Projections of incident atherosclerotic cardiovascular disease and incident type 2 diabetes across evolving statin treatment guidelines and recommendations: A modelling study.

BACKGROUND: Experimental and observational research has suggested the potential for increased type 2 diabetes (T2D) risk among populations taking statins for the primary prevention of atherosclerotic cardiovascular disease (ASCVD). However, few studies have directly compared statin-associated benefits and harms or examined heterogeneity by population subgroups or assumed treatment effect. Thus, we compared ASCVD risk reduction and T2D incidence increases across 3 statin treatment guidelines or recommendations among adults without a history of ASCVD or T2D who were eligible for statin treatment initiation. METHODS AND
FINDINGS: Simulations were conducted using Markov models that integrated data from contemporary population-based studies of non-Hispanic African American and white adults aged 40-75 years with published meta-analyses. Statin treatment eligibility was determined by predicted 10-year ASCVD risk (5%, 7.5%, or 10%). We calculated the number needed to treat (NNT) to prevent one ASCVD event and the number needed to harm (NNH) to incur one incident case of T2D. The likelihood to be helped or harmed (LHH) was calculated as ratio of NNH to NNT. Heterogeneity in statin-associated benefit was examined by sex, age, and statin-associated T2D relative risk (RR) (range: 1.11-1.55). A total of 61,125,042 U.S. adults (58.5% female; 89.4% white; mean age = 54.7 years) composed our primary prevention population, among whom 13-28 million adults were eligible for statin initiation. Overall, the number of ASCVD events prevented was at least twice as large as the number of incident cases of T2D incurred (LHH range: 2.26-2.90). However, the number of T2D cases incurred surpassed the number of ASCVD events prevented when higher statin-associated T2D RRs were assumed (LHH range: 0.72-0.94). In addition, females (LHH range: 1.74-2.40) and adults aged 40-50 years (LHH range: 1.00-1.14) received lower absolute benefits of statin treatment compared with males (LHH range: 2.55-3.00) and adults aged 70-75 years (LHH range: 3.95-3.96). Projected differences in LHH by age and sex became more pronounced as statin-associated T2D RR increased, with a majority of scenarios projecting LHHs < 1 for females and adults aged 40-50 years. This study's primary limitation was uncertainty in estimates of statin-associated T2D risk, highlighting areas in which additional clinical and public health research is needed.
CONCLUSIONS: Our projections suggest that females and younger adult populations shoulder the highest relative burden of statin-associated T2D risk.

Introduction

Statins are a widely prescribed class of lipid-lowering medication used to prevent atherosclerotic cardiovascular disease (ASCVD) [1, 2]. In 2013, the American College of Cardiology (ACC)/American Heart Association (AHA) updated previous cholesterol treatment guidelines, particularly with respect to ASCVD primary prevention [3]; these guidelines were further revised in 2018 [4]. Previous guidelines emphasized low-density lipoprotein cholesterol (LDL-C) levels for guiding statin treatment [5], while the 2013 ACC/AHA guidelines based statin treatment recommendations on predicted 10-year ASCVD risk. As a result, an estimated 10.4 million U.S. adults were newly eligible for statin treatment for the primary prevention of ASCVD, with adults 60–75 years of age versus other age groups being more likely to be newly eligible [6]. While not formal guidelines, other recommendations have the potential to further expand the population eligible for statin treatment for primary prevention of ASCVD, e.g., expanding guidelines to include populations with predicted 10-year risks of ASCVD of 5% or greater [7, 8].

Experimental and observational research has suggested the potential for adverse effects of statins, including type 2 diabetes (T2D), a side effect of particular interest because of its associated adverse health outcomes and impact on quality of life [9-13]. Accumulating evidence has suggested that statins increase the relative risk (RR) of T2D by 5%–55% [10-15], with potentially elevated statin-associated T2D risk in females [16], younger populations compared to older populations, and populations with lower LDL-C compared to populations with higher LDL-C [15]. Such findings merit further investigation in light of evolving statin recommendations that target growing proportions of populations for ASCVD primary prevention, for whom the net effects of statins remain incompletely quantified [17]. Therefore, this study projected the number of expected ASCVD events prevented and incident cases of T2D incurred in primary prevention populations across three 10-year ASCVD risk-based statin treatment guidelines or recommendations [6, 18, 19].

Methods

Motivation for simulation model

This study has been performed according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist and employed a simulation model to examine intended and unintended consequences of statin treatment guidelines or recommendations through synthesis of high-quality observational and experimental data [20]. A simulation model was conceptualized and then developed in September 2018 to evaluate statin guidelines and recommendations because few available studies (1) were contemporary, (2) spanned ages specified by current guidelines or recommendations, and (3) precisely and validly measured ASCVD and T2D incidence within generalizable male and female multi-ethnic populations with long-term follow-up [21]. This study did not consider costs associated with statins, T2D, or ASCVD.

Data sources and inputs

After conceptualizing the problem, the next step was to identify input data. In an attempt to maximize generalizability, we prioritized studies that included multi-ethnic (non-Hispanic African American and non-Hispanic white; capturing 73% of the U.S. population [22]) male and female statin-eligible adults aged 40–75 years, reflecting the ages specified by the AHA/ACC 2013 and 2018 guidelines [3, 4].

Demographic characteristics, 10-year ASCVD risk, and the number of adults eligible for statin therapy initiation were estimated using the biennial, cross-sectional, and nationally representative National Health and Nutrition Examination Survey (NHANES; waves 2007–2014), pooling 4 waves to ensure sufficient precision [23] (see S1 Text). NHANES collects demographic, nutritional, and health status information on a nationally representative probability sample of the U.S. civilian population instituted by the National Center for Health Statistics. We defined the primary prevention population as 40- to 75-year-old males and females of self-reported non-Hispanic African American or non-Hispanic white race/ethnicity who did not report a doctor or health professional diagnosis of ASCVD, T2D, or type 1 diabetes, who reported never taking cholesterol medications (assessed by self-report and medication inventory), and who had measured fasting LDL-C levels ≤ 190 mg/dL. Race/ethnicity-specific, sex-specific, and antihypertensive-therapy–specific predicted 10-year ASCVD risk for each participant of the primary prevention population was then calculated using the Pooled Cohort Equation (see S1 Text and S1 Table) [24]; note that we restricted our evaluation to statin guidelines or recommendations that used the Pooled Cohort Equation.

