place around the same calendar period and the planned range of follow-up was 12 to 24 months. Participants attended clinic visits for random assignment at 12 months and at the end of the study. Telephone or clinic visits were conducted at 1 month and 6 months. When co-incident or proximate, a 12-month visit or a 18-month visit could become the end-of-study visit. Fasting lipids and BP were measured at baseline and visits. Clinic BP was measured with an electronic oscillometric BP monitor (Omron 705CP II; Omron Healthcare, Fremont, USA) and paper printouts were logged.

Outcomes

The outcomes of the present analyses were the use of antiplatelet, statin and BP-lowering therapy in the usual care group compared to the polypill-based treatment group and the treatment effect on LDL-cholesterol, systolic BP and estimated cardiovascular RR reduction in relation to the potency of previous therapy.

Use of medication

Self-reported use of all prescribed and over-the-counter medication was assessed at each trial visit. Use of medication was defined as taking the medication for at least 4 days during the week preceding the visit. All cardiovascular medication was categorised into drug classes.

For each dose of statins the estimated LDL-cholesterol reduction was categorised as described in a large meta-analysis.^12,13 The estimated LDL-cholesterol reduction of simvastatin 40 mg was 37% (95% CI 35% to 40%). Statins with an estimated LDL-cholesterol reduction of 32% to 40% were defined as equally potent as the polypill (fluvastatin 80 mg, pravastatin 80 mg, simvastatin 40 mg, atorvastatin 10 mg and rosuvastatin 5 mg) as these had similar estimated LDL reductions. Statin therapy was defined as less potent than the polypill if the estimated LDL-cholesterol reduction was less than 32% and more potent if the estimated LDL-cholesterol reduction was over 40%.

All BP-lowering agents were classified into drug classes (angiotensin-converting enzyme (ACE) inhibitor, beta-blocker, diuretic, angiotensin receptor blocker, calcium antagonist, alpha blocker, rest group. As a meta-analysis showed similar reductions in BP for each standard dose for all classes of agents, the standard dose equivalent was calculated.¹⁴ The standard dose equivalent was calculated as the used daily dose divided by the defined standard dose.¹⁵ The polypill contained an ACE inhibitor at 1 standard dose equivalent (lisinopril 10 mg) and either a beta-blocker (atenolol 50 mg) at 1 standard dose equivalent in version 1 or a diuretic (hydrochlorothiazide 12.5 mg) at 0.5 standard dose equivalent in version 2.

Cardiovascular risk reduction

It is essential to have an outcome that incorporates changes resulting from all three treatment modalities (antiplatelet therapy, LDL-cholesterol and BP-lowering), given potentially divergent effects on each treatment modality. A common metric is required to compare, for example, a subgroup with improved LDL-cholesterol control but detrimental effects in BP to a subgroup with improved BP control and antiplatelet adherence but no change in LDL-cholesterol. The predicted reductions in cardiovascular risk were therefore estimated using data from systematic reviews and meta-analyses, which have shown that each medication class confers approximately similar proportional reductions in outcomes across a wide range of patient populations, with no major differences between agents after accounting for the extent of risk factor reduction for LDL-cholesterol and systolic BP. In other words, each 1 mmol/L lowering of LDL-cholesterol results in a proportional risk reduction, irrespective of the effect of other agents on risk factors. A similar effect on cardiovascular risk is seen for 10 mmHg systolic BP reduction or the use of aspirin.

The reduction of LDL-cholesterol and systolic BP was calculated as the difference from baseline to follow-up. The combined effects were calculated by multiplying RRs after adjusting for the size of LDL-cholesterol (per 1 mmol/L) and systolic BP reductions (per 10 mmHg) per individual.⁷ Aspirin, 1 mmol/L LDL-cholesterol reduction, and 10 mmHg systolic BP reduction each individually lower the risk of coronary events by 19%, 23% and 22%, respectively (RR 0.81, 0.77 and 0.78, respectively). The expected joint effects of use of aspirin, a 1.5 mmol/L LDL-cholesterol reduction and 5 mmHg systolic BP reduction would result in a 52% lower cardiovascular risk (overall RR 0.81 × 0.77^1.5/1.0× 0.78^5/10= 0.42).^12,15,16 Combining the calculated combined treatment effects with data on current event rates on coronary events, ischaemic strokes and haemorrhagic strokes provides the best estimates of expected treatment effect. These rates were used to calculate the individual cardiovascular risk reduction on major events.⁷ The natural logarithm of the RR was calculated to obtain a mean and accompanying 95% CI of the log transformed RRs. The mean was exponentiated to obtain the overall estimated RR reduction on a linear scale.

