Trial Designs

Rationale and design of a nurse-led

intervention to extend the HIV treatment
cascade for cardiovascular disease
prevention trial (EXTRA-CVD)
Nwora Lance Okeke, a Allison R. Webel, b Hayden B. Bosworth, a,c Angela Aifah, d Gerald S. Bloomfield, a
Emily W. Choi, e Sarah Gonzales, a,c Sarah Hale, a,c Corrilynn O. Hileman, f,g Virginie Lopez-Kidwell, h Charles
Muiruri, a Megan Oakes, a,c Julie Schexnayder, b Valerie Smith, a,c Rajesh Vedanthan, d and Chris T. Longenecker, f,i

Background Persons living with human immunodeficiency virus (PLHIV) are at increased risk of atherosclerotic
cardiovascular disease (ASCVD). In spite of this, uptake of evidence-based clinical interventions for ASCVD risk reduction in
the HIV clinic setting is sub-optimal.
Methods EXTRA-CVD is a 12-month randomized clinical effectiveness trial that will assess the efficacy of a multi-
component nurse-led intervention in reducing ASCVD risk among PLHIV. Three hundred high ASCVD risk PLHIV across three
sites will be randomized 1:1 to usual care with generic prevention education or the study intervention. The study intervention
will consist of four evidence-based components: (1) nurse-led care coordination, (2) nurse-managed medication protocols and
adherence support (3) home BP monitoring, and (4) electronic health records support tools. The primary outcome will be
change in systolic blood pressure and secondary outcome will be change in non-HDL cholesterol over the course of the
intervention. Tertiary outcomes will include change in the proportion of participants in the following extended cascade
categories: (1) appropriately diagnosed with hypertension and hyperlipidemia (2) appropriately managed; (3) at treatment
goal (systolic blood pressure b130 mm Hg and non-HDL cholesterol b National Lipid Association targets).
Conclusions The EXTRA-CVD trial will provide evidence appraising the potential impact of nurse-led interventions in
reducing ASCVD risk among PLHIV, an essential extension of the HIV care continuum beyond HIV viral suppression. (Am Heart
J 2019;216:xxx.)

People living with human immunodeficiency virus persist in the United States, once PLHIV are linked to care,
(PLHIV) can now expect to live near-normal lifespans if rates of viral suppression now exceed 80%. 2
they are diagnosed with the infection at an early stage, are For PLHIV who have achieved viral suppression on
prescribed effective combination antiretroviral therapy ART, providers have an important opportunity to focus
(ART), and achieve suppression of the HIV virus in the on preventing atherosclerotic cardiovascular disease
blood by being optimally adherent to their HIV medica- (ASCVD) and other non-AIDS comorbidities, which
tions. 1 Although gaps in this HIV treatment cascade occur at high rates among PLHIV despite viral suppres-
sion. 3,4 For ASCVD prevention, in particular, one might
From the aDuke University School of Medicine, Durham, NC, USA, bFrances Payne Bolton envision extending the treatment cascade for high blood
School of Nursing, Case Western Reserve University, Cleveland, OH, USA, cDurham VA pressure (BP) and high cholesterol, which account for
Medical Center, Durham, NC, USA, dNew York University School of Medicine, New York,
much of the population-level ASCVD risk in PLHIV, 5 as
NY, USA, eThe University of Texas at Dallas, Dallas, TX, USA, fCase Western Reserve
University School of Medicine, Cleveland, OH, USA, gMetroHealth Medical Center, follows: step 1, appropriate screening and diagnosis; step
Cleveland, OH, USA, hUniversity of North Texas, Denton, TX, USA, and iUniversity 2, appropriate treatment; and step 3, achievement of
Hospitals Harrington Heart & Vascular Institute, Cleveland, OH, USA. guideline-based treatment targets (Figure 1). Yet, BP and
RCT# NCT03643705
Funding: This work is supported in part by the National Institutes of Health (U01HL142099,
cholesterol are suboptimally treated in PLHIV, 6-8 possibly
K23HL137611, and K23HL123341). due to low perceived risk for ASCVD 9 or challenges in
Submitted March 1, 2019; accepted July 10, 2019. primary care coordination between HIV specialists and
Reprint requests: Dr Chris T. Longenecker, MD, Assistant Professor of Medicine, Case
non-HIV providers. 10 Non–physician-led approaches may
Western Reserve University School of Medicine, University Hospitals Harrington Heart and
Vascular Institute, 2103 Cornell Rd. WRB Room 4-533, Cleveland, OH 44106, USA.
address these barriers.
E-mail: [email protected] In this manuscript, we outline the rationale and design
0002-8703 of a mixed-methods implementation research trial to test
 American Heart Journal
2 Okeke et al Month Year

Figure 1
The extended HIV treatment cascade for ASCVD prevention.