In the absence of nationally representative data measuring ASCVD and T2D incidence using validated protocols capturing undiagnosed disease and mortality [25], we leveraged data from the ongoing Reasons for Geographic and Racial Differences in Stroke (REGARDS) study (data used: 2003–2015). The REGARDS study is a contemporary, population-based, longitudinal cohort study designed to evaluate factors underlying the excess stroke mortality in the southeast versus other regions of the U.S. and among African American versus white adults (see S1 Text) [26]. Incident coronary heart disease (CHD) was defined based on medical records, signs and symptoms, diagnostic cardiac enzymes, or electrocardiographic changes consistent with myocardial infarction or a CHD death (S2 Table). Incident stroke was centrally adjudicated by physicians using the World Health Organization definition or by review of final reports from all available neuroimaging studies that were consistent with acute ischemia [27, 28]. Among participants without T2D at baseline who returned to visit 2, T2D incidence was defined by fasting glucose ≥126 mg/dl, non-fasting glucose >200 mg/dl, or use of glucose-lowering medication (S2 Table). Because the REGARDS study only included participants aged ≥45 years and given the comparability of estimated ASCVD and T2D incidence rates between ages 45 and 60, we assigned ASCVD and T2D incidence rates estimated in the 45- to 50-year age group to the 40- to 44-year age group.

Non-ASCVD mortality rates were obtained from the National Center for Health Statistics [29], which compiled death certificates filed in all 50 states and the District of Columbia. Annual non-ASCVD deaths were defined excluding heart disease (International Statistical Classification of Disease, 10th edition codes [ICD-10]: I00–I09, I11, I13, and I20–I51) and cerebrovascular (ICD-10: I60–I69) deaths.

Primary prevention statin-associated ASCVD RRs were obtained from the Cholesterol Treatment Trialists’ meta-analysis of 22 trials (statin treatment versus control), from which we abstracted separate published sex-specific RRs for males (RR = 0.64) and females (RR = 0.84) [30]. In the absence of consistent evidence of association between statins and non-ASCVD mortality [31], the statin–non-ASCVD mortality RR was set to 1.0.

Unlike the statin-ASCVD RRs, for which we only used published estimates from randomized controlled trials (RCTs), we considered meta-analytic estimates from RCTs and observational studies when quantifying statin-associated T2D risk: RR = 1.11 (RCTs only), RR = 1.32 (RCT and observational study pooled), and RR = 1.55 (observational studies only). This decision was motivated by previous reports documenting the potential for industry-funded RCTs to underreport adverse drug reactions [15, 32], the observation that confounding by factors that affect treatment assignment (i.e., “confounding by indication”) is less likely in studies of adverse drug reactions [33], and the average shorter duration (2.6 years) of RCTs compared to observational studies [15].

Assignment of statin treatment was made at study baseline using predicted 10-year ASCVD risk as estimated by the Pooled Cohort Equation [34]. We assigned statin treatment if the predicted 10-year ASCVD risk equaled or exceeded the following 3 statin treatment guidelines or recommendations: ≥10% and the presence of at least one ASCVD risk factor (referred to hereafter as ASCVD risk >10%) [18]; ≥7.5% [3]; and ≥5% [3]. A fourth scenario in which no participant received statin treatment served as our reference group. The ACC/AHA 2018 guidelines were not evaluated due to the unavailability of several ASCVD risk enhancers in the continuous NHANES waves used to inform statin treatment decisions (i.e., persistently elevated LDL-C, persistently elevated triglycerides, preeclampsia, lipoprotein(a) (Lp[a]), ankle-brachial index, and coronary artery calcium were not measured) [4].

Model overview

We contrasted statin-associated ASCVD and T2D incidence across 3 statin treatment guidelines or recommendations using state-transition Markov simulation models [35], which projected annual estimates of T2D and ASCVD incidence based on the previously defined parameters. For each annual cycle, statin-eligible populations could either remain alive and non-diseased (i.e., without ASCVD or T2D) or transition to having T2D, an ASCVD event (fatal or non-fatal), or a non-ASCVD death (S1 Fig). We did not simulate other disease states in addition to non-ASCVD death under our assumption that only ASCVD and T2D incidence were associated with statin use. As we were comparing incidence of T2D and ASCVD, once cohort members had T2D, an ASCVD event or non-ASCVD death, cohort members were removed from the population and were no longer simulated (see example in S1 Fig) [35]. The 1-year cycle was repeated 10 times for each age- (1-year) and sex-specific group and for each intervention scenario to project statin-associated 10-year ASCVD and T2D incidence. As our results project only the incident events for the primary prevention population over 10 years, we cannot make a statement about the population level ASCVD or T2D incidence in 10 years’ time, as this estimate may depend on other factors outside of the scope of our model, e.g., migration patterns or progress in medical technology.

We assumed full implementation of statin treatment recommendations and 100% statin uptake at year 0, after which we assumed a 20% absolute annual decrease through year 4 (S2 Fig) [36]. At year 4, we assumed 20% of adults would continue taking statins through year 10. Given the wide variation in assumed statin adherence found in the literature, we examined statin adherence further in sensitivity analyses.

Using these scenarios, we projected the absolute benefit of statin treatment as the number needed to treat (NNT) to prevent one ASCVD event (1/[statin-associated ASCVD risk difference]), with smaller values indicating a larger absolute risk reduction. The absolute harm of statin treatment was calculated as the number needed to harm (NNH[1/statin-associated T2D risk difference]) to cause one excess incident case of T2D, with larger values indicating that more people needed to receive statin treatment to cause one excess incident case of T2D [37]. In addition, we estimated the likelihood to be helped or harmed (LHH), defined as the ratio of the NNH to NNT; LLH values < 1 indicated that the number of incident cases of T2D incurred exceeded the number of ASCVD events prevented [38]. Statistical analyses were performed using STATA (College Station, TX), TreeAge Healthcare Pro Suite 2018 (TreeAge Software, Williamstown, MA), and SAS (Cary, NC) [39].

