Advertisement

Journal of General Internal Medicine

, Volume 32, Issue 5, pp 524–533 | Cite as

Comparing Strategies for Lipid Lowering in Argentina: An Analysis from the CVD Policy Model–Argentina

  • Jonatan KonfinoEmail author
  • Alicia Fernandez
  • Joanne Penko
  • Antoinette Mason
  • Eugenio Martinez
  • Pamela Coxson
  • David Heller
  • Andrew Moran
  • Kirsten Bibbins-Domingo
  • Eliseo J. Pérez-Stable
  • Raul Mejía
Original Research

Abstract

Introduction

In Argentina, the national guidelines for lipid control emphasize the use of relatively inexpensive low- or moderate-potency statins by patients at high risk (>20 %) of a cardiovascular event. The objective of this study was to compare the impact and costs of the current national CVD prevention guidelines with regard to morbidity and mortality in Argentina with the impact and costs of three strategies that incorporate high-potency statins.

Methods

We used the CVD Policy Model–Argentina to model the proposed interventions. This model is a national-scale, state-transition (Markov) computer simulation model of the CVD incidence, prevalence, mortality, and costs in adults 35–84 years of age. We modeled three scenarios: scenario 1 lowers the risk threshold for treatment to >10 % according the Framingham Risk Score (FRS); scenario 2 intensifies statin potency under current treatment thresholds; and scenario 3 combines both scenarios by lowering the treatment threshold to ≥10 % FRS and intensifying statin potency.

Results

Scenario 1 would translate into 1400 fewer MIs and 500 fewer CHD deaths every year, a 3 % and 2 % reduction, respectively. Scenario 2 would lead to 2000 fewer MIs and 1000 fewer CHD deaths every year. Scenario 3 would result in the greatest reduction in MIs and CHD deaths, with 3400 fewer MIs and 1400 fewer CHD deaths every year, which translates to a 7 % and 6 % reduction, respectively. All scenarios were cost-effective if the cost of a high-potency statin pill was under US$0.25.

Conclusion

Incorporating those individuals with greater than 10 % cardiovascular risk and the use of high-potency statins into Argentina’s national lipid guidelines could result in fewer CHD deaths and events at a reasonable cost.

KEY WORDS

cardiovascular disease Argentina statin coronary heart disease hydroxy-3-methylglutaryl-coenzyme A reductase inhibitor prevention 

INTRODUCTION

Cardiovascular disease (CVD) is the leading cause of death in Argentina, and is responsible for 34 % of total deaths.1 Recent data indicate that the prevalence of hyperlipidemia, hypertension, and diabetes, already among the highest in Latin America, is increasing.2 The search for efficient and cost-effective strategies to improve primary and secondary prevention of cardiovascular disease is a national priority. Although lipid control represents one of the most effective strategies for reducing CVD risk, in Argentina (as in much of the rest of the world), cholesterol medications are underused, and only about 34 % of those with a diagnosis of hypercholesterolemia2 and 15 % of individuals requiring secondary prevention3 report using a medication to reduce low-density lipoprotein cholesterol (LDL-C). By contrast, although in the United States only 19 % of the population aged 30–79 are on statins, this percentage increases to 58 % among people with coronary artery disease.4

Statins are efficacious for primary and secondary prevention, and their use is a key recommendation of international CVD prevention guidelines.5 , 6More recent international guidelines emphasize the use of high-potency statins7 , 8 and suggest lowering treatment thresholds for primary prevention for patients whose overall risk profile conveys moderate cardiac risk. The 2013 American College of Cardiology/American Heart Association (ACC/AHA) guideline on the treatment of blood cholesterol recommends high-potency statins for individuals with a clinical history of CVD or with LDL-C ≥190 mg/dL; moderate-potency statins for diabetics aged 40–75; and moderate- to high-potency statin therapy for primary prevention in those aged 40–75 with CVD risk of 7.5 % or higher.9

In Argentina, the guidelines released by the Ministry of Health in 2009 recommend simvastatin for primary prevention in patients with diabetes or with 20 % or greater chance of having a myocardial infarction in the next 10 years, according to the Pan-American Heart Organization Cardiovascular Risk Calculator for Latin America, region B.10 In Argentina, however, the cost-effectiveness of using high- versus moderate-potency statins for secondary prevention and for high-risk primary prevention is unknown. As many Argentines fall into a moderate-risk category, with a 10–19 % chance of developing a cardiac event over 10 years, the cost-effectiveness of a broadened approach to primary prevention would also need to be established. This lack of data can impede decision-making, particularly among policymakers responsible for government health and medication programs, who must consider whether limited resources should be directed toward higher-cost medications versus other priorities such as expanding the reach of current guidelines.

In order to help address these urgent policy questions, we used a well-established computer simulation model to determine the impact on morbidity and mortality in Argentina of a) using high-potency statins to treat individuals with diabetes or 20 % or greater chance of experiencing a CVD event over the next 10 years; b) using simvastatin or other moderate-potency statins to treat people with 10 % or greater chance of a CVD event over the next 10 years; and to determine c) the cost-effectiveness of each approach from the perspective of the national government, which both administers a large national medication program for low-income people and bears the cost of hospitalization for their CVD events.

METHODS

Cardiovascular Disease Policy Model–Argentina

The CVD Policy Model is a national-scale, state-transition (Markov) computer simulation model of the coronary heart disease and stroke incidence, prevalence, mortality, and costs in adults 35–84 years of age, originally developed for the United States and used for over 25 years to forecast CVD trends and simulate the impact of CVD interventions in the US.11 , 12 The model was adapted for the population of Argentina using Argentina-specific demographic and epidemiologic inputs wherever possible, and calibrating to reproduce CVD events and deaths observed in the Argentine population. The CVD Policy Model–Argentina has been used previously to evaluate population-wide policies aimed at tobacco control13 and dietary salt restriction14 , 15 (see Supplementary Online Appendix 1 for details).

