Precision medicine in Type 2 Diabetes: Targeting SGLT2-inhibitor Treatment For Kidney Protection
================================================================================================

* Thijs T Jansz
* Katherine G Young
* Rhian Hopkins
* Andrew P McGovern
* Beverley M Shields
* Andrew T Hattersley
* Angus G Jones
* Ewan R Pearson
* Richard A Oram
* John M Dennis
* MASTERMIND Consortium

## Abstract

**Background** Current guidelines recommend sodium-glucose cotransporter-2 inhibitors (SGLT2i) for kidney protection to a broad range of people with type 2 diabetes (T2D), but many were not represented in key kidney outcome trials and have unclear benefit. We aimed to identify which of these people are likely to benefit.

**Methods** We studied 134,420 adults with T2D, estimated glomerular filtration rate (eGFR) ≥20mL/min/1.73m2, no cardiovascular disease or heart failure, starting SGLT2i (34%) or dipeptidyl peptidase-4 inhibitors/sulfonylureas (DPP4i/SU, 66%) in UK primary care (Clinical Practice Research Datalink, 2013-2020). We first validated the hazard ratio (HR) for kidney disease progression (≥50% eGFR decline, end-stage kidney disease, or kidney-related death) from SGLT2i trial meta-analysis. We then integrated this with established prediction models (CKD Prognosis Consortium risk score for 3-year risk of kidney disease progression) to estimate SGLT2i benefit (absolute risk reductions [ARR]) and validated the accuracy of these estimates.

**Findings** The multivariable-adjusted SGLT2i HR for kidney disease progression was 0.60 (95%CI 0.52-0.70) compared to DPP4i/SU, consistent with SGLT2i trial meta-analysis and across eGFR/albuminuria subgroups (interaction p=0.36). Predicted SGLT2i benefit was consistent with observed and was substantial (ARR ≥0.85%) in two subgroups: 1) eGFR <60mL/min/1.73m2 or albuminuria ≥30mg/mmol; 2) eGFR ≥60mL/min/1.73m2, albuminuria 3-30mg/mmol, and predicted ARR ≥80th percentile. Benefit was limited (ARR ≤0.38%) in all others with eGFR ≥60mL/min/1.73m2. This latter group with limited benefit comprises 46% of those recommended SGLT2i for kidney protection.

**Interpretation** SGLT2-inhibitor treatment could be targeted to those with substantial predicted kidney protection benefit. Guidelines should consider stratifying treatment recommendations based on predicted benefit.

## INTRODUCTION

Sodium-glucose cotransporter 2 (SGLT2) inhibitors are now widely recognised for their added benefit in reducing the risk of kidney failure1. Consequently, Kidney Disease: Improving Global Outcomes (KDIGO) and American Diabetes Association (ADA) / European Association for the Study of Diabetes (EASD) guidelines recommend SGLT2-inhibitors to all people with type 2 diabetes with an estimated glomerular filtration rate (eGFR) ≥20mL/min/1.73m2 and chronic kidney disease (CKD), i.e. a reduced eGFR (<60mL/min/1.73m2) or albuminuria (urinary albumin/creatinine ratio [uACR] ≥3mg/mmol)2,3. As a result, up to 40% of people with type 2 diabetes could now be eligible for kidney protection treatment with SGLT2 inhibitors2,4. This widespread eligibility could lead to significant prescription costs and have a substantial impact on health budgets.

Although randomised clinical trials have demonstrated SGLT2-inhibitors are effective for kidney protection in people with type 2 diabetes who have severely increased albuminuria (≥30mg/mmol) or markedly reduced eGFR (20-45mL/min)5-7, their benefit in other groups is less well established. Notably, people with type 2 diabetes and preserved eGFR (≥60mL/min/1.72m2) and low-level albuminuria were not included in key SGLT2-inhibitor kidney outcome trials5-7 or subsequent cost-effectiveness analyses8-10. Data from SGLT2-inhibitor cardiovascular outcome trials suggest this group may only have limited kidney protection benefit11,12, although these data are difficult to interpret as subgroups were not consistently defined across different trials.