Sensitivity analyses

To evaluate possible sources of heterogeneity, we projected the expected benefits and harms of statin treatment by sex and age [18, 40]. To examine the robustness of projections to different adherence patterns, we also projected the expected number of ASCVD events prevented and expected number of excess incident case of T2D incurred under 3 annual relative decrease in statin use of 25% and 50%, as well as full adherence (i.e., 0% decrease) across 10 years [36]. Finally, we examined the robustness of projections to variation in 2 sets of study input parameters (sex-specific statin-ASCVD RRs as well as sex- and age-specific ASCVD and T2D incidence rates) by performing 3 probabilistic sensitivity analyses (PSAs). Our first 2 PSAs considered uncertainty from each input parameter sets separately, and the third PSA considered uncertainty from each set of input parameter simultaneously. For each PSA, we ran 1,000 replications that sampled from the probability distribution of each parameter, which were determined based on 95% confidence intervals (S2 Table). For ASCVD and T2D incidence rates, we assumed a normal distribution for the mean as the estimate. For the statin-ASCVD RR, we assumed a lognormal distribution for the RRs [35]. Because the primary analyses already evaluated the influence of variation in statin-T2D RRs by considering 3 separate RRs, we did not consider this parameter in our PSA.

Results

When weighted to the 2014 non-institutionalized, civilian U.S. population, our primary prevention population consisted of 61,125,042 adults (Table 1). Among the primary prevention population, a majority was female (58.5%) and non-Hispanic white (89.4%), with males and older adults having higher estimated 10-year ASCVD risks compared to females and younger adults (S4 Table). The proportion of the primary prevention population eligible for statin treatment initiation ranged from 21.8% (≥10% ASCVD risk threshold) to 45.6% (≥5.0% ASCVD risk threshold).

Table 1

Comparison of demographic and cardiovascular risk profiles for U.S. white and African American primary prevention populations aged 40–75 years overall and according to 3 statin treatment guidelines or recommendations.

Characteristica	Primary prevention populationb	10-year ASCVD risk
		≥10%c	≥7.5%	≥5.0%
N (% of total population)	61,125,042	13,325,617 (21.8)	18,613,696 (30.5)	27,850,426 (45.6)
Female (%)	35,758,150 (58.5)	4,663,966 (35.0)	6,924,295 (37.2)	115,30,076 (41.4)
Non-Hispanic white (%)	54,645,788 (89.4)	11,486,682 (86.2)	16,026,392 (86.1)	24,341,272 (87.4)
Mean age (SD)	54.7 (9.6)	65.4 (6.9)	63.6 (7.3)	61.4 (7.8)
Mean high-density lipoprotein mg/dL (SD)	55 (14.2)	51 (15.3)	52 (15.3)	51 (15.1)
Mean total cholesterol mg/dL (SD)	204 (39.4)	204 (39.1)	205 (38.9)	207 (38.6)
Mean systolic blood pressure mmHg (SD)	122 (19.6)	134 (18.8)	133 (18.4)	129 (18.0)
On hypertension treatment (%)	209,04764 (34.2)	7,968,719 (59.8)	10,405,056 (55.9)	13,953,063 (50.1)
Current smokers (%)	12,591,759 (20.6)	3,877,755 (29.1)	5,323,517 (28.6)	7,714,568 (27.7)

aWeighted means and proportions.

bAfter excluding prevalent statin users and adults with self-reported ASCVD and T2D and upweighting to the 2014 U.S. population.

cIn the presence of at least one ASCVD risk factor.

Abbreviations: ASCVD, atherosclerotic cardiovascular disease; T2D, type 2 diabetes

As the proportion of adults eligible for statin therapy increased, so did the number of ASCVD events prevented (Fig 1). Over 10 years, the ≥10% ASCVD risk threshold guideline was projected to prevent the fewest ASCVD events (N = 103,009) (Fig 1, panel A), whereas the ≥5.0% ASCVD risk threshold was projected to prevent the largest number of ASCVD events (N = 169,370).

Fig 1

Cumulative number of events of ASCVD and T2D (panels A–C) and LHH (NNH/NNT; panels D–F) associated with 3 statin treatment guidelines or recommendations among a primary prevention population of 61,125,042 eligible U.S. African American and white adults in 2014. The shaded area in panels D–F conveys when the NNH > NNT. ASCVD, atherosclerotic cardiovascular disease; LHH, likelihood to be helped or harmed; NNH, number needed to harm; NNT, number needed to treat; RR, relative risk; T2D, type 2 diabetes.

When assuming a statin-associated T2D risk of 1.11, 10-year NNH projections were consistent across guidelines or recommendations, ranging from 444 to 446 (S3 Fig). These projections suggested that for all statin treatment guidelines or recommendations, the number of ASCVD events prevented was at least twice as large as the number of incident cases of T2D incurred (LHH range 2.26–2.90; NNT range 155–215) (Fig 1, panels A and D; S3 Fig). However, projections of absolute and relative harm were sensitive to the assumed statin-associated T2D RR. When the statin-associated T2D RR was increased to 1.32, NNHs decreased to 198–202 and the relative benefits of statin treatment decreased (LHH range: 1.03–1.30; NNT range 155–209) (Fig 1, panels B and E; S3 Fig). Sensitivity analyses that varied adherence to statin treatment resulted in proportional decreases in the number of ASCVD events prevented and incident cases of T2D incurred (S6 Fig), although results for LHH projections remained consistent.

We next examined the benefits and harms of statin treatment by sex. Across all scenarios, females received lower absolute benefits and incurred a higher relative burden of adverse events compared to males (Figs 2 and S4). The absolute and relative benefits of statin treatment guidelines and recommendations also were more variable in females compared to males. For example, when assuming a statin-T2D RR = 1.11, one ASCVD event was prevented for every 196–254 females treated (LHH range: 1.74–2.40; NNH range 430–478) across the 3 statin treatment guidelines or recommendations. For males, one ASCVD event was prevented for every 110–131 males treated (LHH range: 2.55–3.00; NNH range 331–334) across the 3 statin treatment guidelines or recommendations. When assuming a statin-T2D RR = 1.32, LHH ranged from 0.77 to 1.1 in females but remained above 1 (LHH range: 1.18–1.43) for males. Consistent with our findings, PSA indicated that estimates in females were more uncertain than estimates in males or the total population (S7 Fig).

Fig 2

LHH (NNH/NNT) among females (panels A–C) and males (panels D–F) associated with 3 statin treatment guidelines or recommendations among a primary prevention population of 61,125,042 eligible U.S. African American and white adults in 2014. Shaded area describes when NNH > NNT. ASCVD, atherosclerotic cardiovascular disease; LHH, likelihood to be helped or harmed; NNH, number needed to harm; NNT, number needed to treat; RR, relative risk.