The model is composed of three sub-models: the demographic–epidemiological sub-model, the bridge sub-model, and the disease history sub-model. The demographic–epidemiological sub-model predicts coronary heart disease (CHD) and stroke incidence and non-CVD mortality in the population without CVD stratified by age, sex, and six modifiable risk factors: smoking status (active smoker or secondhand smoke-exposed or no exposure), systolic blood pressure (SBP), high-density lipoprotein cholesterol (HDL-C), LDL-C, use of medication to lower cholesterol (statins, other drugs, or no use), and presence or absence of diabetes mellitus. The incidence of CHD, stroke, and non-CVD death is determined by beta coefficients estimated from Framingham Heart Study Original Cohort (exams 13–28) and Offspring Cohort (exams 1–7) data using the counting process extension of the Cox proportional hazards model, which allows for time-dependent covariates.16 20 For the fraction of the population developing CVD, the bridge sub-model characterizes the initial CVD event and related events over the next 30 days. The disease history sub-model then predicts the rate of subsequent CVD events and deaths, stratified by age, sex, and CVD event history (Fig. 1). The model is written in Fortran 95 and compiled using the Lahey Fortran 95 compiler v7.2 (Lahey Computer Systems, Incline Village, NV). Table 1 summarizes the data sources for the CVD Policy Model–Argentina.
Fig. 1

Cardiovascular disease (CVD) policy model structure.

Table 1

Data Sources for the Cardiovascular Disease (CVD) Policy Model–Argentina

Data

Source

Population of Argentina and incoming 35-year-old persons, 2010–2050

Argentina National Statistics and Census Institute21

Prevalence

 Risk factor (means and distributions)

2009 National Risk Factors Survey, Ministry of Health22

Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) Study23

Incidence

 Acute myocardial infarction (MI)

Population-based MI registry in a Buenos Aires district24

 Stroke

National Hospital Discharge Registry25

 MI and stroke prevalence 2010

Population-based risk factor telephone survey in Buenos Aires, Ministry of Health10

Mortality

 Coronary heart disease*

Statistics and Information Department, Ministry of Health25

 Stroke

Statistics and Information Department, Ministry of Health25

One-day and 28-day CHD case fatality

 CHD

Argentine national hospital survey26

Ministry of Health admissions database25

 Stroke

Argentine National Registry (RENACER)27

 Stroke 28-day case fatality

Iquique Stroke Study (PISCIS)28

*International Classification of Diseases, 10th revision (ICD-10) codes I21, I22 for myocardial infarction; ICD-10 codes I20, I23–I25 for angina and other CHD; ICD-10 codes I461, I469, I472, I490, I460, I500, I501, I509, I514, I515, I516, I519, I709 for poorly defined cardiovascular disease events and death

ICD-10 codes I60-I69 for stroke deaths

A more detailed explanation of the CVD Policy Model and its translation to the Argentine population can be found in supplementary Online Appendix 1.

Modeling Assumptions and Simulated Interventions

We modeled four statin interventions, including one simulation representing wider implementation of current Argentine statin use guidelines and three scenarios reflecting variations to national guidelines, with input assumptions as detailed in Table 2. For our main simulations, we assumed 50 % compliance among those who qualify for a statin under the given scenario but are not currently taking a statin. The four simulated interventions were as follows:
Table 2

Input Parameters and Assumptions for CVD Policy Model Simulations

Parameter

Base case value

Range for sensitivity analyses

Reference

Treatment-associated reduction in LDL-C, %

  

29 , 30

 Moderate-potency

40

36–44

 

 High-potency

54.77

50.77–58.77

Annual cost of statin treatment (US$/person treated)

  

31

 Moderate-potency

43.80

21.90–65.70

 

 High-potency

91.25

45.63–136.88*

 

Treatment compliance/adherence

50 %

25 %–75 %

Assumed

Costs associated with medication management

  

32 , 33

 Physician visit at treatment initiation

12.90

6.45–19.35

 

 Liver panel at treatment initiation

9.16

4.58–13.74

 

 Annual lipid panel during treatment

12.60

6.30–18.90

 

Costs and QALY decrements from adverse events

 Myopathy

   

  Annual risk of episode

0.0001

 

29

  Annual QALY penalty per episode

0.00996

0.00498–0.0149

34

  Annual cost, &/episode

33.51

16.76–50.27

32 , 33

 Stroke

   

  Annual risk of episode

0.0001

 

29

  Annual QALY penalty per episode

0.3120

0.156–0.468

34

  Annual cost, &/episode

1083.69

541.85–1625.54

32 , 33

 Diabetes

   

  Annual risk of episode

0.00098

0.00029–0.001667

35

  Annual QALY penalty per episode§

0.0762

0.0381–0.114

34

  Annual cost, &/episode

4451.1

2225.55–6676.65

32 , 33

Annual discount rate

3 %

0–5 %

Assumed

* Sensitivity to this assumption tested both with probabilistic sensitivity analyses using the range listed and with deterministic analyses that evaluate the cost per pill values: $0.15, $0.20, $0.25 (main assumption), $0.30, $0.35

Sensitivity to this assumption tested with deterministic sensitivity analyses that evaluate the low and high end of the range

Cost for each episode of myopathy from one physician visit and three creatinine kinase blood tests

§QALY decrements associated with neuropathy and infections of the extremities

  1. (1)

    Current guidelines. For those without a history of cardiovascular disease, we treated 50 % of those not on a statin who had either greater than 20 % cardiovascular risk based on the Framingham Risk Score (FRS) or a history of diabetes. For those with a history of CVD, we assumed that 15 % were already on a statin, based on responses to a national survey3, and treated 50 % of the remaining CVD population. In alignment with current national guidelines, those newly treated were assumed to be on a moderate-dose statin (such as simvastatin 40 mg), with a corresponding decrease in mean LDL-C of 40 %.29 , 30

     
  2. (2)

    Scenario 1 – expansion of primary prevention target population. For scenario 1, we treated the population described for the current guidelines scenario and expanded the primary prevention population to include 50 % of treatment-naïve individuals with an FRS score of 10–20 %, assuming treatment with a moderate-dose statin.

     
  3. (3)

    Scenario 2 – use of high-potency statins with current guideline’s target population. For scenario 2, we treated the same population as described for the current guideline scenario, but simulated the effects of a high-potency statin (such as atorvastatin 40–80 mg, or rosuvastatin 20–40 mg), with an assumed decrease in LDL-C of 55 %.29 , 30 Those already taking a statin were assumed to experience the incremental benefit of moving from a moderate-dose to a high-potency statin.

     
  4. (4)

    Scenario 3 – expansion of primary prevention and use of high-potency statin with high-risk population. Scenario 3 involved a combination of interventions in scenarios 1 and 2, with those qualifying under current national guidelines treated with a high-potency statin and those with no history of CVD but an FRS score of 10–20 % treated with a moderate-dose statin.