a nurse-led intervention to extend the HIV treatment non-HIV chronic conditions in an increasingly medically
cascade for cardiovascular disease prevention (EXTRA- complex patient population, novel HIV clinic-based
CVD). The EXTRA-CVD trial is part of a consortium support initiatives are needed.
funded by the National Heart, Lung, and Blood Institute
under RFA-HL-18-007 “ImPlementation REsearCh to Perceived ASCVD risk
Develop interventions for People Living with HIV Persons living with HIV have 1.5-2 times higher risk of
(PRECLuDE).” The consortium includes other projects ASCVD compared to uninfected persons, 4,22 a risk that is
on primary ASCVD prevention (U01 HL142107), statin underestimated by current ASCVD risk calculators. 23,24
use (U01 HL142104), addressing trauma in CVD preven- Additionally, PLHIV personally underestimate their risk of
tion care (U01142109), and chronic obstructive pulmo- ASCVD. 9 Misperceptions of risk, whether patient initiat-
nary disease care (U01 HL142103). Implementation ed or driven by faulty risk assessment tools, directly result
strategies developed by this consortium may be applica- in a lack of communication between PLHIV and their
ble to other chronic disease populations (eg, rheumatoid providers about how to manage modifiable ASCVD risk
arthritis, diabetes, and chronic kidney disease) that share factors. For example, a survey of PLHIV from the United
similar barriers to effective ASCVD prevention care. States and Canada revealed that only 8% of PLHIV have
discussed heart disease with their HIV provider despite
reported hypertension and hyperlipidemia rates of 32%
Rationale and 40%, respectively. 25
The changing HIV workforce and the patient-provider
experience for PLHIV BP and cholesterol targets matter
The population of PLHIV in the United States is Because absolute risk reduction depends on absolute
increasing by approximately 30,000 persons per year, baseline risk, recent guidelines from the American Heart
but growth in the HIV provider workforce is not keeping Association (AHA) and American College of Cardiology
pace, 11 and the Institute of Medicine and the HIV (ACC) recommend pharmacologic therapy thresholds
Medical Association have warned of an impending HIV and treatment targets for BP 26 and cholesterol 27 that are
provider shortage. 12,13 These trends are particularly appropriately tailored to patients' risk. For hypertension,
worrisome in light of high rates of dissatisfaction among initiation of pharmacologic therapy is recommended at a
HIV providers 11 and an aging PLHIV population 14 that is threshold of 130/80 for those with N10% ASCVD risk or
experiencing rising rates of non–AIDS-related chronic high-risk comorbidity, although HIV is not specifically
diseases. 15,16 These epidemiologic shifts may exacerbate mentioned as a high-risk condition. On the other hand,
provider stressors and dissatisfaction as HIV specialists the 2018 Cholesterol Clinical Practice Guidelines 27 do
have expressed their discomfort and lack of support in specifically mention HIV as a risk-enhancing condition
managing non–HIV-related conditions. 11,17-20 Converse- when considering statin therapy, and the National Lipid
ly, non-HIV primary care providers also feel inadequately Association (NLA) has recognized PLHIV as a special high-
trained to manage chronic HIV infection. 21 Although risk population 28 for whom a non–high-density lipopro-
assigning care unrelated to HIV to primary care providers tein (HDL) treatment target of b130 mg/dL (3.36 mmol/L)
may appear to be an easy solution, PLHIV dislike the care is reasonable when at least 1 other major risk factor such
fragmentation that occurs with having multiple longitu- as hypertension is present.
dinal providers. 10 To alleviate both patient and provider People living with HIV are conditioned to care about
stress in navigating the intersection between HIV and their “numbers” and have surprisingly accurate
American Heart Journal
Volume 216, Number 0
Okeke et al 3

Table I. Overall clinic demographics and estimated numbers of potentially eligible PLHIV engaged in care at the 3 academic HIV-specialty clinic
sites selected for this study
Total patients Age (IQR) % Female % Black % Hispanic HIV viral load b200 High BP⁎ High cholesterol⁎ Both⁎

MetroHealth
1759 47 (35-55) 24% 50% 13% 1500 (85%) 491 501 286
(Cleveland, OH)
Duke Health (Durham, NC) 1890 50 (40-58) 28% 59% 4% 1349 (71%) 605 397 291
UH (Cleveland, OH) 1101 51 (40-58) 23% 64% 4% 975 (89%) 550 485 334

⁎ For the purposes of estimating eligible subjects, this feasibility analysis used billing codes, chart diagnosis, OR on antihypertensive or cholesterol medication. The numbers for
hypertension and hypercholesterolemia reflect ONLY HIV patients with HIV viral load b200 copies/mL.

knowledge of their viral load and CD4+ T-cell count. 29 extended treatment cascade to identify patients who
Presenting patients with cholesterol and BP treatment merit more clinical attention and ease medication
targets may thus resonate for PLHIV if the implementa- algorithm use through decision support tools embedded
tion of guideline-based therapy is streamlined. These in the EHR platform. 39
principles guided the development of our proposed We believe that if proven effective for ASCVD risk
intervention described below. factor control, our nurse-led intervention may be scaled
up to address a broad range of preventive care services
A multipronged approach is required to address for PLHIV, thus increasing its population impact. Our
barriers to ASCVD prevention care model may be especially relevant in the context of a
The use of nonphysician providers is expanding in the changing HIV specialty workforce that will increasingly
United States, a trend that is also true in HIV-specialist rely on nonphysician providers and increased coordina-
care. 11,30 The quality of HIV care provided by these tion with non-HIV primary care providers and specialists.
nonphysician specialists is comparable to physician
specialists, 31 but the quality of and comfort level with
ASCVD preventive care are poorly understood. Our Design
experience in other US populations suggests that nurse- Overall design
led management of cardiovascular risk factors is highly The overall aims of the EXTRA-CVD study are to (1)
effective. 32-35 For example, a meta-analysis of nurse- and conduct a formative assessment of ASCVD preventive
pharmacist-led cholesterol medication adherence inter- care and perceptions of ASCVD risk in the HIV specialty
ventions showed substantial improvements in adherence clinic environment, (2) evaluate the 12-month efficacy of
and 15- to 20-mg/dL (0.39-0.52 mmol/L) reductions in a prevention nurse-led intervention to improve BP and
total cholesterol. 36 Based on this premise and an lipid control on PLHIV, and (3) conduct a process
understanding of the complexities of primary preventa- evaluation of the prevention nurse-led intervention.
tive ASCVD care in PLHIV, we have designed a The EXTRA-CVD study uses a mixed-methods clinical
multicomponent intervention led by an EXTRA-CVD effectiveness trial design. We will first conduct a baseline
prevention nurse consisting of (1) home BP monitoring, assessment of ASCVD prevention care in HIV clinics. These
(2) care coordination, (3) nurse-managed medication baseline data will then inform an intervention design team
protocols and adherence counseling, and (4) electronic who will use principles of human-centered design to adapt
health record tools. our intervention to the local context. The effectiveness of
Home BP measurements have greater predictive power the intervention will be tested in a participant-level
for mortality as compared to office-based measure- randomized controlled trial. Finally, an extensive process
ments, 37 and home BP monitoring is a class I recommen- evaluation will assess implementation of the intervention,
dation in the 2017 ACC/AHA guidelines. 26 Furthermore, including how the intervention alters trust and communica-
algorithm-based care to reduce practice variation and tion ties between PLHIV and their providers.
clinical inertia has long been recommended to ensure The protocol is institutional review board (IRB)
that patients are not “stuck” at subtherapeutic doses of approved at University Hospitals (UH) Cleveland Medical
medications. 38 By using algorithms and clear decision Center (Protocol #03-18-16), with reliant review at all
rules to guide medication titration based on home BP participating sites in accordance with the National
measurements, the prevention nurse will make recom- Institutes of Health single IRB policy (Duke IRB Protocol
mendations to providers to improve care by reducing #00092437; MetroHealth IRB Protocol #00000685). 40
clinical inertia, reducing variation, and allowing nonphy- Subjects throughout all phases of the study will sign
sician staff members to assist in care. Finally, electronic written informed consent, except for telephone inter-
health records (EHRs) can generate reports on the views for which they will give verbal consent according
 American Heart Journal
4 Okeke et al Month Year