We also examined benefits and harms of statin treatment by age. Regardless of scenario, the absolute and relative benefits of statin treatment were lowest in populations aged 40–50 years and highest in populations aged 71–75 years (Figs 3 and S8). As an example, when assuming a statin-associated T2D RR of 1.11, adults aged 40–50 received the lowest absolute benefits of statin treatment (NNT range: 322–378) and incurred the highest relative burden of adverse events (LHH range: 1.0–1.14) compared to other age groups (Fig 3, panel A). When a statin-associated T2D RR of 1.32 was assumed, adults aged 40–50 continued to receive the lowest absolute benefits of statin treatment (NNT range: 313–367), with the relative burden of adverse events suggesting that every ASCVD event prevented was associated with at least 2 incident cases of T2D (LHH range: 0.43–0.49) (Fig 3, panel E). In contrast, adults aged 71–75 years received the highest absolute benefit (NNT range: 106–107) and the lowest relative burden of adverse events (LHH range: 1.92–1.93), projections that showed very modest differences across statin treatment guidelines or recommendations or assumed statin-associated T2D RR (Fig 3; panel D).

Fig 3

LHH (NNH/NNT) associated with 3 statin treatment guidelines or recommendations among 40–50 (panels A, E, I), 51–60 (panels B, F, J) 61–70 (panels C, G, K), and 71–75 (panels D, H, L) baseline age groups among a primary prevention population of 61,125,042 eligible African American and white U.S. adults in 2014. ASCVD, atherosclerotic cardiovascular disease; LHH, likelihood to be helped or harmed; NNH, number needed to harm; NNT, number needed to treat; RR, relative risk; T2D, type 2 diabetes.

Discussion

In this study, we examined the net effects of statins across 3 treatment guidelines or recommendations in a contemporary, biracial adult primary prevention population. We projected that 13 to 28 million non-Hispanic African American and white adults would be newly eligible for statin treatment, among whom one ASCVD event would be prevented for every 155–197 adults treated. Benefits of statin treatment were even more pronounced in males and older populations compared to females and younger populations. Quantifying harms associated with statin treatment was more complex, as changes in statin-associated T2D risk produced large differences in projected harms. Overall, these results suggest that further efforts are needed to better quantify statin-associated T2D risk across a range of populations, particularly female and younger adult populations for whom statin treatment may introduce a large relative burden of adverse events.

One major challenge for research examining the net effects of statins is comparing intended benefits and unintended harms, which may not be equivalent. Here, the broad clinical spectrum associated with ASCVD that spans a biomarker-diagnosed silent myocardial infarction to a disabling stroke to sudden death would have different implications for patients as well as different clinical courses compared to a new diagnosis of T2D. That being said, as the goal of statin therapy is to prevent ASCVD, and considering that T2D is a clinically important harm associated with statin therapy that has a similar increase in risk as a coronary event for a subsequent ASCVD event [41], this contrast remains a relevant comparison. Particularly of interest are scenarios assuming higher statin-associated T2D RR that compared females versus males and younger versus older populations. In these scenarios, for females and younger populations, the number of incident cases of T2D incurred were often greater, sometimes by several orders of magnitude, then the number of ASCVD events prevented. One way forward may be to incorporate ASCVD and T2D risk prediction models in patient-tailored decision-support tools to select treatment options that balance risks and benefits of statins. Yet, despite the fact that all guidelines emphasize shared decision making, none currently provide the necessary tools [42].

Our projections also suggested that males and older adults received greater benefits from statin treatments compared to females and younger adults. Understanding the risk–benefit profile of statin treatment in younger and female populations is important, given the large proportions of females eligible for statin treatment (35%–41%) as well as uncertainties surrounding adverse effects associated with long-term statin use [3]. Differences between males and females in the net benefits of statin treatment may reflect several factors, including estimates of statin-associated ASCVD risk, which showed a more protective effect in males than females. Heterogeneity by sex in age-specific ASCVD incidence rates also is long-described [43], which could further decrease the net benefits of statin therapy for ASCVD reduction. Interestingly, available studies also support the potential for net harms to be greater in females than males [15, 16, 44, 45]. For example, the Justification for the Use of statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) trial reported a 50% increase in physician-reported T2D in females compared to males, corresponding to an estimated 11 incident T2D diagnoses per 1,000 females taking statins over 1.9 years [16]. However, our simulation assumed a constant statin-associated T2D RR by sex. Additional efforts that enable quantification of the relative and absolute net effects of statin treatment by sex are warranted, particularly in light of disparities in statin treatment by sex [(46], the continued under-representation of female participants in RCTs supported by the US Food and Drug Administration for approval of new molecular entities [46, 47], and the limited number of studies that have examined evidence of heterogeneity in statin-associated T2D risk by sex [30, 48].

In addition, the youngest age groups, which composed 26% of the primary prevention population, did not realize the same statin-associated benefits as the oldest age groups. The importance of research quantifying the net benefits of statins across adulthood is underscored by the fact that statin initiation in younger ages may be associated with decades of statin treatment, despite mixed evidence of long-term effectiveness or greater risk reduction in younger populations [30]. For example, the Anglo-Scandinavian Cardiac Outcomes Trial-Lipid Lowering Arm (ASCOT-LLA) found a statistically significant reduction in ASCVD associated with statins approximately 2 years post follow-up (hazard ratio [HR] = 0.64 [95% CI: 0.53–0.78]) but not 11 years post follow up (HR = 0.89 [95% CI: 0.72–1.11]) [49, 50]. Among additional trials, information on post-trial statin use among those initially randomized to statins or placebo was not always known, further suggesting that the association between long-term, persistent use of statins and adverse events remains incompletely quantified [49, 51].