     

We conducted 10-year simulations for the adult population of Argentina aged 35–84 years, a relevant time frame for near-term decision-making by the Argentine government. Results from scenarios 1, 2, and 3 were compared to those from the current guidelines simulation to describe the impact of alterations to national guidelines on myocardial infarctions, CHD deaths, and cost-effectiveness outcomes. All four scenarios were compared to a baseline simulation assuming current levels of statin use as observed in national surveys to determine the numbers needed to treat under each scenario to prevent one MI or CHD death over 10 years.

Cost-Effectiveness Estimates

Health care costs were estimated from the perspective of the health care system, and treatment costs assumed the direct cost of statin pills, medication management costs including laboratory tests and physician visits, and costs associated with adverse events including myopathy, diabetes, and stroke (Table 2). Indirect costs such as work loss or family assistance were not included in the analysis. Health care costs were obtained from Ministry of Health publications,36 the National Institute of Statistics and Census,37 reviews from non-governmental organizations,32 , 33 , 38 , 39 and published papers in peer-reviewed journals.40 42 The estimated cost per person per year of statin treatment was based on the direct cost of medications that the government currently pays for moderate-potency statins. Costs for moderate-potency statins were US$0.12 per day.43 To estimate the cost of high-potency statins from the government’s perspective, we applied a price ratio of high- to moderate-potency statins based on the average over-the-counter price difference between atorvastatin 40 mg pill (high) and simvastatin 40 mg pill (moderate), which yielded a ratio of 2.08 for high- to moderate-potency costs.31 Based on this adjustment, we estimated that a high-potency statin would cost US$0.25 per day.

Model inputs for QALYs associated with cardiovascular disease outcomes were derived from the Global Burden of Disease study.34 In each of our statin intervention simulations, we included quality-of-life decrements associated with adverse events, but excluded QALYS for the burden of taking a pill every day (Table 2), with our base case analyses assuming the same rate of toxicity for moderate- and high-potency statins. In our base case simulations, we used an annual discount rate of 3 %,44 with costs and QALYs discounted at the same rate. All monetary values are expressed in current US dollars at December 2013 conversion rates.

Sensitivity Analyses

We used both probabilistic and deterministic analyses to evaluate the sensitivity of our findings to our input assumptions. Monte Carlo simulations were used to generate 95 % confidence interval (95 % CI) around our primary outcomes measures for each intervention scenario. In addition to inputs relating to statin effectiveness, costs, and QALYs described in Table 2, we also varied beta values defining the relationship between LDL-C and incident CHD. There were 1000 random draws from a standard normal distribution, scaled to the mean and confidence interval for each varied parameter. The Monte Carlo program was written in Python and results analyzed using Microsoft Excel 2010. We used deterministic approaches to evaluate the sensitivity of our results to assumptions about the degree of statin compliance (range evaluated: 25–75 %), the discount rate (range evaluated: 0–5 %), and the rate of toxicity in moderate- vs. high-potency statins (sensitivity analysis assumed twofold greater toxicity for high-potency statins).

Model Calibration

The CVD-Policy Model–Argentina is based on assumptions originally derived from the Framingham population. Although the beta coefficients have a similar effect in different populations,45 we compared the predicted number of CHD deaths derived from the CVD Policy Model–Argentina to the actual CHD deaths captured in Argentina’s vital statistics for the years 1997–2009. We used the total CHD deaths in ages 35–84 years, which included definite CHD deaths (International Statistical Classification of Diseases and Related Health Problems, 10th revision [ICD-10] codes I20–I25) plus a percentage of poorly defined deaths (named “garbage” codes) that could be attributed to CHD (ICD-10 codes I461, I469, I472, I490, I460, I500, I501, I509, I514, I515, I516, I519, I709).46 We found a difference of less than 5 % between the total CHD deaths mentioned in the national statistics in the last year available (24,246 for 2009) and the first estimation of total CHD deaths from the CVD Policy Model (25,640 for 2010).1 Regarding stroke deaths, the difference observed was less than 10 %.

RESULTS

Projected Impact of Change in Lipid Control Guidelines

Estimates from the CVD Policy Model suggest that if half of those who qualified for statins under current guidelines were treated, approximately 3.7 million people would be treated. The three proposed scenarios predict a significant reduction in MIs and CHD deaths, compared with treating more people under current guidelines (Table 2). Scenario 1 (expand primary prevention with moderate statins) lowers the risk threshold to >10 % FRS and would translate into 1400 fewer MIs (95 % UI: 900–2000) and 500 fewer CHD deaths (95 % UI: 200–800) every year, a 3 % and 2 % reduction, respectively.

Scenario 2 (high-potency statins for high-risk patients) intensifies statin potency under current guidelines; this intensification would lead to 2000 fewer MIs (95 % UI: 1400–2700) and 1000 fewer CHD deaths (95 % UI: 700–1300) every year. Scenario 3 (lower primary treatment threshold plus high-potency statin for high-risk patients) combines scenarios 1 and 2, and would result in the greatest reduction in MIs and CHD deaths, with 3400 fewer MIs (95 % UI: 2500–4200) and 1400 fewer CHD deaths (95 % UI: 1100–1800) every year, which translates to reductions of 7 % and 6 %, respectively.

The combined impact on morbidity and mortality using QALYs is also illustrated in Table 3. Compared to treating more people under current guidelines, an additional 2500 (95 % UI: 1500–3400), 3300 (95 % UI: 2300–4300), and 5600 (95 % UI: 4200–7000) QALYs could be gained if scenarios 1, 2, and 3 were implemented, respectively.
Table 3

Projected Average Annual Myocardial Infarctions, Coronary Heart Disease Deaths and Quality-Adjusted Life Years, and Changes in these Outcomes Relative to Expanding Statin Use Under Current Guidelines, Estimated Using the CVD Policy Model–Argentina

Simulation scenario*

Average annual number treated§

Total MIs (95 % UI)

% Change in MI

Total CHD deaths (95 % UI)

% Change in CHD deaths

Total QALYs

Change in QALYs (95 % UI)

Current guidelines (comparator)

3,700,900

47,300 (46,100–48,500)

 

23,000 (22,500–23,400)

 

15,633,000 (15,632,000–15,634,000)

 

Scenario 1: lower primary prevention treatment threshold

4,900,800

45,900 (44,600–47,170)