Table II. Full inclusion and exclusion criteria for EXTRA-CVD trial participants
Inclusion criteria Exclusion criteria

1. Age ≥ 18 y 1. On lipid-lowering medication solely for secondary prevention of ASCVD

events with evidence of premedication non-HDL which was already below
100 mg/dL (2.59 mmol/L)
2. On antihypertensive medications solely for a nonhypertension indication
(eg, systolic heart failure)
2. Confirmed HIV+ diagnosis (HIV+ ELISA with confirmatory PCR) 3. Severely hearing or speech impaired, or other disability that would limit
3. Undetectable HIV viral load: defined as the most recent HIV viral load participation in the intervention components
b200 copies/mL, checked within the past year (assessed via medical 4. In a nursing home and/or receiving in-patient psychiatric care
record abstraction) 5. Terminal illness with life expectancy b4 m
4. Hypertension: defined as systolic BP N130 mm Hg on ≥2 occasions in the 6. No reliable access to a telephone
past 12 m or on an antihypertensive medication (assessed via medical 7. Pregnant, breast-feeding, or planning a pregnancy during the study
record abstraction) and period
5. Hyperlipidemia: defined as a non-HDL cholesterol N130 mg/dL 8. Planning to move out of the area in the next 12 m
(3.36 mmol/L) or on cholesterol-lowering medication 9. Non-English speaking

ELISA, Enzyme-linked immunosorbent assay; PCR, polymerase chain reaction.

to an IRB-approved script. The study is registered at ASCVD risk reduction medications, and perceptions of
clinicaltrials.gov (NCT03643705). nonpharmacologic ASCVD risk reduction modalities.
For HIV health care workers, a preinterview survey will
Setting collect information on demographics and practice
EXTRA-CVD will be conducted at 3 Ryan White environment. All health care workers will then be
Program federally funded academic medical centers that interviewed about general perceptions of ASCVD risk in
provide HIV specialty care for a diverse patient panel PLHIV and ideas for interventions that would improve
representative of US HIV+ population (Table I). More ASCVD risk reduction in their setting. For HIV specialty
than 20% of patients at the UH Cleveland and Duke clinics providers, a segment of the interview will focus on
reside in rural counties. Less than 1% of Cleveland area prescribing patterns for BP and cholesterol medication.
patients receive outpatient HIV care at both MetroHealth The interviews will be coded and analyzed using standard
and UH Cleveland clinics in a given calendar year. qualitative research methodology, and a summative
report will be prepared to aid with the intervention
Formative assessment adaptation phase of the study.
For the formative assessment, we will enroll approxi-
mately 60 PLHIV (20 per site) with hypertension and A human-centered design approach to intervention
hyperlipidemia along with 36 health care workers (12 per adaptation
site) including HIV care providers, primary care pro- Human-centered design is pillared by a focus on
viders, nurses, and other support staff. interventions that directly meet the needs of targeted
PLHIV will complete a 1-time survey and focus group stakeholders and by incorporating the input of stake-
discussion or key informant interview conducted by an holders in every step of the design process in a systematic
experienced study investigator (A. R. W., J. S., or S. G.). and iterative manner. 42 The framework is often divided
Telephone interviews will be offered to eligible partici- into 5 phases: (1) empathizing with stakeholders, (2)
pants who cannot make it to the clinic site. A defining the problem, (3) conceptualizing the problem in
preinterview questionnaire will include demographics, an inclusive manner, (4) prototyping the intervention,
HIV and medical history, adherence to ASCVD-related and (5) testing the intervention. 42,43 The intervention
medications, and perceptions of CVD risk. We will assess adaptation phase of the EXTRA-CVD study will adapt our
perceived cardiovascular risk using the Health Beliefs for proposed intervention to the local context by integrating
Cardiovascular Disease Scale, a 25-item Likert scale–based feedback from PLHIV and health care team members into
questionnaire developed by Tovar et al, 41 to assess 4 the intervention design process, with a particular focus
separate constructs of the Health Belief Model (perceived on phases 3-5 of the aforementioned framework.
susceptibility, perceived severity, perceived barriers, per- Two design consultation teams will be assembled: a
ceived benefits) as they pertain to ASCVD risk perception. combined Cleveland site team (10-12 individuals) with
The 20- to 25-minute individual interview or focus group representation from UH and MetroHealth and a team at
discussion will cover perceptions of CVD risk, perceived Duke (6-8 individuals). Sessions will be facilitated by team
ASCVD risk associated with ART, barriers to adherence to members with training in human-centered design princi-
ples (A. A., J. S., and L. O.). Design team members will
American Heart Journal
Volume 216, Number 0
Okeke et al 5