Our results also underscore the influence of assumed statin-associated T2D risk when quantifying statin net benefit. Although the majority of published studies examining statin-associated harms and benefits leveraged statin-associated T2D RRs from RCTs [17], RCTs may underestimate adverse drug effects due to under-reporting of harms or limited follow-up time [52, 53]. A modest—albeit growing—body of literature also has examined the implications of poor external validity in RCTs, which has the potential to limit transportability of RCT-derived RR estimates to external populations [53]. Specifically, if factors including age, sex, or health characteristics modified the association between statin treatment and T2D, then application of such estimates to populations with different age, sex, or health characteristic has the potential to inaccurately quantify statin-associated T2D (or ASCVD) risk [54]. In this study, RCTs used to quantify statin-associated T2D risk enrolled older (mean age = 63.6 years) and predominantly male (64.1% male) populations compared to the statin-eligible primary prevention population simulated herein (41.5% male, mean age = 54.7 years) or observational studies included in prior meta-analyses (48.5% male, mean age = 57.4 years). Although it remains difficult to anticipate the magnitude by which differences in population characteristics affect estimates of statin-associated T2D risk, prior studies reporting potential heterogeneity by age, sex, and health characteristics support evaluating a range of potential RR estimates rather than relying on a single estimate [10, 15, 55].

Despite many strengths, there are limitations that merit consideration. First, we were unable to examine other adverse events associated with statin use, including rhabdomyolysis [56-58], although the rarity of the event (potentially impacting 546–1,344 adults who were newly eligible for treatment in our study [56]) likely resulted in a very modest underestimate of harm. We also did not consider additional potential benefits of statin treatment, including studies reporting a decreased risk of breast cancer recurrence among females treated with statins [59]. This decision reflected our prioritization of ASCVD, for which the evidence base was the strongest, although future simulations may consider breast cancer recurrence or other potential benefits or harms as evidence accumulates [59-61]. Second, we limited our study to projecting incidence in a primary prevention populations 40–75 years of age of non-Hispanic African Americans or white race/ethnicity given limited input data (e.g., T2D and ASCVD incidence) in other racial/ethnic groups. Although this decision may limit generalizability, 73% of the U.S. population are non-Hispanic African American or white [22]. We also reported projections in the total population and not by race/ethnicity, anticipating imprecise model inputs by race/ethnicity. Future work may wish to expand simulations to include additional populations and evaluate heterogeneity by race, ethnicity, or other potentially modifying factors. Third, our estimate of T2D incidence is based on REGARDS participants returning for the second visit and may be affected by participant attrition. We estimated T2D incidence using REGARDS data because other available sources, e.g., national incidence estimates [62], were based on self-report, which is moderately sensitive and cannot capture the large burden of undiagnosed diabetes in the U.S. [63, 64]. Fourth, 2 model inputs included non-U.S. data: meta-analyzed estimates of statin-T2D RRs and meta-analyzed estimates of the statin-ASCVD RRs. For the former estimate, our previous meta-analysis did not detect significant heterogeneity by country of residence (P > 0.05), supporting pooling. For the latter estimate, we could not identify a meta-analysis that tested for heterogeneity by country of residence, although prior population-based studies conducted before widespread statin use suggested comparable LDL-C levels in European and North American adults. However, the degree to which heterogeneity from variation in the epidemiology of ASCVD by country of residence affects estimates of statin-associated ASCVD RRs cannot be evaluated in depth [65, 66]. Finally, our prioritization of guidelines based on the Pooled Cohort Equation led us to exclude the National Institute for Health and Care Excellence (NICE) [67], the Canadian Cardiovascular Society (CCS) [68], and the European Society of Cardiology (ESC)/European Atherosclerosis Society (EAS) [69] guidelines. However, recent reports have suggested overlap between guidelines, and comparable estimated NNTs have been reported for the CCS, ACC/AHA, and NICE guidelines as well as the U.S. Preventive Services Task Force (USPSTF) and ESC/EAS guidelines [70]. We also did not explicitly consider the ACC/AHA 2018 guideline given data limitations. However, both the 2013 [3] and 2018 [4] ACC/AHA guidelines use the Pooled Cohort Equation with 5% and 7.5% treatment thresholds as starting points for considering statin therapy among adults age 40–75 being evaluated for primary ASCVD prevention. The USPSTF suggests a threshold of 10% for initiating statin therapy for primary prevention [18]. Thus, the 5%, 7.5%, and 10% thresholds for statin eligibility aligns appropriately with current guidelines and physician practice [71] when determining cut-points for initiating statin therapy.

In conclusion, this simulation study adds to a growing body of literature examining the net effects of statins in primary prevention populations. Our results suggest that the highest relative burden of T2D occurred among female and younger adult populations and highlight areas in which additional clinical and public health research is needed.

Estimation of race- and sex-specific ASCVD risk using the ASCVD Pooled Cohort risk equations.

ASCVD, atherosclerotic cardiovascular disease.

(DOCX)

Click here for additional data file.

Model input parameters stratified by 5-year age groups and sex.

(DOCX)

Click here for additional data file.

Markov model parameters.

(DOCX)

Click here for additional data file.

U.S. African American and white primary prevention populations aged 40–75 years according to 10-year ASCVD thresholds stratified by 5-year age groups and sex.

ASCVD, atherosclerotic cardiovascular disease.

(DOCX)

Click here for additional data file.

Markov model conceptual diagram for projections of ASCVD, T2D, and non-ASCVD mortality among an eligible primary prevention population.

Rectangles correspond to disease states, and arrows represent the allowed transitions. Absorbing states are shaded. ASCVD, atherosclerotic cardiovascular disease; T2D, type 2 diabetes.

(TIF)

Click here for additional data file.

Proportion of adults adhering to statin treatment guidelines or recommendations over 10 years among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014.

(TIF)

Click here for additional data file.

NNT or NNH associated with 3 statin treatment guidelines or recommendations among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014.

NNH, number needed to harm; NNT, number needed to treat.

(TIF)

Click here for additional data file.

NNT or NNH among females (panels A–C), and males (panels D–F) associated with 3 statin treatment guidelines or recommendations among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014. NNH, number needed to harm; NNT, number needed to treat.

(TIF)

Click here for additional data file.

Cumulative number of events of ASCVD and T2D among females (panels A–C) and males (panels D–F) associated with 3 statin treatment guidelines or recommendations from a projected population of 61,125,042 eligible U.S. African American and white adults in 2014. ASCVD, atherosclerotic cardiovascular disease; T2D, type 2 diabetes.

(TIF)

Click here for additional data file.

LHH (NNH/NNT) associated with 3 statin treatment guidelines or recommendations among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014.

Grey line describes threshold when NNH > NNT. Statin-T2D RR = 1.11. LLH, likelihood to be helped or harmed; NNH, number needed to harm; NNT, number needed to treat; RR, relative risk; T2D, type 2 diabetes.

(TIF)

Click here for additional data file.

