Correspondence
1 Schwalm J-D, McCready T, Lopez-Jaramillo P, controlled trials, and multiplies relative
et al. A community-based comprehensive
intervention to reduce cardiovascular risk in
hypertension (HOPE 4): a cluster-randomised
controlled trial. Lancet 2019; 394: 1231–42.
2 Jafar TH, Samad Z, Bloomfield GS. Parallel
community solutions for cardiovascular risk
reduction. Lancet 2019; 394: 1207–08.
3 McManus RJ, Mant J, Haque MS, et al. Effect of
self-monitoring and medication self-titration
on systolic blood pressure in hypertensive
patients at high risk of cardiovascular disease:
the TASMIN-SR randomized clinical trial. JAMA
2014; 312: 799–808.
4 Tucker KL, Sheppard JP, Stevens R, et al. Self-
monitoring of blood pressure in hypertension:
a systematic review and individual patient data
meta-analysis. PLoS Med 2017; 14: e1002389.
Congratulations to Jon-David Schwalm
and colleagues1 of the HOPE 4 study
team for convincingly showing that
a non-physician health worker care
model reduces 10-year cardiovascular
disease risk in middle-income coun­
tries. However, the reduction in risk
attributable to the care model is almost
certainly larger than the –6·40% points
reported for the primary outcome for
two reasons.
First, a change in Framingham risk
score (FRS) is known to inaccurately
reflect longitudinal changes in cardio­
vascular disease risk over time.2 For
example, the FRS might accurately
predict that a 60-year-old man,
whose untreated systolic blood
pres­­sure (SBP) is 150 mm Hg, has a
10-year cardiovascular risk of 19·2%.
However, if he lowers his SBP with
medication to 140 mm Hg, the FRS
predicts a higher risk of 22·8%.3 This
counterintuitive increase conflicts
with evidence from randomised con­
trolled trials that each 10 mm Hg
reduction in SBP lowers cardiovascular
disease risk by approximately 20%.2,4
To approach this issue, the authors
should apply methods comparable to
those of the Million Hearts longitudinal
atherosclerotic cardiovascular disease
risk assessment tool, developed sub­
sequent to the design and inception
of HOPE 4, to estimate longitudinal
changes in cardiovascular risk.2 The
tool directly estimates relative risk
reductions for individuals using
observed changes in risk factors com­
bined with evidence from random­ised
310
systems might be theoretically optimal
risk estimates by predicted absolute risk
to calculate an absolute risk reduction.2
Second, the FRS is validated in
primary prevention populations; its
use is inappropriate among individuals
with existing myocardial infarction,
stroke, congestive heart failure, or atrial
fibrillation, who comprise 4·0%, 2·2%,
4·2%, and 1·2% of HOPE 4 participants,
respectively. Until appropriate absolute
risk equations are developed and
validated in these populations, risk
reductions can only be presented on a
relative scale.5
I declare no competing interests.
Adam Richards
adam.is.emailing@gmail.com
Community Partners International Research,
San Francisco, CA 94104, USA
1 Schwalm J-D, McCready T, Lopez-Jaramillo P,
et al. A community-based comprehensive
intervention to reduce cardiovascular risk in
hypertension (HOPE 4): a cluster-randomised
controlled trial. Lancet 2019; 394: 1231–42.
2 Lloyd-Jones DM, Huffman MD, Karmali KN, et al.
Estimating longitudinal risks and benefits
from cardiovascular preventive therapies
among Medicare patients: the Million Hearts
longitudinal ASCVD risk assessment tool:
a special report from the American Heart
Association and American College of
Cardiology. Circulation 2017; 135: e793–813.
3 D’Agostino RB, Sr, Vasan RS, Pencina MJ, et al.
General cardiovascular risk profile for use in
primary care: the Framingham Heart Study.
Circulation 2008; 117: 743–53.
4 Ettehad D, Emdin CA, Kiran A, et al.
Blood pressure lowering for prevention of
cardiovascular disease and death: a systematic
review and meta-analysis. Lancet 2016;
387: 957–67.