This study therefore aimed to identify which people with type 2 diabetes, eligible for treatment with SGLT2-inhibitors for kidney protection under current guidelines, have clinically relevant benefits from this treatment, with a focus on clinically actionable subgroups defined by eGFR and albuminuria. To achieve this, we developed a framework to estimate and validate predictions of absolute risk benefit with SGLT2-inhibitors for kidney protection, integrating relative risk estimates for kidney protection from SGLT2-inhibitor trials with established absolute risk prediction models for kidney disease progression.

## METHODS

### Study design and participants

This cohort study used routine UK primary care data from Clinical Practice Research Datalink Aurum (CPRD). CPRD is broadly representative of the UK population13 and contains patient information recorded during routine general practice, e.g. demographic characteristics, diagnoses and symptoms, prescriptions, laboratory tests, and physiological measurements. It also provides data linkage to information from inpatient secondary care, death registries, and socioeconomic measures. Primary care data were recorded with Read and Systematized Nomenclature of Medicine Clinical Terms codes. Secondary care data with International Classification of Disease (ICD) 10 and Office of Population Censuses and Surveys Classification of Surgical Operation and Procedures (4th revision) codes, and death registry data with ICD-10 codes (code lists available at: [https://github.com/Exeter-Diabetes/CPRD-Codelists](https://github.com/Exeter-Diabetes/CPRD-Codelists)). Approval for CPRD data access and the study protocol was granted by the CPRD Independent Scientific Advisory Committee (eRAP protocol number: 22_002000).

As per our published protocol14, we identified people with type 2 diabetes initiating SGLT2-inhibitors, or as a comparator group dipeptidyl peptidase-4 (DPP4) inhibitors or sulfonylureas (which do not have a drug-specific effect on kidney disease progression15,16), between January 21st, 2013, and October 15th, 2020, with at least 3 months of registration data before treatment initiation. We excluded individuals as per the following criteria: eGFR <20mL/min/1.73m2 or end-stage kidney disease (ESKD), missing baseline eGFR or uACR, concurrent treatment with glucagon-like peptide-1 (GLP1) receptor agonists (given the kidney protection benefit from these drugs17), and additionally body mass index (BMI) <20kg/m2, age <20 years, or >80 years due to limitations in risk score calculation. We also excluded those with a prior diagnosis of heart failure or atherosclerotic vascular disease (ischaemic heart disease or angina, peripheral vascular disease, revascularisation, stroke, or transient ischaemic attack), as these individuals have a clear indication for SGLT2-inhibitors due to cardiovascular benefit18.

### Subgroups

The cohort included individuals with preserved eGFR (≥60mL/min/1.73m2) and normal (<3mg/mmol) or low-level normal albuminuria (uACR 3-30mg/mmol), as well as individuals representative of key kidney outcome trials5-7 as comparator groups, i.e. reduced eGFR at any level of albuminuria (<60mL/min/1.73m2) or severely increased albuminuria with preserved eGFR (uACR ≥30mg/mmol).

### Treatment

We defined treatment as a first ever prescription for an SGLT2-inhibitor, DPP4-inhibitor, or sulfonylurea, with baseline defined as the date of first prescription. Follow-up lasted up to 3 years or until the outcome event, death, deregistration from a participating GP practice, or October 15th, 2020. We censored individuals if they also started a GLP1-receptor agonist. If individuals were treated with a DPP4-inhibitor or sulfonylurea and subsequently started an SGLT2-inhibitor, they were censored and re-entered into the SGLT2-inhibitor arm.

Individuals were not censored if additionally starting a DPP4-inhibitor or sulfonylurea. DPP4-inhibitors and sulfonylureas were considered a single control group due to their similar risk profiles for kidney disease progression (Supplemental Figure 1).

### Outcomes

The primary outcome was kidney disease progression. This was a composite of a sustained ≥50% decline in eGFR, ESKD (requirement for renal replacement therapy or sustained eGFR <15mL/min/1.73m2), and death with any cause listed as kidney-related19. Secondary outcomes included a composite outcome of ≥40% decline in eGFR, ESKD, and death due to kidney-related causes, progression to severely increased albuminuria (≥30mg/mmol), diabetic ketoacidosis, amputation, and mycotic genital infections.