Figure 2
EXTRA-CVD trial design.

include a representative sample of providers (HIV with other inclusion and exclusion criteria. To simplify
specialists, primary care providers, cardiologists), clinic the enrollment process, we have set uniform cut points
support staff (nurses, pharmacists, social workers), and for the definitions of high BP (systolic BP N130 mm Hg)
HIV clinic patients. The teams will meet for 3 initial and high cholesterol (non-HDL cholesterol N130 mg/dL or
sessions: (1) brainstorming, (2) conceptualization, and 3.36 mmol/L). However, individual treatment targets will
(3) creation. These will be followed by 2 iteration be defined by the site prevention nurse through consultation
meetings which will be informed by acceptability and with patient and providers based on 10-year ASCVD risk,
feasibility testing conducted during a 6-week pilot trial of comorbidities, and risks of adverse treatment effects.
the intervention. We will use the electronic medical records at the 3 sites
to identify potential subjects. Potential subjects will
Clinical trial initially be mailed a recruitment letter signed by his or
After thoroughly adapting the intervention to the local her primary HIV provider and will have the opportunity
context, we will test our intervention in a randomized to opt out of the study by calling a toll-free number. A
clinical trial. research assistant will contact all subjects who do not opt
Subjects. We will enroll 300 subjects randomized 1:1 out. Following a telephone script, the research assistant
to the prevention nurse intervention or education will describe the study in detail, ensure the patient is
control. Randomization will be stratified by site with eligible and willing to participate, and schedule a baseline
goal enrollment distributed equally (n = 100) across sites. study visit at the next clinical visit with an HIV provider
All subjects will have suppressed HIV-1 viral load on where they will be enrolled following confirmation of
antiretroviral therapy and will have a diagnosis of both entry criteria and written informed consent.
high BP and high cholesterol as defined in Table II along
 American Heart Journal
6 Okeke et al Month Year

Figure 3
Hypothetical examples of variation in dose or exposure to the EXTRA-CVD intervention. A, Participant with lower intensity requirements. B,
Participant with higher intensity requirement. C, Control participant. Squares represent in-person visits and lines are telephone contact.

Treatment assignment. Figure 2 describes the overall recognizing that wrist monitors are less accurate and
trial design. The intervention group will receive the less precise. 44 Participants will receive training about
locally adapted multicomponent nurse-led intervention, how to use the device at the enrollment visit, and these
and the education control group will receive generic non- principles will be reinforced at each contact with the
AIDS comorbidity prevention education only. This active prevention nurse. Participants will be expected to take
comparator is appropriate because participants have their BP every day prior to taking morning medications.
multiple risk factors for ASCVD and other non-AIDS At each telephone or in-person follow-up visit, the
comorbidities. The generic prevention educational mod- prevention nurse will request BP values for the past 2
ules will be delivered to all subjects at 4 in-person visits weeks. Participants with poor BP control (determined by
(enrollment and at 4, 8, and 12 months) and will consist home BP readings) will receive calls every 2 weeks, with
of evidence-based material on diet, exercise, smoking, study algorithm-based management changes (Figure 3).
sexually transmitted infections, and cancer prevention. At each visit (in-person or telephone) where recent
Intervention. Intervention subjects will undergo an average weekly home BP values exceed 130/80 mm Hg,
initial risk assessment based on AHA materials including the prevention nurse will review the patient's medica-
10-year ASCVD risk scoring. The prevention nurse will tions, including recent changes in medication and
conduct a baseline medication assessment, including potential adverse effects. 45 The prevention nurse will
participant's knowledge of the purpose and adverse also provide medication adherence and adverse effect
effects of each BP or cholesterol medication and current mitigation counseling to participants. Patients will also
or potential adherence strategies. receive a personalized medication schedule to enhance
Beginning with initial enrollment, the prevention nurse medication adherence. The prevention nurse will decide
will coordinate BP and cholesterol management for all on recommendations for medication change based on
participants in the intervention arm. Care coordination study algorithms (Supplemental Figures 1 and 2) and will
will consist of tailored discussions with the participant approach the designated responsible provider (HIV or
and his/her providers about which provider will take primary care provider) for prescriptions and laboratory
primary responsibility for BP and cholesterol manage- orders. The provider will ultimately decide on final
ment, informed by patient preference and provider management decisions and may request to have the
comfort with CVD risk management. The prevention participant be taken OFF management protocols as
nurse will direct subsequent management decisions to clinically indicated (eg, recent ASCVD events or advanced
the designated provider but will facilitate communication CKD), in which case the participant would continue all
by notifying the nondesignated provider of any changes other components of the EXTRA-CVD intervention.
to the treatment plan. For BP, we will use an algorithm adapted from Kaiser
Intervention participants will receive the Omron 7- Permanente 46 and used in our prior studies. Once-daily
series upper arm monitor (Omron Healthcare, Lake medication and combination therapy will be recom-
Forest, IL). A variety of cuff sizes are available; however, mended when possible. A follow-up basic chemistry
very obese subjects may require a wrist monitor, panel will be ordered when adding angiotensin-
American Heart Journal
Volume 216, Number 0
Okeke et al 7

Table III. Sample size estimates to detect a range of plausible and clinically significant effect sizes for the primary and secondary outcomes of the
EXTRA-CVD trial
BP effect size Non-HDL effect size

5 mm Hg 6 mm Hg 7 mm Hg 10 mg/dL (0.26 mmol/L) 15 mg/dL (0.39 mmol/L) 20 mg/dL (0.52 mmol/L)