−3.03 %

22,500 (21,900–23,000)

−2.13 %

15,635,000 (15,634,000–15,637,000)

2500 (1500–3500)

Scenario 2: use high-potency statin for high-risk patients

3,700,900

45,300 (43,900–46,700)

−4.24 %

22,000 (21,300–22,600)

−4.35 %

15,636,000 (15,634,000–15,638,000)

3300 (2300–4300)

Scenario 3: use high-potency statins for those at risk under current guidelines; use moderate-potency with those 10–20 % FRS

5,000,900

44,000 (42,400–45,500)

−7.11 %

21,500 (20,800–22,200)

−6.30 %

15,638,500 (15,636,000–15,641,000)

5600 (4200–7000)

MI myocardial infarction, CHD coronary heart disease, QALY quality-adjusted life years

* Current guidelines: treat anyone with a history of cardiovascular disease (CVD), diabetes, or >20 % cardiovascular risk based on the Framingham Risk Score (FRS) with a moderate potency statin. Scenario 1: expand current guidelines to include those with >10 % risk based on FRS. Scenario 2: treat those qualifying under current guidelines with a high potency instead of moderate-potency statin. Scenario 3: combination of scenarios 1 and 2. All scenarios assume 50 % compliance among those who qualify for but are not taking a statin

Changes in % MIs, % CHD deaths, and QALYs shown for simulations 1, 2, and 3 result from a comparison to the “current guidelines” simulation, which models the expansion of treatment with moderate-dose statins to 50 % of those who qualify under current national guidelines but who are not currently treated with statins

50 % of untreated gain the effect of going from no statin to a high-potency statin; those already on statins gain the additional LDL lowering by moving from moderate to high potency

§ Includes those currently taking statins

We also examined the number needed to treat (NNT) to avoid an MI or CHD death in the next decade (Fig. 2). Compared to baseline, an additional 39 (95 % UI: 32–47) individuals who qualify for statins under current guidelines but are not receiving them would need to be treated each year with a moderate-dose statin to avoid one MI or CHD death over 10 years. The corresponding NNT over 10 years to avoid a CHD death or MI for scenarios 1, 2, and 3 was 45 (95 % UI: 37–55), 45 (95 % UI: 37–55), and 48 (95 % UI: 41–59), respectively.
Fig. 2

Projected numbers needed to treat (NNT) annually, in addition to those currently treated, to avoid one coronary heart disease death or myocardial infarction over 10 years. Current guidelines: treat anyone with a history of cardiovascular disease (CVD), diabetes, or >20 % cardiovascular risk based on the Framingham Risk Score (FRS) with a moderate-potency statin. Scenario 1: expand current guidelines to include those with >10 % risk based on FRS. Scenario 2: treat those qualifying under current guidelines with a high-potency instead of moderate-potency statin. Scenario 3: combination of scenarios 1 and 2. To estimate the numbers needed to treat, each scenario (including “current guidelines”) was compared to a simulation assuming no change in statin treatment from status quo. Numbers treated include only treatment-naïve individuals (i.e., those not on statins under status quo conditions) for the current guidelines simulation and for scenario 1; in scenarios 2 and 3, the total treated include those on statins at baseline who are assumed to move from a moderate- to high-dose statin. Error bars represent 95 % uncertainty intervals from Monte Carlo simulation results.

Cost and Cost-Effectiveness Analysis

We estimated that achieving 50 % compliance with current statin use guidelines would cost approximately US$108 million every year in medication costs. Scenario 1 would cost an additional US$66 million per year but would save US$17 million annually in terms of total health care costs. Scenarios 2 and 3 would cost US$15 million and US$21 million more, respectively, than increasing the number treated under current guidelines, but would save US$11 million and US$28 million in annual health care costs, respectively (Table 4).
Table 4

Projected Average Annual Costs and Cost-Effectiveness for Simulation Scenarios Compared to Current Guidelines, Estimated using the CVD Policy Model–Argentina (in US$)

Simulation scenario

Cost of intervention (95 % UI)

Change in cost of intervention (95 % UI)

Total health care costs§ (95 % UI)

Change in total health care costs (95 % UI)

ICER (95 % UI)

Current guidelines (assumed to reach 50 % currently untreated)

(comparator)

107,878,000 (53,594,000 to 160,225,000)

 

79,414,310,000 (79,152,317,000 to 79,607,511,000)

  

Scenario 1: expand current guidelines to include 50 % of FRS 10–20 %

173,939,000 (89,441,000 to 253,980,000)

66,061,000 (−31,558,000 to 162,110,000)

79,397,070,800 (79,134,772,000 to 79,591,351,000)

−17,239,000 (−13,128,000 to −21,729,000)

19,900 (cost saving, 66,800)

Scenario 2: use high-potency statins for those at risk under current guidelines*

256,399,000 (116,972,000 to 379,007,000)

148,522,000 (5,866,000 to 281,032,000)

79,403,200,000 (79,142,744,000 to 79,594,989,000)

−11,110,000 (−8,223,000− −14,655,000)

42,200 (cost saving, 93,400)

Scenario 3: use high-potency statins for those at risk under current guidelines*; use moderate-potency with those 10–20 % FRS

322,087,800 (165,585,100 to 492,008,500)

214,210,300 (37,991,000 to 396,783,000)

79,386,273,000 (79,122,326,000 to 79,580,634,000)

−28,036,800 (−22,291,000 to −34,120,000)

33,100 (2300 to 67,500)

ICER incremental cost-effectiveness ratio (additional cost per quality-adjusted life years gained)

*50 % of untreated gain the effect of going from no statin to a high-potency statin; those already on statins gain the additional LDL lowering by moving from moderate to high potency

Includes cost of medication, chronic disease management related to statin delivery and assessment, and cost associated with adverse events (diabetes, stroke, and myopathy)

Changes in costs and ICERs are computed for each scenario relative to a simulation assuming current guidelines are expanded to reach 50 % of the untreated population

§Includes costs associated with coronary heart disease, stroke, and background health care costs