Statistical analysis

The primary analyses of changes in mean LDL-cholesterol and systolic BP at 12 months between the polypill-based treatment and usual care groups were conducted using an analysis of covariance including the randomised treatment. Data from baseline, 1 month, 6 and 12 months were used in the analyses as all participants were scheduled to attend these visits. Adjusted analyses included treatment and baseline value of the corresponding risk factor. Continuous variables are presented as mean and standard deviation (mean ± SD) in a normal distribution. Categorical variables are presented as a percentage of the total (n (%)). All analyses were conducted on an intention-to-treat basis with no imputation of missing data. Results are considered to be statistically significant at a P value less than 0.05.

Results

Clinical characteristics

In total, 2004 individuals were randomly assigned (1000 in India, 1004 in Europe) between July 2010 and July 2011. Among the 1002 participants randomly assigned to the polypill, version 1 was used by 589 (59%) participants and version 2 by 413 (41%) participants. Baseline characteristics of the usual care group and polypill-based treatment strategy group are displayed in Table 1.

Antiplatelet agents

At baseline 881 (88%) participants in the usual care group and 892 (89%) in the polypill-strategy group reported the use of aspirin. After randomisation the use of aspirin increased in both the usual care and the polypill group.

At 12 months aspirin use was reported by 834 (91%) in the usual care group versus 883 (95%) in the polypill group (RR polypill versus usual care (RR_{PP vs. UC}) 1.05; 95% CI 1.02 to 1.07) (Supplementary Table 1).

Statin therapy

Statin use was reported by 874 (87%) participants in the usual care group versus 882 (88%) in the polypill group at baseline. After randomisation the reported use of statins increased in both groups, more so in the polypill group (929 (94%) vs. 982 (99%) participants, respectively; RR_{PP vs. UC} 1.08; 95% CI 1.06 to 1.10). The proportion of patients taking more potent statin regimens remained constant in the usual care group (at 40%) but decreased to 2% in the polypill group (see Supplementary Table 2).

At 12 months 839 (91%) participants in the usual care group reported the use of a statin versus 885 (95%) participants in the polypill group (RR_{PP vs. UC} 1.05; 95% CI 1.04 to 1.07) (Table 3). There remained a large difference in the proportion of patients taking more potent statin regimens (39% usual care vs. 7% polypill group) (see Supplementary Table 2).

Baseline statin therapy and LDL-cholesterol

The LDL-cholesterol levels at baseline and 12 months are displayed in Table 4. Overall, at 12 months the mean LDL-cholesterol was 0.16 mmol/L (95% CI 0.09 to 0.21) lower in the polypill group after adjusting for baseline LDL-cholesterol levels (Table 2).

In patients with no statin at baseline the difference between the treatment groups was largest (mean difference _{PPvs. UC} –0.37 mmol/L; 95% CI –0.59 to –0.16). In those reported to use a less potent statin at baseline, switching to a polypill resulted in 0.22 mmol/L (95% CI 0.02 to 0.41) lower mean LDL-cholesterol. Also in those with an equally potent statin compared to the polypill at baseline the mean LDL-cholesterol was 0.14 mmol/L (95% CI 0.05 to 0.22) lower in the polypill group. In those prescribed a more potent statin at baseline switching to a polypill did not result in a statistically significant difference in levels of LDL-cholesterol (mean difference _{PPvs. UC} –0.07 mmol/L; 95% CI –0.16 to 0.03) (Table 2).

BP-lowering therapy

At baseline one BP-lowering agent was reported to be used by 213 (21%) participants, two agents by 435 (43%) and three or more agents by 288 (29%) in the usual care group and by 256 (26%), 405 (40%) and 263 (26%) in the polypill group, respectively. At 12 months in the usual care group 197 (20%) participants reported using one agent, 411 (45%) two agents and 188 (22%) three or more agents compared to 52 (6%), 574 (62%) and 279 (30%) in the polypill group, respectively (see Supplementary Table 3).