70% Power 278 190 140 248 110 64

80% Power 350 234 178 310 148 80
90% Power 466 340 232 424 184 104

Values in italics represent sample sizes that are less than our proposed sample size (n = 300).

converting enzyme/angiotensin receptor blocker, thia- nurse which will appear on his/her dashboard or as a
zide diuretic, or potassium-sparing diuretic. Medication recurring pdf report. During the intervention phase, the
uptitrations will be recommended at intervals of 2- prevention nurse will have regular access to this graphic
4 weeks until control is achieved. Measures not shown and will receive names of specific patients who have
in Supplemental Figure 1 will include but will not be fallen out of each cascade category. Additional ideas for
limited to (1) adding agents such as hydralazine, EHR tools to be refined during the intervention adapta-
terazosin, and clonidine; (2) considerations for comorbid tion include decision support tools to identify recom-
kidney disease or prior ASCVD event; and (3) avoiding mended BP and cholesterol medications based on the
combination use of heart rate–slowing drugs. algorithms described above.
For cholesterol, we will use an algorithm (Supplemen- Analyses and outcomes. The primary outcome will
tal Figure 2) that reconciles the NLA guidelines for HIV- be systolic BP at 12 months and secondary outcome will
infected patients 28 with recent ACC/AHA cholesterol be non-HDL cholesterol at 12 months, both measured at 4
practice guidelines. 27 As a first step, the prevention nurse time points (0, 4, 8, and 12 months). All BPs used for
will determine non-HDL target for each individual outcomes will be obtained by a blinded research
participant based on the guidelines. For most participants assistant, and cholesterol levels will be measured by
in the trial, the target non-HDL will be b130 mg/dL laboratory personnel who are also blinded to treatment
(3.36 mmol/L); however, high-risk patients (such as group. Linear mixed-effects models will be used to
those with history of prior ASCVD event) will have a examine differences in systolic BP and non-HDL choles-
more aggressive goal (b100 mg/dL or 2.59 mmol/L). Our terol over time between the study arms. Linear mixed-
algorithm will address drug-drug interactions with ART, effects models will allow us to implicitly account for the
particularly the safe use of higher-potency statins such as correlation between a patient's repeated measurements
rosuvastatin and atorvastatin when interactions are over time. The tertiary outcome will be change in the
present. In accordance with recent ACC/AHA guidelines, proportion of subjects falling into each extended cascade
the algorithm will include consideration of combination category ([1] % appropriately diagnosed, [2] % appropri-
therapy with ezetimibe and proprotein convertase ately managed, and [3] % at treatment goal). For this
subtilisin/kexin type 9 inhibitors as appropriate, which analysis, we will calculate an ordinal 4-level variable at
may require referral to a specialist. Lipid profiles will be baseline and at 4, 8, and 12 months. We will use a
checked at every in-person study visit. The prevention proportional odds model fit via generalized estimating
nurse will have access to all cholesterol fractions, but the equations to examine differences over time between
algorithm will focus on non-HDL as the primary target. study arms. The proportional odds assumption will be
Barriers to appropriate statin use in the general population assessed using score tests, and the model will be relaxed
and among PLHIV are well documented and include statin- to partial proportional odds if necessary. All analyses will
associated muscle symptoms. 27,47 The prevention nurse will be conducted following an intention-to-treat principle.
call 2 weeks after statin initiation to discuss adherence and Although we are underpowered to assess the impact of
possible adverse effects. The prevention nurse will use an the intervention on all-cause mortality, cardiac-associated
evidence-based approach to evaluation and management of mortality, and major adverse cardiovascular events
statin-associated muscle symptoms as recommended by NLA (myocardial infarction, stroke, coronary revasculariza-
guidelines (Supplemental Figure 3). 48,49 This approach will tion, peripheral vascular diagnosis and/or intervention),
include evaluation for other causes, drug-drug interactions, we will report on all of these outcomes as dictated by the
checking creatinine kinase levels, trial off statin, retrial of study protocol.
different statin, nondaily dosing of longer-acting statin (ie, Power. Simulation-derived power estimates for the
rosuvastatin), and/or referral to a specialist. primary outcome were generated based on the following
EHR tools to help the prevention nurse will include an assumptions which are based on preliminary data from
extended treatment cascade graphic for the prevention our sites: mean baseline systolic BP of 145 mm Hg,
 American Heart Journal
8 Okeke et al Month Year