The incremental cost-effectiveness ratio (ICER; additional cost/QALY gained) is projected to be 19,900 for scenario 1 (95 % UI: cost saving, 66,800), 42,200 for scenario 2 (95 % UI: cost saving, 93,400), and 33,100 for scenario3 (95 % UI: 2300–67,500), with each scenario compared to increasing to 50 % compliance with current guidelines. With a gross domestic product (GDP) per capita in Argentina (2009–2013) of approximately US$14,760,47 all three scenarios would be considered cost-effective, as ICERs less than 1 times the per capita GDP are typically considered very good values and ICERs 1-3 times the per capita GDP are acceptable values under the WHO guidelines on cost-effectiveness.48 50

Sensitivity Analyses

Table 5 shows results from scenarios 2 and 3 using different values for the cost of high-potency statins. Both scenarios were cost-effective at a cost per pill of US$0.25 or lower.
Table 5

Projected Cost and Cost-Effectiveness Outcomes Assuming Varying Costs of High-Potency Statins, Estimated Using the CVD Policy Model–Argentina (in US$)

Scenario

Price per pill (high-potency)

Annual cost of intervention

Change relative to current guidelines

Annual total health care cost

Change relative to current guidelines*

ICER

Scenario 2

0.35

370,749,000

262,872,000

79,403,200,000

−11,109,000

77,300

Scenario 3

436,154,000

328,276,000

79,386,273,000

−28,037,000

53,400

Scenario 2

0.3

313,582,000

205,705,000

79,403,200,000

−11,109,500

59,800

Scenario 3

379,129,000

271,251,000

79,386,273,000

−28,037,000

43,300

Scenario 2

0.25

256,399,000

148,522,000

79,403,200,000

−11,109,000

42,200

Scenario 3

322,088,000

214,210,000

79,386,273,000

−28,037,00

33,100

Scenario 2

0.2

199,233,000

91,355,000

79,403,200,000

−11,109,000

24,700

Scenario 3

265,063,000

157,186,000

79,386,273,000

−28,037,000

23,000

Scenario 2

0.15

142,066,000

34,188,000

79,403,200,000

−11,109,000

7100

Scenario 3

208,039,000

100,161,000

79,386,273,000

−28,037,000

12,800

*Differences between each scenario and the base case of current guidelines are shown in parentheses

ICER = incremental cost-effectiveness ratio (additional cost per quality-adjusted life years gained). ICERs are relative to the base case of current guidelines

Although changes in the compliance rate (25 and 75 %) modified the magnitude of events avoided, the three scenarios remained effective in terms of reducing myocardial infarctions and CHD deaths. The three scenarios remained cost-effective under a broad range of assumptions. Results were moderately sensitive to changes in compliance rate. Only scenario 2, when assuming 75 % of compliance, did not remain cost-effective (ICER 48,100). Results were somewhat sensitive to the level of toxicity assumed for high-potency statins; scenario 2 was not cost-effective (ICER 51,600) when high-potency statins were assumed to have twofold increased toxicity compared to moderate-dose statins. Results were not sensitive to varying discount rates.

Willingness-to-pay acceptability curves for scenarios 2 and 3 are presented in Fig. 3.
Fig. 3

Willingness-to-pay acceptability curves.

DISCUSSION

In this study, we modeled the potential health impact, cost, and cost-effectiveness of changing the national guidelines for lipid control for cardiovascular disease prevention in Argentina. We found that the three strategies examined—lowering primary prevention treatment thresholds, using high-potency statins for high-risk patients, and doing both simultaneously—would not only avoid thousands of cardiac events and deaths every year, but would also be cost-effective under international criteria.48 50

These results are similar to published data from other countries, which found that 5651 to 104 people52 needed to be treated with a statin for 5 years to avoid a major CVD event. A separate study found that high-potency statins in national guidelines could be cost-effective only if the cost of the pills were sufficiently low.41 Our study extends these observations to an Argentine context.

Our study adopts a government perspective on cost tradeoffs between medications and hospitalizations. As in many countries in Latin America, Argentina’s government plays a large role in ensuring health care. In early 2014, Argentina included simvastatin in the national medication program REMEDIAR, which over the past 10 years has provided “essential drugs” at no cost to 16 million people receiving health care within the national network of primary care centers.53 , 54 In general, in Argentina, access to essential medications such as lipid-lowering drugs has been considered a social good that should be made readily available to all.55 As the government also bears the costs of CVD hospitalization for approximately 40 % of the population, cost-effectiveness studies such as this one are highly relevant for government policymakers.

Argentina has used evidence-based strategies in the past to inform national policy. For example, other modeling studies using the CVD Policy Model–Argentina on tobacco control13 and salt consumption14 , 15 suggested policies that were subsequently implemented. Other Latin American countries including Chile and Brazil also have national CVD prevention guidelines56 , 57 that focus on lipids; modeling exercises like this one, which are commonly used, for example, in countries of the United Kingdom,58 can be instrumental in the development of evidence-based guidelines that incorporate resource impact.

Given the limited resources available, only those interventions that can lead to large reductions in the CVD burden at relatively low cost are likely to be sustainable.59 The scenarios modeled in this study highlight the importance of lipid-lowering drugs and guidelines for cardiovascular disease prevention in Argentina. Two of our scenarios extend drug treatment to large numbers of the population. This raises the potential for “over-medicalization” and for the tacit acceptance of unhealthy elements of the Argentine diet. We made no attempt to consider other alternatives, such as government-funded social marketing campaigns that could contribute to sustained cholesterol lowering, as we have no data on their effectiveness from an Argentine perspective. Much more research on the effective promotion of broad changes in diet is needed in Argentina and in Latin America as a whole.

Some other limitations of our study should be considered. We assumed that everyone who was currently on statin therapy in Argentina was treated with a moderate-potency statin, as is currently suggested by national guidelines. As there are no national data available on which statins patients are actually using, we were unable to test this assumption. Second, as noted, we considered the health care costs from the government’s perspective; these results might not be generalizable to the private health sector, which has different health care costs. Finally, all modeling studies are limited by the integrity of their inputs. We used the best available data from Argentina to adapt the US CVD Policy Model for Argentina. Although there is debate on how relevant the Framingham Risk Score is to non-US populations,60 because of a lack of Argentine data, we used the Framingham data to determine the association between CVD risk factors and CVD outcomes. Prior studies adapting Framingham to other international contexts have suggested that the associations between risk factors and CVD events (beta) appear to be similar across populations, although actual event rates (alpha) may differ among populations.61 Event rates for the CVD Policy Model–Argentina were calibrated with the best available epidemiological data from Argentina, comparing the predictions of the CVD Policy Model with local vital statistics; differences in event rate are less than 5 %. Previously published papers have used the CVD Policy Model to accurately predict outcomes in the Argentine population, which increases our confidence in the model’s accuracy.13 15 , 62

Despite these limitations, we believe that this study contributes to the ongoing debate in Argentina by projecting the health benefits of three possible strategies for lipid control in cardiovascular disease prevention. To our knowledge, this study is the first to analyze the impact of current national guidelines for cardiovascular disease prevention in Latin America or to model alternatives to current guidelines based on lowering the primary prevention threshold or treating high-risk patients with high-potency statins, as is currently recommended in many other international guidelines. Incorporating modeling strategies in the development of national guidelines could help provide a better understanding of cost-effective interventions in the region. This type of approach may also be useful in the future for modeling additive interventions in this population, such as control of hypertension and use of low-dose aspirin. The data provided here should be key inputs for policymakers who seek to improve cardiovascular disease prevention in Argentina.