Boxplots of the uncertainty intervals for the LHH (NNH/NNT) associated with the 7.5% ASCVD 10-year risk threshold statin treatment guideline among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014.

Uncertainty was quantified through 3 PSAs: 2 PSAs that considered uncertainty from each of the input parameters separately and a third PSA that considered uncertainty from each input parameter simultaneously. Gray shading indicates the portions of the uncertainty intervals for which the NNH exceeds the NNT. ASCVD, atherosclerotic cardiovascular disease; LLH, likelihood to be helped or harmed; PSA, probabilistic sensitivity analysis; NNH, number needed to harm; NNT, number needed to treat.

(TIF)

Click here for additional data file.

NNT or NNH among 40- to 50-year-olds (panels A–C), 51- to 60-year-olds (panels D–F), 61- to 70-year-olds (panels G–I), and 71- to 75-year-olds (panels J–L) associated with 3 statin treatment guidelines or recommendations among a projected population of 61,125,042 eligible U.S. African American and white adults in 2014. NNH, number needed to harm; NNT, number needed to treat.

(TIF)

Click here for additional data file.

STROBE Statement—Checklist of items that should be included in reports of cohort studies.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

(DOCX)

Click here for additional data file.

28 Apr 2020

Dear Dr. Engeda,

Thank you very much for submitting your manuscript "Projections of incident atherosclerotic cardiovascular disease and incident type 2 diabetes across evolving statin treatment guidelines and recommendations" (PMEDICINE-D-19-04150) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by May 19 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Emma Veitch, PhD

PLOS Medicine

On behalf of Clare Stone, PhD, Acting Chief Editor,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

*Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question, beginning with the main concept. The study design (eg in this case - "modelling study," etc.) should appear in the subtitle (ie, after a colon).

*Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions) -"methods and findings" should be a single subsection. Please also rephrase some of the sections so these are written as full sentences rather than sentence fragments ("to compare.." etc).

*In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

*At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

*Ideally please change the referencing format (this should be simple if referencing software has been used) to change citation callouts in the text to numbers in square brackets (ie, [1, 2] etc). Many thanks

*Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

-----------------------------------------------------------

Comments from the reviewers:

Reviewer #1: The aim of this study by Engeda and colleagues was to investigate (and project) the potential impact of statin use in primary prevention, that is, reducing the risk of atherosclerotic cardiovascular disease (ASCVD) on the one hand, and the potential harm - risk of incident type 2 diabetes mellitus in a population of adult Americans aged 40 - 75 years. A Markov model was developed and used to compare the projected the impact (benefit and harms) on population groups stratified by 10-year ASCVD risk, based on American College of Cardiology (ACC)/ American Heart Association (AHA) guidelines.

Authors report that in the overall cohort, the number of averted ASCVD events outweigh the number incident type 2 diabetes mellitus cases from statin treatment for primary prevention, with a likelihood to help or harm of 2.26 - 2.90. The magnitude of this impact was greater in men and older people as opposed to women and younger people with more modest impact. Their findings are sensitive to the varied relative risks of statin use and risk of type 2 diabetes mellitus. Their findings are interesting as it provides insights to a less-well investigated relationship explored in the same population.

I have a few comments/questions on the methods and reporting of results that require clarification.

Methods

1) Page 6, "The first step in constructing the simulation model was to collect input data." I understand authors are referring to model development when they talk about 'constructing' a simulation model? If that be the case, it appears this does not quite sit well with the International Society for Pharmacoeconomics and Outcomes research (ISPOR) guidelines for good research practices (See: Med Decis Making. 2012 Sep-Oct;32(5):678-89.), which suggest that conceptualization and model development to answer the decision question should precede the search of input data. The model structure should therefore not [initially] be dictated by the range of available data but rather the other way around. Admittedly, this goes through an iterative process before finally adopting the final model. Authors might want to revise this statement to better reflect their process aligns with guidelines?

2) Page 8, "Primary prevention statin-associated ASCVD RRs were obtained from the Cholesterol Treatment Trialists' meta-analysis of 22 trials (statin treatment versus control), from which we abstracted separate RR point estimates for males (RR=0.78) and females (RR=0.84).(28)" I have some questions in relation to this;

a) A good number of the trials included in this meta-analysis were conducted in the UK and Europe. Authors may want to comment on the transferability/applicability of these estimates to the U.S., given there's likely differences in the epidemiology of, and treatment of CVD events between these countries and the U.S.? Was there not pooled evidence from the literature with U.S specific data that could be used instead?

b) Authors purport to have extracted point estimates of relative risks for males and females from this study. Authors should clarify how this was done. In addition, it is unclear if there was any uncertainty e.g. 95% confidence intervals for these estimates. I would expect some uncertainty around the relative risk of events, which needs to be accounted for in the probabilistic sensitivity analysis, to get a comprehensive sense of the risk/impact. Could authors clarify how this was managed?

c) Relative risk of vascular events would vary by age, e.g. differ for 40-50yrs, 51-60yrs, 61-70yrs, etc. Using just a single point estimate of RR is likely to undermine this heterogeneity [at least by age]. In a later part of the results, authors discuss variation by age, but it is unclear or not immediately apparent how this was approached, at least, in terms of the relative risks. Could authors comment on this with respect to their modelling (possible limitation to be discussed)?

3) Page 9, Model overview: "For each annual cycle, statin-eligible populations could either remain alive and non-diseased or transition to having T2D, an ASCVD event, or a non-ASCVD death". Did the health state 'ASCVD event' include fatal and non-fatal cases? Otherwise, how were the fatal ASCVD events modelled?

4) Regarding heterogeneity, it would be interesting to see if there was a differential in the impact of statin treatment on ASCVD events between Black Americans, Hispanic and non-Hispanic whites. In addition, variation in impact by socio-economic status (SES). I wonder if authors considered exploring these (at least in sensitivity analysis) as there are likely to be differences in risk by ethnicity and SES.

5) The paper uses the REGARDS study for input data on T2D and ASCVD incidence. There are some issues with this:

a) Authors should clarify why the data from REGARDS was chosen in preference to other data. For example, there seems to be data available from the Centre for Disease Control (page 19 and reference 58).

b) While REGARDS is a national survey, it is unclear if it is nationally representative, given that the study aim was to include 30% of participants from the "Stroke Belt" states. Please confirm the study is nationally representative.

c) The REGARDS study only included participants from ages 45 years, but the paper starts with populations at age 40 and uses the data from the 45 - 50 age group for the 40 - 45 group and "assessed different specifications via meta analyses". Please clarify what this means.