reduction in control group of 1 mm Hg by 12 months, primary network analysis, we will assess the intervention
15% dropout, SD of 17, and a within-individual correla- effect on subjects' trust and communication ties. If the
tion of 0.4 among repeated systolic BP measurements. intervention has an effect on both clinical outcomes and
Similarly, for non-HDL, we assumed a baseline value of trust/communication ties, we will examine whether
132 mg/dL (3.41 mmol/L) with an SD of 41 and a within- changes in these ties act as mediators of the overall
individual correlation of 0.7. After generating 1,000 intervention effect. Anonymized data collection and
simulated data +sets under these assumptions, we fit analysis for this portion of the process evaluation will
linear mixed models to each and assessed the effect of be overseen by experts in social network analysis (V. K.
interest using 2-sided tests with a type I error rate of 0.05. and E. C.).
Power calculations for a range of BP and non-HDL effects
are shown in Table III. Based on these results, we will Funding and author responsibilities
have N80% power to detect a 6–mm Hg lower systolic BP.
This work is supported in part by the National Institutes of
The target difference of 6 mm Hg was based on data from
Health (U01HL142099, K23HL137611, and K23HL123341).
similar BP reduction trials with nurse/pharmacist-based
The authors are solely responsible for the design and
telephone interventions coupled with home BP monitor-
conduct of this study, all study analyses, the drafting and
ing. 33,50,51 Based on our sample size calculations, we will
editing of the paper, and its final contents.
also have N90% power to detect a 15-mg/dL (0.39 mmol/
L) lower non-HDL cholesterol in the intervention arm
versus education control. This target was based on the Discussion
data from 2 meta-analyses of nurse-based interventions to Clinical outcomes for PLHIV have improved dramati-
reduce cardiovascular disease risk. 36,52 A 6–mm Hg cally over the last 30 years thanks to substantial
improvement in systolic BP is associated with a ~20% government and private industry investments in antire-
decrease in ASCVD events, 53 and a 15-mg/dL (0.39 mmol/ troviral drug discovery and HIV care infrastructure (eg,
L) improvement in cholesterol is associated with ~10% the Ryan White Act 57 ). Going forward, continued
decrease in clinical ASCVD events. 54 improvements in clinical outcomes for PLHIV will
depend in large part on strategies to target non-AIDS
Process evaluation comorbidity such as cardiovascular disease. As part of the
An extensive process evaluation of the intervention will National Heart, Lung, and Blood Institute's PRECLUDE
be conducted. We will evaluate key implementation consortium, EXTRA-CVD aims to contribute to the
process measures across the following domains: fidelity development of an evidence base for interventions that
(quality), dose delivered (completeness), dose received can be scaled across the United States and beyond.
(exposure and satisfaction), recruitment, reach (partici- Scalability and sustainability are 2 critical domains of
pation rate), and context with both PLHIV and health implementation science that ultimately define the reach
care team participants. 55,56 (and thus population impact) of an effective intervention.
We will quantify the dose of the intervention received We have conceived of EXTRA-CVD and aim to iteratively
with a variety of measures including (1) number and refine our intervention with these principles in mind. For
duration of telephone calls from the prevention nurse to example, the treatment cascade model is already in use
patient subjects and to providers or providers' staff; (2) nationwide in HIV specialty care clinics and is a key
number and nature of provider communication notes in evaluation framework for the federally funded Ryan
the EHR; (3) frequency of BP and cholesterol algorithm White program. Ryan White currently provides extra
use; (4) number of referrals to BP or cholesterol funding to HIV clinics that prove they can effectively
specialists; and (5) frequency of use of EHR tools. We implement evidence-based strategies to retain PLHIV in
will assess fidelity through the use of prevention nurse care and improve rates of HIV viral suppression. We
observation and checklists. After study completion, we believe that, if proven effective, services provided by
will conduct exit focus groups with study participants prevention nurses as envisioned in EXTRA-CVD might be
and with a sample of health care providers. supported by Ryan White funding. Already, measures
Finally, we will conduct a network analysis of subjects' such as rates of cholesterol screening are used by Ryan
trust and communication ties with their HIV provider, White to evaluate programs; however, more is clearly
HIV nurse, primary care provider (if they have one), and needed to improve ASCVD outcomes. In addition,
any non-HIV specialty care providers. At each in-person prevention nurses may be used to improve preventive
visit, all subjects in the intervention and control group care of other non-AIDS comorbidities such as cancer, for
will be asked to complete surveys of their trust/ which PLHIV are at increased risk. 58
communication ties with individual providers using Several limitations of the EXTRA-CVD study deserve
validated instruments. The prevention nurses will also mention. Although numerous reports have described the
be asked to assess the extent of their trust and increased prevalence and cardiovascular risk of tobacco
communication ties with individual providers. For the use in persons living with HIV, 59,60 we were unable to
American Heart Journal
Volume 216, Number 0
Okeke et al 9