Notes

Compliance with Ethical Standards

Disclaimer

The content and views expressed in this manuscript are those of the authors and should not be construed to represent the views of the National Institutes of Health or any of the sponsoring organizations and agencies of the US government.

Funding

Dr. Fernandez was supported by grant K24DK102057 from the National Institutes of Health (NIH) and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK).

Conflict of Interest

The authors declare that they do not have a conflict of interest.

Supplementary material

11606_2016_3907_MOESM1_ESM.doc (262 kb)
ESM 1 (DOC 261 kb)

References

  1. 1.
    Ministerio de Salud de la Nación. ESTADISTICAS VITALES - INFORMACION BASICA AÑO 2014. In: DIRECCION DE ESTADISTICAS E INFORMACION DE SALUD, ed. http://deis.msal.gov.ar/wp-content/uploads/2016/07/BoletinNro153.pdf. Accessed: October 12, 2016.
  2. 2.
    Ministerio de salud de la Nación, Instituto Nacional de Estadística y Censos. Tercera Encuesta Nacional de Factores de Riesgo. Resumen ejecutivo. Buenos Aires; 2014.Google Scholar
  3. 3.
    Yusuf S, Islam S, Chow CK, Rangarajan S, Dagenais G, Diaz R, et al. Use of secondary prevention drugs for cardiovascular disease in the community in high-income, middle-income, and low-income countries (the PURE study): a prospective epidemiological survey. Lancet. 2011;378(9798):1231–1243.CrossRefPubMedGoogle Scholar
  4. 4.
    Johansen ME, Green LA, Sen A, Kircher S, Richardson CR. Annals journal club: cardiovascular risk and statin use in the United States. Ann Fam Med. 2014;12(3):215–23.CrossRefPubMedPubMedCentralGoogle Scholar
  5. 5.
    Ansell BJ, Watson KE, Fogelman AM. An evidence-based assessment of the NCEP adult treatment panel II guidelines. National cholesterol education program. JAMA. 1999;282(21):2051–2057.CrossRefPubMedGoogle Scholar
  6. 6.
    Prosser LA, Stinnett AA, Goldman PA, Williams LW, Hunink MG, Goldman L, et al. Cost-effectiveness of cholesterol-lowering therapies according to selected patient characteristics. Ann Intern Med. 2000;132(10):769–779.CrossRefPubMedGoogle Scholar
  7. 7.
    Minder CM, Blaha MJ, Horne A, Michos ED, Kaul S, Blumenthal RS. Evidence-based use of statins for primary prevention of cardiovascular disease. Am J Med. 2012;125(5):440–446.CrossRefPubMedGoogle Scholar
  8. 8.
    Mortensen MB, Falk E. Real-life evaluation of European and American high-risk strategies for primary prevention of cardiovascular disease in patients with first myocardial infarction. BMJ Open. 2014;4(10):e005991.CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Stone NJ, Robinson J, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2013.Google Scholar
  10. 10.
    Ministerio de Salud de la Nación. Factibilidad de Implementación de un sistema de vigilancia telefónico para factores de riesgo de enfermedades crónicas no transmisibles. Bol Vigilancia. 2011;5:5–21.Google Scholar
  11. 11.
    Weinstein M, Coxson P, Williams L, Pass T, Stason W, Goldman L. Forecasting coronary heart disease incidence, mortality, and cost: the coronary heart disease policy model. Am J Public Health. 1987;77(11):1417–1426.CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Moran AE, Odden MC, Thanataveerat A, Tzong KY, Rasmussen PW, Guzman D, et al. Cost-effectiveness of hypertension therapy according to 2014 guidelines. N Engl J Med. 2015;372(5):447–455.CrossRefPubMedPubMedCentralGoogle Scholar
  13. 13.
    Konfino J, Ferrante D, Mejía R, Coxson P, Moran A, Goldman L, et al. Impact on cardiovascular disease events of the implementation of Argentina’s national tobacco control law. Tob Control. 2012.Google Scholar
  14. 14.
    Konfino J, Mekonnen TA, Coxson PG, Ferrante D, Bibbins-Domingo K. Projected impact of a sodium consumption reduction initiative in Argentina: an analysis from the CVD policy model - Argentina. PLoS One. 2013;8(9):e73824.CrossRefPubMedPubMedCentralGoogle Scholar
  15. 15.
    Ferrante D, Konfino J, Mejía R, Coxson P, Moran A, Goldman L, et al. Relación de la costo utilidad de la disminución del consumo de sal y su efecto en la incidencia de enfermedades cardiovasculares en la Argentina. Rev Panam Salud Pública. 2012;32(4):274–280.CrossRefPubMedGoogle Scholar
  16. 16.
    Framingham Heart Study-Cohort (FHS-Cohort) data request site. Biologic Specimen and Data Repository Information Coordinating Center: National Heart L, and Blood Institute. Available from: http://biolincc.nhlbi.nih.gov/studies/framcohort/?q=framingham. Accessed: October 12, 2016.
  17. 17.
    Framingham Heart Study-Offspring Study (FHS-OS) data request site. Biologic Specimen and Data Repository Information Coordinating Center: National Heart L, and Blood Institute. Available from: http://biolincc.nhlbi.nih.gov/studies/framoffspring/?q=framingham. Accessed: October 12, 2016.
  18. 18.
    Dawber TR. The Framingham Study: the epidemiology of atherosclerotic disease. Cambridge, MA1980Google Scholar
  19. 19.
    Feinleib M, Kannel WB, Garrison RJ, McNamara PM, Castelli WP. The Framingham offspring study. Design and preliminary data. Prev Med. 1975;4(4):518–525.CrossRefPubMedGoogle Scholar
  20. 20.
    Kleinbaum DG, Klein M. Survival analysis: a self-learning text. New York: Springer; 2012:366–379.CrossRefGoogle Scholar
  21. 21.
    Instituto Nacional de Estadísticas y Censos. Ministerio de Economía. 2011 [cited 2016 October 12]. Available from: www.indec.gov.ar
  22. 22.
    Ministerio de Salud de la Nación. Segunda Encuesta Nacional de Factores de Riesgo. Buenos Aires: Ministerio de Salud de la Nación; 2011.Google Scholar
  23. 23.
    Schargrodsky H, Hernandez-Hernandez R, Champagne B, Silva H, Vinueza R, Silva Aycaguer L, et al. CARMELA: assessment of cardiovascular risk in seven Latin American cities. Am J Med. 2008;1(121):58–65.CrossRefGoogle Scholar
  24. 24.
    Caccavo A, Alvarez A, Bello F, Ferrari A, Carrique A, Lasdica S, et al. Incidencia poblacional del infarto con elevación del ST o bloqueo de rama izquierda a lo largo de 11 años en una comunidad de la provincia de Buenos Aires. Rev Argent Cardiol. 2007;75:185–188.Google Scholar
  25. 25.
    Ministerio de Salud de la Nación. ESTADISTICAS VITALES - INFORMACION BASICA AÑO 2010. In: DIRECCION DE ESTADISTICAS E INFORMACION DE SALUD, ed. 2012: http://deis.msal.gov.ar/wp-content/uploads/2016/01/Boletin136.pdf
  26. 26.
    Blanco P, Gagliardi J, Higa C, Dini A, Guetta J, Di Toro D, et al. Infarto agudo de miocardio. Resultados de la Encuesta SAC 2005 en la República Argentina. Rev Argent Cardiol. 2007;75:163–179.Google Scholar
  27. 27.
    Sposato L, Esnaola M, Zamora R, Zurru M, Fustinoni O, Saposnik G. Quality of ischemic stroke care in emerging countries: the Argentinean national stroke registry (ReNACer). Stroke. 2008;39:3036–3041.CrossRefPubMedGoogle Scholar
  28. 28.
    Lavados P, Sacks C, Prina L, Escobar A, Tossi C, Araya F, et al. Incidence, 30-day case-fatality rate, and prognosis of stroke in Iquique, Chile: a 2-year community-based prospective study (PISCIS project). Lancet. 2005;365(9478):2206–2215.CrossRefPubMedGoogle Scholar