At baseline the mean total number of BP-lowering agents reported to be used was 2.1 ± 1.0 in the usual care group and 2.0 ± 1.1 in the polypill group. The mean total standard dose equivalent was 2.3 ± 1.6 and 2.3 ± 1.6, respectively. At 12 months the reported total number of BP-lowering agents was 2.1 ± 1.0 in the usual care group versus 2.3 ± 0.9 in the polypill group. The mean standard dose equivalent was 2.3 ± 1.6 and 2.2 ± 1.1, respectively (see Supplementary Table 3). The differences in drug classes are displayed in Supplementary Table 4.

Baseline BP-lowering therapy and systolic BP

The systolic BP levels at baseline and 12 months are displayed in Table 3. Overall, at 12 months the mean systolic BP was 3.3 mmHg (95% CI 1.9 to 4.8) lower in the polypill group, after adjusting for baseline systolic BP levels. In patients with none or one BP-lowering agent at baseline the use of the polypill resulted in a 5.4 (95% CI 11.4 to 0.6) and 6.2 mmHg (95% CI 9.4 to 3.1) lower systolic BP. In participants taking two BP-lowering agents at baseline, there were no clear differences between groups in the average number of agents (2.0 vs. 2.1) or total standard dose (2.1 vs. 2.0). However, switching to the polypill resulted in a 3.3 mmHg (95% CI 1.2 to 5.4) lower mean systolic BP. In those with three or more BP-lowering agents at baseline, by 12 months the average number of agents was similar in usual care compared to polypill (3.2 vs. 3.1) and there was higher total dosage in the usual care group (3.5 vs. 3.1), but switching to a polypill did not result in statistically different levels of systolic BP (mean difference _{PP vs. UC} –0.4 mmHg 95% CI –3.4 to 2.7). Similar patterns were observed when BP regimens were stratified according to the number of standard doses per day.

Cardiovascular risk reduction

The overall estimated mean cardiovascular risk reduction was 12.6% (95% CI 6.6 to 18.2) greater in the polypill group. Switching cardiovascular preventive treatment to a polypill-based strategy resulted in a greater cardiovascular risk reduction across a wide range of usual care patterns of antiplatelet, statin and BP-lowering therapy prescribing (Table 4). Of note, there was no heterogeneity in the predicted cardiovascular risk reduction when treatment regimens were categorised by the intensity of individual treatment modalities. Essentially lesser reductions in one modality were offset by greater reductions in other modalities. There was a trend to increasing the overall predicted cardiovascular risk reduction with decreasing levels of baseline treatment.

In the subgroup of 192 (10%) patients taking aspirin, a more potent statin and three or more BP-lowering agents at baseline there was no clear evidence of adverse risk factor trends or predicted cardiovascular risk reduction after switching to the polypill.

Discussion

Switching to a polypill-based treatment strategy increased the use of aspirin, and lowered levels of both LDL-cholesterol and systolic BP when compared to the continuation of usual care among a high-risk population with a wide variety of baseline treatment regimens. In participants taking less or equally potent medication at baseline as compared to the components of a polypill, switching to a polypill resulted in improvements in LDL-cholesterol and systolic BP. In participants switched from more potent treatments, there was no difference in the levels of LDL-cholesterol or systolic BP. Switching to a polypill resulted in predicted cardiovascular risk reduction across a wide range of usual care patterns of antiplatelet, statin and BP-lowering therapy prescribing.

The overall effect of therapy is determined by the potency of the drugs prescribed and whether those drugs are actually taken long term. This study showed that a polypill with moderately potent components achieves similar outcomes as individual medicines that are more potent. These findings are novel, given this is the first large trial to assess the use of FDC trials compared to usual care, with all the heterogeneity that that entails. One potential concern of a polypill-based strategy is that reduced choice of medications, limited flexibility to titrate agents and/or unfamiliarity with the use of a polypill as background treatment leads to suboptimal risk factor control.^17,18 Yet this trial showed that these issues were not sufficient to offset improvements in risk factor control brought about by improved adherence. In part, this was due to the fact that a polypill strategy does not rule out tailored care on top of a FDC.

Improvement of adherence in patients with a high cardiovascular risk is an important principle of a polypill strategy in high-risk individuals.⁶ There are several mechanisms whereby a polypill strategy may diminish barriers to adherence, which include patients’ acceptability, number of pills used a day and ease of using the medication. Although this issue has not been explored extensively, there is evidence similar to our data that indicates better medication adherence when combination treatments are used in the management of hypertension and other conditions.^19,20 Previous articles have suggested that targeting multiple risk factors to achieve a general cardiovascular risk reduction is often likely to be more productive than focussing on treatment targets for individual risk factors.^21,22 The World Health Organization suggested that increasing adherence to therapy may have a greater impact on the health of the population than any new intervention.²³

Further research is required to determine if polypills containing more potent agents and/or more dose versions could improve rather than match outcomes compared to optimal usual care. Such information may come from ongoing initiatives. However, the most important finding from the current research is that currently available polypills can result in clinically important improvements and risk factor control and possibly adherence among patients currently undertreated – so called ‘step-up substitution’.