incorporate smoking cessation as a primary outcome for 2. Selected National HIV. Prevention and care outcomes in the United
the trial due primarily because of 2 key considerations. States. https://www.cdc.gov/hiv/pdf/library/factsheets/cdc-hiv-
First, smoking cessation is difficult, and interventions national-hiv-care-outcomes.pdf 2018. Accessed December 26,
2018.
have had limited effectiveness, especially for PLHIV. 61,62
3. Freiberg MS, Chang CC, Kuller LH, et al. HIV infection and the risk of
For example, a recent Cochrane review reported that the
acute myocardial infarction. JAMA Intern Med 2013:1-9.
absolute success rates of motivational interviewing– 4. Longenecker CT, Triant VA. Initiation of antiretroviral therapy at high
based smoking cessation interventions (modality best CD4 cell counts: does it reduce the risk of cardiovascular disease?
suited for the EXTRA-CVD protocol) above usual care is Curr Opin HIV AIDS 2014;9(1):54-62.
approximately 3%. 61 Our study could not feasibly be 5. Althoff K, Palella F, Gebo K, et al. Impact of smoking, hypertension &
powered to detect such a difference. Second, 2 of the 3 cholesterol on myocardial infarction in HIV+ adults. Conference on
clinic sites already have existing smoking cessation Retroviruses and Opportunistic Infections. Seattle. USA: WA; 2017.
programs associated with their HIV clinics. At both the 6. Clement ME, Park LP, Navar AM, et al. Statin utilization and
recommendations among HIV- and HCV-infected veterans: a cohort
Duke and UH sites, all HIV-positive smokers are readily
study. Clin Infect Dis 2016;63(3):407-13.
referred to a smoking cessation clinic with access to 7. Al-Kindi SG, Zidar DA, McComsey GA, et al. Gender differences in
smoking cessation specialist who have expertise in statin prescription rate among patients living with HIV and hepatitis C
evidence-based cessation strategies. The availability of virus. Clin Infect Dis 2016;63(7):993-4.
such programs to prospective participants raised basic 8. Myerson M, Poltavskiy E, Armstrong EJ, et al. Prevalence, treatment,
questions about clinical equipoise and the ethics of and control of dyslipidemia and hypertension in 4278 HIV
randomization. Therefore, we will offer smoking cessa- outpatients. J Acquir Immune Defic Syndr 2014;66(4):370-7.
tion services to all participants in the trial and secondarily 9. Cioe PA, Crawford SL, Stein MD. Cardiovascular risk-factor
track and report tobacco use–associated outcomes. knowledge and risk perception among HIV-infected adults. J Assoc
Nurses AIDS Care 2014;25(1):60-9.
We will be unable to assess long-term maintenance
10. Cheng QJ, Engelage EM, Grogan TR, et al. Who provides primary
effects of our intervention within the time frame of this care? An assessment of HIV patient and provider practices and
study but will seek to evaluate maintenance effects using preferences. J AIDS Clin Res 2014;5(11).
medical records review and follow-up interviews with 11. Weiser J, Beer L, West BT, et al. Qualifications, demographics,
participants in the future. A cost-effectiveness analysis is satisfaction, and future capacity of the HIV care provider workforce in
also beyond the scope of our current project; however, the United States, 2013-2014. Clin Infect Dis 2016;63(7):966-75.
we will collect cost data to inform the design of such 12. Association HM. Averting a crisis in hiv care: a joint statement of the
analyses in the future. Finally, our assessment of American Academy of HIV Medicine (AAHIVM) and the HIV
Medicine Association on the HIV Medical Workforce, 2009. 2009.
medication adherence using a validated self-report
13. Institute of Medicine. HIV screening and access to health care: health
questionnaire is not as accurate as pill counting or
care system capacity for increased HIV testing and provision of care
pharmacy data; however, these more sophisticated 2011.
measures would not be feasible for nurses who would 14. Centers for Disease Control and Prevention. HIV among people aged
be hired to fill this role in a real-world context. 50 and older 2018; https://www.cdc.gov/hiv/group/age/
In conclusion, barriers such as low perceived risk for olderamericans/index.html. Accessed December 11 2018.
ASCVD or challenges in primary care coordination 15. Berry SA, Fleishman JA, Moore RD, et al. Trends in reasons for
between HIV specialists and non-HIV providers may hospitalization in a multisite United States cohort of persons living with
explain suboptimal ASCVD risk management for PLHIV. HIV, 2001-2008. J Acquir Immune Defic Syndr 2012;59(4):368-75.
16. Smith CJ, Ryom L, Weber R, et al. Trends in underlying causes of
To address these barriers, we have designed a multicom-
death in people with HIV from 1999 to 2011 (D:A:D): a multicohort
ponent nurse-led intervention that will be further adapted
collaboration. Lancet 2014;384(9939):241-8.
to the local HIV specialty clinic context. If proven 17. Okeke NL, Chin T, Clement M, et al. Coronary artery disease risk
effective to reduce both BP and cholesterol as postulated, reduction in HIV-infected persons: a comparative analysis. AIDS Care
EXTRA-CVD will have substantial clinical impact among 2016;28(4):475-82.
high-risk PLHIV, potentially reducing ASCVD events by 18. Burkholder GA, Tamhane AR, Salinas JL, et al. Underutilization of
more than a quarter. 53,54 aspirin for primary prevention of cardiovascular disease among HIV-
infected patients. Clin Infect Dis 2012;55(11):1550-7.
19. Fultz SL, Goulet JL, Weissman S, et al. Differences between infectious
Appendix A. Supplementary data diseases-certified physicians and general medicine-certified physi-
Supplementary data to this article can be found online cians in the level of comfort with providing primary care to patients.
at https://doi.org/10.1016/j.ahj.2019.07.005. Clin Infect Dis 2005;41(5):738-43.
20. Lichtenstein KA, Armon C, Buchacz K, et al. Provider compliance with
guidelines for management of cardiovascular risk in HIV-infected
patients. Prev Chr Dis 2013;10, E10.
References 21. Barnes R, Koester KA, Waldura JF. Attitudes about providing HIV
1. Nakagawa F, Lodwick RK, Smith CJ, et al. Projected life expectancy of care: voices from publicly funded clinics in California. Fam Pract
people with HIV according to timing of diagnosis. AIDS 2012;26(3): 2014;31(6):714-22.
335-43.
 American Heart Journal
10 Okeke et al Month Year