  29. 29.
    Stone NJ, Robinson JG, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2014;63:2889–2934.CrossRefPubMedGoogle Scholar
  30. 30.
    Cholesterol Treatment Trialists’ Collaborators MB, Emberson J, Blackwell L, Keech A, Simes J, Barnes EH, Voysey M, Gray A, Collins R, Baigent C. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet. 2012;380(9841):581–590.CrossRefGoogle Scholar
  31. 31.
    K@iros. Revista de Ciencia y Tecnologia para su farmacia 2014 [cited 2014 February 20th]. Available from: http://ar.kairosweb.com/laboratorios/producto-simvastatina-fabra-20519, http://ar.kairosweb.com/laboratorios/producto-atorvastatina-fabra-19965
  32. 32.
    Base de Datos de Costos Sanitarios Argentinos, Documento Técnico N°3 [Internet]. 2010.Google Scholar
  33. 33.
    Nomenclador Bioquímico Único – PMO y Prácticas Especiales- Versión 2010. Confederación Unificada Bioquímica de la Republica Argentina.Google Scholar
  34. 34.
    The Institute for Health Metrics and Evaluation. Global burden of disease study 2010 Disability Weights 2010 [cited 2016 October 12]. Available from: http://ghdx.healthdata.org/record/global-burden-disease-study-2010-gbd-2010-disability-weights
  35. 35.
    Sattar N, Preiss D, Murray HM, Welsh P, Buckley BM, de Craen AJ, et al. Statins and risk of incident diabetes: a collaborative meta-analysis of randomised statin trials. Lancet. 2010;375(9716):735–742.CrossRefPubMedGoogle Scholar
  36. 36.
    Nomenclador del Hospital Público de Gestión Descentralizada-Actualización 2011- Ministerio de Salud de la Republica Argentina.Google Scholar
  37. 37.
    Instituto Nacional de Estadística y Censos. Índice de Precios al Consumidor – Capítulo: Atención Médica y Gastos para la Salud. 2011.Google Scholar
  38. 38.
    Pichon-Riviere A AA, Bardach A, Augustovski F, Caporale J, Caccavo F. Carga de Enfermedad atribuible al Tabaquismo en Argentina. Documento Técnico IECS N° 7. Instituto de Efectividad Clínica y Sanitaria, Buenos Aires, Argentina. Mayo de 2013 (www.iecs.org.ar).
  39. 39.
    Riapari N, MoscosoN, Budassi N. Estudio de Costos de Cirugías Cardiovasculares. Anales de la Asociación Argentina de Economía Política. 2010.Google Scholar
  40. 40.
    Rubinstein A, Colantonio L, Bardach A, Caporale J, Marti SG, Kopitowski K, et al. Estimation of the burden of cardiovascular disease attributable to modifiable risk factors and cost-effectiveness analysis of preventative interventions to reduce this burden in Argentina. BMC Public Health. 2010;10:627.CrossRefPubMedPubMedCentralGoogle Scholar
  41. 41.
    Pletcher MJ, Lazar L, Bibbins-Domingo K, Moran A, Rodondi N, Coxson P, et al. Comparing impact and cost-effectiveness of primary prevention strategies for lipid-lowering. Ann Intern Med. 2009;150(4):243–254.CrossRefPubMedGoogle Scholar
  42. 42.
    Borraci R, Rubio M, Insua J. Análisis de costos médicos y resultados de la cirugía coronaria sin circulación extracorpórea. Rev Argent Cardiol. 2006;74(5).Google Scholar
  43. 43.
    Ministerio de Salud de la Nación, Programa REMEDIAR + Redes, Personal Communication. 2014.Google Scholar
  44. 44.
    Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the economic evaluation of health care programmes. Fourthth ed. Oxford: Oxford University Press; 2015.Google Scholar
  45. 45.
    D’Agostino R, Grundy S, Sullivan LM, Wilson P. CHD Risk Prediction Group. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001;286(2):180–187.CrossRefPubMedGoogle Scholar
  46. 46.
    Naghavi M, Foreman K, O’Brien J, Pourmalek F, Lozano R. Algorithms for enhancing public health utility of national causes-of-death data. Popul Health Metrics. 2010;8(9).Google Scholar
  47. 47.
    World Bank. GDP per capita 2015 [cited 2015 september 25th]. Available from: http://data.worldbank.org/indicator/NY.GDP.PCAP.CD
  48. 48.
    Ferrante D, Lew D, Peruga A, Compton C, Romano E. The role of public policies in reducing smoking prevalence and deaths: the Argentina tobacco policy simulation model. Rev Panam Salud Pública. 2007;21(1):37–49.CrossRefPubMedGoogle Scholar
  49. 49.
    Levy DT, Nikolayev L, Mumford EA. Recent trends in smoking and the role of public policies: results from the SimSmoke tobacco control policy simulation model. Addiction. 2005;100(10):1526–1537.Google Scholar
  50. 50.
    World Health Organization. Who Guide to Cost - Effectiveness Analysis 2003. Available from: http://www.who.int/choice/publications/p_2003_generalised_cea.pdf
  51. 51.
    Taylor F HM, Macedo A, Moore THM, Burke M, Davey Smith G, Ward K, Ebrahim S. Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev 2013, Issue 1. Art. No.: CD004816; doi:  10.1002/14651858.CD004816.pub5.
  52. 52.
    Ray KK, Seshasai SR, Erqou S, Sever P, Jukema JW, Ford I, et al. Statins and all-cause mortality in high-risk primary prevention: a meta-analysis of 11 randomized controlled trials involving 65,229 participants. Arch Intern Med. 2010;170(12):1024–1031. doi: 10.1001/archinternmed.2010.182.
  53. 53.
    Ministerio de Salud de la Nación. Programa REMEDIAR + Redes. Available from: http://www.msal.gov.ar/
  54. 54.
    Cerezo L, Dias JM, Fernández Prieto A, La Cava G, Luque PS, Monsalvo M, et al. Impacto redistributivo del programa Remediar en el gasto en medicamentos: estudio cuantitativo. Rev Argent Salud Pública. 2014;5(20):40–43.Google Scholar
  55. 55.
    Dondo M, Monsalvo M, Garibaldi L. Determinants of equity in financing medicines in Argentina: an empirical study. Cad Saúde Pública. 2016;32(1):1–14.CrossRefGoogle Scholar
  56. 56.
    Ministerio de Salud de Chile. Dislipemias 2000. Available from: http://www.minsal.gob.cl/portal/url/item/75fefc3f8128c9dde04001011f0178d6.pdf
  57. 57.
    Ministerio da Saude Brazil. DISLIPIDEMIA: PREVENÇÃO DE EVENTOS CARDIOVASCULARES E PANCREATITE 2013 [cited 2016 October 12]. Available from: http://portalsaude.saude.gov.br/images/pdf/2014/abril/02/pcdt-dislipidemia-livro-2013.pdf
  58. 58.
    NICE. National Institute for Health and care Excellence. [cited 2016 October 12]. Available from: https://www.nice.org.uk/gui
  59. 59.
    Gaziano TA. Cardiovascular disease in the developing world and its cost-effective management. Circulation. 2005;112(23):3547–3553.CrossRefPubMedGoogle Scholar
  60. 60.
    Cortes-Bergoderi M, Thomas RJ, Albuquerque FN, Batsis JA, Burdiat G, Perez-Terzic C, et al. Validity of cardiovascular risk prediction models in Latin America and among Hispanics in the United States of America: a systematic review. Rev Panam Salud Publica. 2012;32(2):131–139.CrossRefPubMedGoogle Scholar
  61. 61.
    D’Agostino RB Sr, Grundy S, Sullivan LM, Wilson P, CHD Risk Prediction Group. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001;286(2):180–187.CrossRefPubMedGoogle Scholar
  62. 62.
    Moran A, Degennaro V, Ferrante D, Coxson P, Palmas W, Mejía R, et al. Coronary heart disease and stroke attributable to major risk factors is similar in Argentina and the United States: the coronary heart disease policy model. Int J Cardiol. 2011;150(3):332–337.CrossRefPubMedPubMedCentralGoogle Scholar