5 Richards A, Jackson NJ, Cheng EM, et al.
Derivation and application of a tool to
estimate benefits from multiple therapies
that reduce recurrent stroke risk. Stroke 2020;
published online March 23. https://doi.org/
10.1161/STROKEAHA.119.027160.
Authors’ reply
We thank Marleen Hendriks and col­
leagues and Adam Richards for their
comments regarding the HOPE 4 trial.1
Hendriks and colleagues draw attention
to the generalisability of our findings
as the non-physician health workers
were paid by the project and the
study medications were provided
to the intervention group free of
charge. Although working within the
infrastructure of existing health-care
for uptake and sustainability, most
health systems do not adequately
address cardiovascular disease control.
Access to health-care providers and
medications were considerable barriers
to cardiovascular disease prevention
and care in Colombia and Malaysia, as
in many countries around the world.
These barriers could not be ignored.
Study medications were provided
without charge to participants in the
intervention group when needed,
but use was not a requirement for
participation. Approximately half of
the intervention group took advantage
of the free medication, whereas the
rest obtained treatments available
locally. Both countries in the HOPE 4
trial provide subsidies for medications
through the public health-care sys­
tem; however, numerous barriers
to medication access still exist. The
comprehensive HOPE 4 intervention
mitigated these barriers to effectively
reduce cardiovascular disease risk.
Providing barrier-free access to essential
medications to those in need has
also been shown to be effective in a
contemporary randomised clinical trial.2
The non-physician health workers
in the HOPE 4 trial were compensated
through research funding. Although
this compensation is not a sustainable
means of support, this study empha­
sised the substantial impact a collab­
orative model of care can have on
cardiovascular disease risk. We hope
that health-care systems around
the world will adapt to include the
strategy that has been shown to be
effective in the HOPE 4 trial. While we
have shown that the HOPE 4 model
of care is feasible, we agree that an
economic evaluation is needed, and
this is currently underway.
Richards suggests inadequacies
of the Framingham risk score as a
marker of longitudinal changes in
cardiovascular disease risk. However,
given that our analysis is based on the
comparisons of longitudinal changes
between the intervention and control
groups, these concerns do not apply
www.thelancet.com Vol 396 August 1, 2020

to our randomised trial as any noise
or errors would be expected to be
similar in the two randomised groups.
Richards states that the Framingham
risk score is validated only in primary
prevention. Re-analysis of the data,
confined to the 90% of people who
do not have previous cardiovascular
disease or atrial fibrilliation, showed
that the intervention impact was very
similar and highly significant, with an
absolute difference in change between
intervention and control groups of
–4·6% (95% CI –7·1 to –2·1%, p=0·0003).
We declare no competing interests.
*Jon-David Schwalm, Tara McCready,
Shofiqul Islam, Martin McKee,
Salim Yusuf
schwalj@mcmaster.ca
Population Health Research Institute, McMaster
University and Hamilton Health Sciences, Hamilton,
ON L8L 2X2, Canada (J-DS, TM, SI, SY); Department
of Health Research Methods, Evidence and Impact,
McMaster University Faculty of Health Sciences,
Hamilton, ON, Canada (SY); and Department of
Health Services Research and Policy, London School
of Hygiene & Tropical Medicine, London, UK (MM)
1 Schwalm J-D, McCready T, Lopez-Jaramillo P, et al.
A community-based comprehensive
intervention to reduce cardiovascular risk in
hypertension (HOPE 4): a cluster-randomised
controlled trial. Lancet 2019; 394: 1231–42.
2 Persaud N, Bedard M, Boozary AS, et al.
Effect on treatment adherence of distributing
essential medicines at no charge—the CLEAN
Meds randomized clinical trial.
JAMA Intern Med 2020; 180: 27–34.
Individual patient risk
assessment and cost–
benefit analysis of
patiromer in AMBER
We commend Rajiv Agarwal and
colleagues 1 for completion of the
AMBER phase 2 trial. This is an
important contri­bution to a habitually
understudied patient population.