22. Freiberg MS, Chang CC, Kuller LH, et al. HIV infection and the risk of education: a cluster randomized trial. Ann Intern Med 2006;145(3):
acute myocardial infarction. JAMA Intern Med 2013;173(8):614-22. 165-75.
23. Feinstein MJ, Nance RM, Drozd DR, et al. Assessing and refining 39. Moja L, Kwag KH, Lytras T, et al. Effectiveness of computerized
myocardial infarction risk estimation among patients with human decision support systems linked to electronic health records: a
immunodeficiency virus: a study by the Centers for AIDS Research systematic review and meta-analysis. Am J Public Health 2014;104
Network of Integrated Clinical Systems. JAMA Cardiol 2017;2(2): (12):e12-22.
155-62. 40. Final NIH Policy on the use of a single institutional review board for
24. Thompson-Paul AM, Lichtenstein KA, Armon C, et al. Cardiovascular multi-site research. https://grants.nih.gov/grants/guide/notice-
disease risk prediction in the HIV outpatient study. Clin Infect Dis files/NOT-OD-16-094.html. Accessed May 31, 2019.
2016;63(11):1508-16. 41. Tovar EG, Rayens MK, Clark M, et al. Development and
25. Sherer R, Solomon S, Schechter M, et al. HIV provider-patient psychometric testing of the Health Beliefs Related to Cardiovascular
communication regarding cardiovascular risk: results from the AIDS Disease Scale: preliminary findings. J Adv Nurs 2010;66(12):
Treatment for Life International Survey. J Int Assoc Provid AIDS Care 2772-84.
2014;13(4):342-5. 42. IDEO. The field guide to human-centered design: design kit IDEO.
26. Whelton PK, Carey RM, Aronow WS, et al. ACC/AHA/AAPA/ABC/ 2015.
ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the 43. Matheson GO, Pacione C, Shultz RK, et al. Leveraging human-
prevention, detection, evaluation, and management of high blood centered design in chronic disease prevention. Am J Prev Med
pressure in adults: a report of the American College of Cardiology/ 2015;48(4):472-9.
American Heart Association Task Force on Clinical Practice 44. Irving G, Holden J, Stevens R, et al. Which cuff should I use? Indirect
Guidelines. Hypertension 2017;2017. blood pressure measurement for the diagnosis of hypertension in
27. Grundy SM, Stone NJ, Bailey AL, et al. AHA/ACC/AACVPR/AAPA/ patients with obesity: a diagnostic accuracy review. BMJ Open
ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNa guideline on the 2016;6(11), e012429.
management of blood cholesterol: a report of the American College of 45. Mueller SK, Kripalani S, Stein J, et al. A toolkit to disseminate best
Cardiology/American Heart Association Task Force on Clinical practices in inpatient medication reconciliation: multi-center medica-
Practice Guidelines. J Am Coll Cardiol 2018;2018. tion reconciliation quality improvement study (MARQUIS). Jt Comm J
28. Jacobson TA, Maki KC, Orringer CE, et al. National Lipid Association Qual Patient Saf 2013;39(8):371-82.
recommendations for patient-centered management of dyslipidemia: 46. Jaffe MG, Lee GA, Young JD, et al. Improved blood pressure control
part 2. J Clin Lipidol 2015;9(6 Suppl):S1-S122. e121. associated with a large-scale hypertension program. JAMA
29. Sewell J, Daskalopoulou M, Nakagawa F, et al. Accuracy of self- 2013;310(7):699-705.
report of HIV viral load among people with HIV on antiretroviral 47. Calza L, Magistrelli E, Colangeli V, et al. Significant association
treatment. HIV Med 2017;18(7):463-73. between statin-associated myalgia and vitamin D deficiency among
30. The American Academy of HIV Medicine. HIV specialist in crisis: the treated HIV-infected patients. AIDS 2017;31(5):681-8.
HIV workforce. 2016; https://aahivm.org/wp-content/uploads/ 48. Guyton JR, Bays HE, Grundy SM, et al. The National Lipid Association
2017/03/FINAL-MARCH-2016.pdf. Accessed May 31, 2019. Statin Intolerance P. An assessment by the Statin Intolerance Panel:
31. Wilson IB, Landon BE, Hirschhorn LR, et al. Quality of HIV care 2014 update. J Clin Lipidol 2014;8(3 Suppl):S72-81.
provided by nurse practitioners, physician assistants, and physicians. 49. Rosenson RS, Baker SK, Jacobson TA, et al. The National Lipid
Ann Intern Med 2005;143(10):729-36. Association's Muscle Safety Expert P. An assessment by the Statin
32. Ho PM, Lambert-Kerzner A, Carey EP, et al. Multifaceted intervention Muscle Safety Task Force: 2014 update. J Clin Lipidol 2014;8(3
to improve medication adherence and secondary prevention Suppl):S58-71.
measures after acute coronary syndrome hospital discharge: a 50. Margolis KL, Asche SE, Bergdall AR, et al. Effect of home blood
randomized clinical trial. JAMA Intern Med 2014;174(2):186-93. pressure telemonitoring and pharmacist management on blood
33. Bosworth HB, Powers BJ, Olsen MK, et al. Home blood pressure pressure control: a cluster randomized clinical trial. JAMA 2013;310
management and improved blood pressure control: results from a (1):46-56.
randomized controlled trial. Arch Intern Med 2011;171(13): 51. Santschi V, Chiolero A, Colosimo AL, et al. Improving blood pressure
1173-80. control through pharmacist interventions: a meta-analysis of ran-
34. Bosworth HB, Olsen MK, McCant F, et al. Hypertension Intervention domized controlled trials. J Am Heart Assoc 2014;3(2), e000718.
Nurse Telemedicine Study (HINTS): testing a multifactorial tailored 52. Shaw RJ, McDuffie JR, Hendrix CC, et al. Effects of nurse-managed
behavioral/educational and a medication management intervention protocols in the outpatient management of adults with chronic
for blood pressure control. Am Heart J 2007;153(6):918-24. conditions: a systematic review and meta-analysis. Ann Intern Med
35. Bosworth HB, Olsen MK, Grubber JM, et al. Two self-management 2014;161(2):113-21.
interventions to improve hypertension control: a randomized trial. 53. Blood Pressure Lowering Treatment Trialists C, Ninomiya T, Perkovic
Ann Intern Med 2009;151(10):687-95. V, et al. Blood pressure lowering and major cardiovascular events in
36. van Driel ML, Morledge MD, Ulep R, et al. Interventions to improve people with and without chronic kidney disease: meta-analysis of
adherence to lipid-lowering medication. Cochrane Database Syst Rev randomised controlled trials. BMJ. 2013;347:f5680.
2016;12, CD004371. 54. Silverman MG, Ference BA, Im K, et al. Association between lowering
37. Tsuji I, Imai Y, Nagai K, et al. Proposal of reference values for home LDL-C and cardiovascular risk reduction among different therapeutic
blood pressure measurement: prognostic criteria based on a interventions: a systematic review and meta-analysis. JAMA
prospective observation of the general population in Ohasama. 2016;316(12):1289-97.
Japan Am J Hypertens 1997;10(4 Pt 1):409-18. 55. Devaney B, Rossi P. Thinking through evaluation design options.
38. Roumie CL, Elasy TA, Greevy R, et al. Improving blood pressure Children and Youth Services Review 1997;19(7):587-606.
control through provider education, provider alerts, and patient