6) The paper uses 10-year ASCVD risk to determine eligibility for statin treatment. This reflects the 2013 Guidelines but does not reflect the 2018 AHA/ACC Guideline on the Management of Blood Cholesterol (2018 Guidelines), which use other parameters to determine statin eligibility. Further, it is unclear whether the use of 10-year ASCVD risk to determine eligibility for statin treatment reflects current practice. Authors should provide information to demonstrate that the statin eligibility criteria used in the intervention groups reflect current practice. This is important to demonstrate that the population eligible for statins in the intervention groups accurately reflect those people that would be eligible for statins in the population.

7) The current analysis seems not to have performed an uncertainty analysis? I would assume authors would consider uncertainty distributions around key input parameters e.g. relative risk, to allow for a probabilistic sensitivity analysis?

Results and discussion

8) Page 16, paragraph 1: The paper states "one ASCVD event would be prevented for every 155 - 197 adults treated" and "benefits of statin treatment were even more pronounced in males and older populations compared to females and younger populations". These statements rely on the NNB produced by the model. However, the results section does not disclose the NNB - it focuses on the changes in ASCVD events prevented between different guidelines and the changes to LHH between different statin-associated T2D RRs, sex and age groups. The authors should include in the results the NNB, including how this varies by age and sex, in order to support this part of the discussion.

9) Page 20, paragraph 1: The paper states "In addition, we did not consider the 2018 guidelines given data limitations although we anticipate projections of these guidelines would fall between 5% and 7.5% ASCVD risk thresholds we considered, providing some degree of information." This sentence appears incomplete. Authors should clarify what information is provided. For example, do these projections provide information about the risk-benefit analysis under the 2018 Guidelines? The authors should clarify the basis of their anticipated projections - how were they reached? The cited reference (62) does not seem to support the statement and is relevant to the previous point (comparison of other published guidelines to the 2013 Guidelines).

Minor comments and considerations

* The calculation of statin eligibility for the intervention groups relies upon calculation of the 10-year ASCVD risk using the Pooled Cohort Equation. The Equation is used to estimate 10-year ASCVD risk for an individual. Authors explain in the supplementary material (page 3) how they calculate a 10-year ASCVD risk at population level. The authors should clarify that section to make clear how this is achieved and avoid confusion.

* Page 6, end of first paragraph: The reference should be amended to include reference 3 and 4 because the sentence refers to both the 2013 and 2018 Guidelines.

* Page 9, paragraph 2 & page 10, paragraph 2: Can authors provide some justification for their choice of adherence rates? including those in the sensitivity analyses. The cited study (reference 34) aimed to identify factors that predict adherence to statins rather than the adherence rates. However, it does not seem to include data about adherence rates (page 1413 of reference 34). Authors may want to provide clarification on how this reference informed their adherence rates.

* Page 16, paragraph 1: "overall, these results suggest that further efforts are needed to more precisely characterize populations for whom statin treatment may introduce a large burden of adverse effects". Authors should clarify this sentence to provide a more specific conclusion that follows from the preceding sentences. The problem alluded to in the preceding sentence is the lack of a precise estimate of statin-associated T2D risk. If that is correct, a more pertinent conclusion might be that further efforts are required to estimate this risk.

* Page 16, paragraph 2: "particularly of interest are scenarios where the number of incidence cases of T2D incurred were greater than, sometimes by several orders of magnitude, the number of ASCVD events prevented". The author should clarify whether they are referring to scenarios produced by this study and, if so, which scenarios.

* Page 19 - 20: In response to the limitation of not using multiple guidelines, authors state: "recent reports have suggested overlap between guidelines considered herein with other published guidelines. For example, comparable estimated NNTs have been reported" between other published guidelines and those used by the study. To demonstrate the suggested overlap, the authors may want to consider comparing these other published guidelines and the 2013 Guidelines by reference to the populations that would be eligible for statin therapy rather than the effects of statin therapy (i.e. NNTs). For example, to compare the 2018 Guidelines and the 2013 Guidelines, the authors refer to projections of the ASCVD risk thresholds produced by the 2018 Guidelines. Perhaps a similar projection could be made for the other published guidelines.

Supplementary material

1) Authors state, "The definition of stroke in the REGARDS study was: Prevalent stroke was defined as a positive response to either "Were you ever told by a physician that you had a stroke?" or "Were you ever told by a physician you had a mini-stroke or TIA, also known as a transient ischemic attack?" I anticipate that the risk of TIA and the risk of ischaemic stroke are different. Any considerations in this risk differential?

2) In S4 Table, frequencies seem to be in parenthesis and percentage out, contrary to what authors refer to as for example, Male (%). Please, consider adjusting.

References

Reference 15, 28 and 58 and 62 seem incomplete.

-----------------------------------------------------------

Reviewer #2: This is a well conducted study on the projections of incident ASCVD and incident type 2 diabetes using different assumptions of statin treatment recommendations. The study design, datasets, statistical methods and analyses, and presentation (tables and figures) and interpretation of the results are mostly adequate. However, there are still a few issues needing attention.

1) This is a simulation study using the Markov model, popular in the health economics field, which is fine. It's based on many assumptions and also the parameters estimated from the existing datasets. The key question is the believability and robustness of the model, which we don't really know. Are there any validation procedures for the proposed models? In other words, to what degree, should we trust the model that can reflect what's going on in the real world and in the future?

2) Although the Markov model can be used for simulation, it seems a bit simplistic. A few issues remains. Firstly, it's highly likely the model will rely on the projected prevalence of ASCVD and T2D in the next 10 years. Not sure how exactly the authors did it. It seems that they used the REGARDS study data to estimate? Then what prediction models were used for the projection, linear, Poisson or exponential? They are many ways to predict the prevalence using past and future information based on different models. Some good examples can be seen from the latest Global Burden of Disease (GBD) studies. Robust and reliable estimates of future prevalence of the events will have a big impact on the simulation results, and at least this should be discussed in the limitation. Secondly, in the Markov model, for different transition states, what happens if a person is still alive but developed diseases other than ASCVD or T2D later? It seems they were not assigned to any state in the study.