The strengths of the UMPIRE trial include the large sample size, duration of follow-up and completeness of data collection. This study was the first trial to assess the effect a polypill-based treatment strategy compared to usual care in individuals at high risk of cardiovascular events, making the results generalisable to standard clinical practice. However, there are several issues to be considered when interpreting the results. Most importantly, the trial is likely to underestimate the benefits in a general secondary prevention setting with typical adherence levels, as volunteers for clinical trials tend to be relatively more motivated and clinical management in a trial setting tends to be more intensive than usual care.²⁴ The proportion of untreated or partially treated patients was far lower in UMPIRE than in the general routine secondary prevention setting population.²⁵ Improvements in the polypill group were observed compared to a usual care group in whom treatment rates initially rose and remained higher than baseline throughout the study, whereas adherence typically reduces over time.^26,27 However, at least some of the benefits in the polypill group (especially when compared to more potent treatment regimens) can be attributed to this high level of care, with 42% of patients being prescribed BP-lowering medications on top of the polypill and 7% of patients being switched from the polypill to a more potent statin. In addition, as stated previously in the main paper, regulations in India and in Europe mandated the distribution of a trial treatment by the approved trial teams and prohibited charging for the prescription of trial treatments. Consequently, the polypill group received their treatment free of charge from the trial centres, which might have affected the adherence of i.e. expensive medication. However, a large US trial showed that the elimination of co-payments for core cardiovascular medicines improved adherence by about 5% in absolute terms.²⁸ Current data suggest a greater increase in adherence. Furthermore, post hoc analysis in the main study comparing adherence between patients exempt from medication charges and those who were not exempt showed similar effects of the polypill strategy.^10,29,30 Currently marketed polypills cost around 10–15 rupees (0.15–20 Euros) per day, which is within the range of the sum of separate components. Preferably pricing will be reduced in the near future, which might also increase adherence due to budgetary reasons in real life.

In the era of evidence based-medicine physicians call for randomised trial data on the effect of the FDC on clinical outcome rather than on predicted risk. To establish superiority in trials comparing two good established approaches is a tremendous undertaking with thousands of patients. One may wonder whether the time, effort and budget should be spent on such huge initiatives, in particular because all the components of the cardiovascular FDC have already shown that their use is related to a reduction of risk.

Patient-related outcomes are currently deemed important in providing care. So treatment that is effective and well appreciated by patients may be preferred. With respect to cardiovascular FDC pills, the data from multiple studies suggest that by far the majority of the patients randomly assigned to the polypill favour that treatment above the treatment with separate pills.

In conclusion, adherence benefits from switching to a polypill resulted in risk factor changes that were at least as good as usual care across a wide variety of treatment patterns. The benefits of switching to polypill-based care were greatest among those stepped up from partial treatment or less potent treatment. However, there was also evidence of benefit when switched from equally potent regimens, and switching from a more potent regimen did not result in a detriment in risk factor control or estimated cardiovascular risk reduction.

Acknowledgements

The authors gratefully acknowledge the members of the UMPIRE study group.

Author contribution

All authors contributed to the conception or design of the work. All contributed to the acquisition, analysis, or interpretation of data for the work. ML, WS, FV and AR drafted the manuscript. All authors critically revised the manuscript. All gave final approval and agree to be accountable for all aspects of the work ensuring integrity and accuracy. The lead author (the manuscript’s guarantor) affirms that the manuscript is an honest, accurate and transparent account of the study being reported, that no important aspects of the study have been omitted, and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/coi_disclosure.pdf and declare George Health Enterprises, the social enterprise arm of the George Institute for Global Health, has received investment to develop aspirin, a statin, blood pressure-lowering combination product.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The UMPIRE study work was supported by the European Commission, 7th framework programme under the theme Health-2009-3.1-4: Improved treatment of chronic diseases in developed countries (grant no. 241849). The substudy on CIMT progression was supported by a grant from the Netherlands Organisation for Health research and Development (ZonMw) (grant no. 200320013).

References

Author notes