Copyright information

© Society of General Internal Medicine 2016

Authors and Affiliations

  • Jonatan Konfino
    • 1
    Email author
  • Alicia Fernandez
    • 1
    • 2
  • Joanne Penko
    • 2
  • Antoinette Mason
    • 3
  • Eugenio Martinez
    • 4
  • Pamela Coxson
    • 2
  • David Heller
    • 5
  • Andrew Moran
    • 6
    • 7
  • Kirsten Bibbins-Domingo
    • 2
  • Eliseo J. Pérez-Stable
    • 8
    • 9
  • Raul Mejía
    • 1
    • 2
    • 3
    • 4
    • 5
    • 6
    • 7
    • 8
    • 9
    • 10
  1. 1.Centro de Estudios de Estado y Sociedad (CEDES)Buenos AiresArgentina
  2. 2.Division of General Internal Medicine, Department of MedicineUniversity of California San FranciscoSan FranciscoUSA
  3. 3.University of California San Diego School of MedicineLa JollaUSA
  4. 4.Instituto de Investigaciones Económicas, Facultad de Cs. EconómicasUniversidad Nacional de SaltaSaltaArgentina
  5. 5.Arnhold Institute for Global HealthIcahn School of Medicine at Mount SinaiNew YorkUSA
  6. 6.Division of General Internal MedicineColumbia University Medical CenterNew YorkUSA
  7. 7.New York College of Physicians and SurgeonsColumbia UniversityNew YorkUSA
  8. 8.Division of General Internal Medicine, Department of Medicine, Medical Effectiveness Research Center for Diverse PopulationsUniversity of California San FranciscoSan FranciscoUSA
  9. 9.Office of the Director, National Institute of Minority Health and Health DisparitiesNational Institutes of HealthBethesdaUSA
  10. 10.Hospital de ClinicasUniversity of Buenos AiresBuenos AiresArgentina

Personalised recommendations