3) The presentation of Table 1 needs to improve and to use standard format. For continuous variables such as age and total cholesterol, they should be summarised as mean and SD if normal distribution, or median and IQR if non-normal distribution. For categorical data, they need to be summarised as count and percentage.

-----------------------------------------------------------

Reviewer #3:

Projections of incident atherosclerotic cardiovascular disease and incident type 2 diabetes across evolving statin treatment guidelines and recommendations

In this study, Dr Engeda and co-workers have estimated the CVD risk reduction from statins used against induced Type 2 diabetes in adults free of prior CVD, eligible to statins use according to contemporary guidelines. Using simulation models (Markov models) they found that the number of incident CVD averted was at least twice as higher the number of incident diabetes induced. Women and younger adults had the lowest absolute benefits and the investigators concluded that these were areas in need for additional clinical and public health research.

While the question investigated has merit, the approach used (simulation), which is essentially based on assumptions, has limitations, many of which are highlighted by the investigators. Where the study really falls short in my view is that the investigators should have included possible scenario to mitigate diabetes risk from statin use. Just like they have used absolute risk model to predict CVD risk, similar models exist to predict incident diabetes risk. Furthermore, lifestyles and other interventions the reduced the unset of diabetes in people at risk are also know. The investigators could therefore consider accounting for the impact interventions to prevent diabetes in those at high risk at baseline, on the future risk of type 2 diabetes in people started on statins. The reality is that, in spite of the authors' findings, chances of withholding statins in eligible patients because of the fear of diabetes risk are very low, and therefore additional focus should be on how to mitigate the risk of diabetes in people started on statin. Other considerations: Incident diabetes was based on history, fasting or random glucose, which is not optimal.

-----------------------------------------------------------

Any attachments provided with reviews can be seen via the following link:

[LINK]

26 May 2020

Submitted filename: engedaResponseMay26_Final_Formatted.docx

Click here for additional data file.

3 Jul 2020

Dear Dr. Engeda,

Thank you very much for re-submitting your manuscript "Projections of incident atherosclerotic cardiovascular disease and incident type 2 diabetes across evolving statin treatment guidelines and recommendations: A modelling study" (PMEDICINE-D-19-04150R1) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

Our publications team (plosmedicine@plos.org) will be in touch shortly about the production requirements for your paper, and the link and deadline for resubmission. DO NOT RESUBMIT BEFORE YOU'VE RECEIVED THE PRODUCTION REQUIREMENTS.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Jul 10 2020 11:59PM.

Sincerely,

Adya Misra, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

Abstract-please add summary demographics for the cohort.

Author summary

Please tone down “We also found that females and the youngest adults received lower absolute benefits of statin treatment when compared to males and the oldest adults” to reflect that this is a finding from a modelling study. I suggest “found” is replaced with “our models suggest”.

The Data statement requires revision, as you say all data are available but then you say “Some data are available from https://www.cdc.gov/nchs/nhanes/index.htm The remaining data cannot be shared publicly because they are from the REGARDS…”. Can you please clarify within the data statement which data can and cannot be accessed freely or otherwise. Please note that authors cannot be the contact persons for data requests and must be deposited with a data committee or an ethics committee or similar. For each data source used in your study:

a) If the data are freely or publicly available, note this and state the location of the data: within the paper, in Supporting Information files, or in a public repository (include the DOI or accession number).

b) If the data are owned by a third party but freely available upon request, please note this and state the owner of the data set and contact information for data requests (web or email address). Note that a study author cannot be the contact person for the data.

c) If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (again, this cannot be a study author).

The link to REGARDS researchers and institutions in the acknowledgements is broken. Please correct as needed.

The four guidelines need to be clearly highlighted in the methods as it currently doesn’t stand out

The data sources are mentioned in suppl methods but brief details in main methods needed

Please can you provide p-values throughout, as needed. Note that we require exact p values, unless p<0.001

Could you replace the word “harm” with an alternative, such as adverse effects or side effects ? The same goes for benefits, perhaps replace with effects?

References should be in Vancouver style please

The STROBE checklist must be called out in the methods and please remove page numbers from the checklist as these are subject to change. Instead please use paragraphs and sections.

Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

Comments from Reviewers:

Reviewer #1: Authors have addressed most of my comments.

Minor correction

In your Author summary, Page 4, you write: "Projected differences by age and sex also were become more pronounced as the effect of statins on T2D was increased". You may want to delete either "were" or "become".

Thank you.

Reviewer #2: Thanks authors for their great effort to improve the manuscript. All my comments were professionally addressed. I am satisfied with the response and the revision. No further issues needing attention.

Any attachments provided with reviews can be seen via the following link:

[LINK]

19 Jul 2020

Submitted filename: engedaResponseJuly16_Final_Formatted.docx

Click here for additional data file.

22 Jul 2020

Dear Dr. Engeda,

On behalf of my colleagues and the academic editor, Dr. Leopold Aminde , I am delighted to inform you that your manuscript entitled "Projections of incident atherosclerotic cardiovascular disease and incident type 2 diabetes across evolving statin treatment guidelines and recommendations: A modelling study" (PMEDICINE-D-19-04150R2) has been accepted for publication in PLOS Medicine.

PRODUCTION PROCESS

Before publication you will see the copyedited word document (in around 1-2 weeks from now) and a PDF galley proof shortly after that. The copyeditor will be in touch shortly before sending you the copyedited Word document. We will make some revisions at the copyediting stage to conform to our general style, and for clarification. When you receive this version you should check and revise it very carefully, including figures, tables, references, and supporting information, because corrections at the next stage (proofs) will be strictly limited to (1) errors in author names or affiliations, (2) errors of scientific fact that would cause misunderstandings to readers, and (3) printer's (introduced) errors.

If you are likely to be away when either this document or the proof is sent, please ensure we have contact information of a second person, as we will need you to respond quickly at each point.

PRESS

A selection of our articles each week are press released by the journal. You will be contacted nearer the time if we are press releasing your article in order to approve the content and check the contact information for journalists is correct. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact.

PROFILE INFORMATION

Now that your manuscript has been accepted, please log into EM and update your profile. Go to https://www.editorialmanager.com/pmedicine, log in, and click on the "Update My Information" link at the top of the page. Please update your user information to ensure an efficient production and billing process.

Thank you again for submitting the manuscript to PLOS Medicine. We look forward to publishing it.

Best wishes,

Adya Misra, PhD

Senior Editor

PLOS Medicine

plosmedicine.org