However, their results raise challenging
questions regarding both individual
patient risk assessment and future
cost–benefit analysis of patiromer use
in this setting.1
The addition of patiromer did
not improve the most meaningful
clinical endpoint, namely mean blood
www.thelancet.com Vol 396 August 1, 2020
pressure between groups. When
performing cost–benefit analysis of
patiromer in this setting, the only
remaining benefit is thus reducing the
risk of hyperkalaemia associated with
spironolactone.
The risk of adverse outcomes
follow­ i ng hyperkalaemia for any
individual patient with chronic kidney
disease is difficult to quantify, not
least because there is a suggestion
that mild hyperkalaemia might be
tolerated slightly better by different
patients with varying severities of
chronic kidney disease. 2,3 Further
complicating the extrapolation of
the risk of hyperkalaemia, the data
that do exist that describe the risk
of hyperkalaemia are observational
and describe population outcomes
following hyperkalaemia rather
than spironolactone induced hyper­
kalaemia specifically (which is a
reversible cause in a stable patient).
Although, in total, patiromer
resulted in five fewer patients with
hyper­k alaemia than did placebo
(14 in the placebo group vs nine in
the patiromer group), after baseline
there two more patients had a
serum potassium concentration
below 3·8 mmol/L (one patient in
the patiromer group had a the serum
potassium concentration below
3·5 mmol/L but greater than or equal
to 3·0 mmol/L) following patiromer
treatment compared with the
placebo group (six in the patiromer
group vs four in the placebo group).1
In the context of chronic kidney
disease, patients with low potassium
concentrations might have a higher
risk of adverse outcomes than those
with mild hyperkalaemia, and there
is a symmetrical U-shaped risk curve
for patients with varying potassium
concentrations.2,4
Pragmatically, an appropriate
risk–benefit ratio might be achieved
in selected patients who have a
propensity towards hyperkalaemia
after spironolactone use. A compelling
future study would involve selecting
patients specifically intolerant of
Correspondence
spironolactone to assess whether
patiromer allows the reintroduction of
the drug.
We declare no competing interests.
*Eoin D O’Sullivan, Iain M MacIntyre
eoin.osullivan@ed.ac.uk
Department of Renal Medicine, Royal Infirmary of
Edinburgh, Edinburgh EH16 4SA, UK
1 Agarwal R, Rossignol P, Romero A, et al.
Patiromer versus placebo to enable
spironolactone use in patients with resistant
hypertension and chronic kidney disease
(AMBER): a phase 2, randomised, double-
blind, placebo-controlled trial. Lancet 2019;
394: 1540–50.
2 Luo J, Brunelli SM, Jensen DE, Yang A.
Association between serum potassium
and outcomes in patients with reduced kidney
function. Clin J Am Soc Nephrol 2016;
11: 90–100.
3 Einhorn LM, Zhan M, Hsu VD, et al.
The frequency of hyperkalemia and its
significance in chronic kidney disease.
Arch Intern Med 2009; 169: 1156.
4 Korgaonkar S, Tilea A, Gillespie BW, et al.
Serum potassium and outcomes in CKD:
insights from the RRI-CKD cohort study.
Clin J Am Soc Nephrol 2010; 5: 762–69.
Authors’ reply
Eoin O’Sullivan and Iain MacIntyre
raise important questions about the
relevance of the AMBER study, relating
to individual-level decision making
and the cost–benefit of our approach.1
The primary objective of the
AMBER study was to assess the
ability of patiromer to enable the use
of spironolactone for 12 weeks.2 Of
the 148 patients randomly assigned
to spironolactone and placebo,
98 (66%) remained on spironolactone.
Of the 147 patients randomly
assigned to spironolactone and Science Photo Library/Getty Images
patiromer, 126 (84%) remained on
spironolactone. Of the 50 patients
stopping spironolactone in the
placebo group, 34 (68%) stopped
because of hyperkalaemia. Of the
21 patients stopping spironolactone
in the patiromer group, 10 (48%)
stopped because of hyperkalaemia, a
two-fold difference with compelling
statistical evidence of the superiority
of patiromer when considering time
to hyperkalaemia (p<0·0001). By
contrast, O’Sullivan and Macintyre
only refer to the 14 (placebo group)
versus nine (patiromer group) patients
311