### Covariates

We predefined covariates based on clinical knowledge, which included age (years), sex (male/female), deprivation quintile (index of multiple deprivation), ethnicity (white, black, South Asian, mixed and other or unknown), calendar year at baseline, BMI (kg/m2), HbA1c (mmol/mol), total cholesterol (mmol/L), eGFR (using the Chronic Kidney Disease Epidemiology Collaboration formula, in mL/min/1.73m2)20, uACR (mg/mmol), systolic blood pressure (mmHg), CKD Prognosis Consortium (CKD-PC) risk score for kidney disease progression21, number of current glucose-lowering treatments (1, 2, and 3 or more), concurrent prescriptions for statins, insulin, angiotensin-converting enzyme inhibitors or angiotensin-II receptor blockers, smoking status (non-smoker, ex-smoker, and current smoker), duration of type 2 diabetes at baseline (years), previous diagnosis of hypertension, atrial fibrillation, and hospitalisation in the year before baseline. For physiological or laboratory measurements, the closest value to baseline in the previous 2 years was taken.

### Statistical analyses

#### Baseline clinical features

We present data as mean ±standard deviation, median (interquartile range [IQR]), and number (percentage) for normally distributed, non-normally distributed, and categorical variables respectively. We assumed that missing clinical data were missing at random (blood pressure 0.2%, BMI 3.1% [i.e. either height or weight missing], diabetes duration 5.8%, HbA1c 0.2%, index of multiple deprivation 0.1%, smoking status 0.6%, total cholesterol 0.3%) and imputed these 10 times using multivariate imputations with chained equations. Index of multiple deprivation and smoking status were imputed with multinomial regression, BMI with passive imputation, and all other variables with predictive mean matching. We performed all analyses in each individual imputed dataset and pooled the results according to Rubin’s rules22.

#### Evaluation of the relative benefit for kidney disease progression of SGLT2-inhibitors

We estimated the relative risk reduction of kidney disease progression in our cohort using Cox proportional hazards models, comparing SGLT2-inhibitors versus DPP4-inhibitors/sulfonylureas, and compared estimates to SGLT2-inhibitor trial meta-analyses (0.62, 95% confidence interval [CI] 0.56-0.68)19. We derived hazard ratios (HRs) from three analytical approaches: multivariable adjustment, overlap-weighted analyses with multivariable adjustment23, and inverse probability of treatment-weighted analyses with multivariable adjustment (weights truncated at 2nd and 98th percentiles)24. Weights were derived from a multivariable logistic regression-based propensity score model (excluding calendar year at baseline). We tested consistency of estimates in subgroups defined by eGFR and albuminuria as described previously. In those with preserved eGFR (≥60mL/min/1.73m2), we also evaluated potential treatment effect heterogeneity by continuous baseline CKD-PC risk score, fitted as a non-linear term (5-knot restricted cubic spline) interacting with treatment.

#### Risk score optimisation to estimate absolute kidney disease progression risk

We predicted the 3-year absolute risk of kidney disease progression using the established CKD-PC risk score21 for sustained ≥50% decline in eGFR, ESKD, or death due to kidney-related causes. This risk score uses distinct models for individuals with eGFR <60 and ≥60mL/min/1.73m2. We evaluated both models separately and recalibrated the risk score by fitting a Cox proportional hazards model with the linear predictor as the only variable, calculating a new baseline hazard and calibration slope as appropriate25. We performed internal validation using 500 bootstrap samples. We assessed discrimination using the C-statistic. We assessed prediction accuracy using calibration plots comparing predicted risk with Kaplan-Meier estimates per risk score decile. We assessed overall fit using the Brier score.

#### Derivation and validation of predicted SGLT2-inhibitor absolute benefits

We estimated individual-level absolute SGLT2-inhibitor benefit by integrating the SGLT2-inhibitor relative risk reduction from trial meta-analyses19 (*HR*) with an individual’s predicted kidney disease progression-free survival (*S*(0)*t*) which was calculated as one minus an individual’s predicted absolute risk of kidney disease progression. The predicted benefit or absolute risk reduction (*PARR*) was calculated as follows: *P**ARR* = *S*(0)*t**HR* – *S*(0)*t* . We then assessed the accuracy of predicted benefits by estimating observed SGLT2-inhibitor benefit (i.e. the observed absolute risk reduction) by eGFR and albuminuria subgroup. Observed benefit was estimated as the difference in survival probabilities between SGLT2-inhibitors and DPP4-inhibitors/sulfonylureas at 3 years derived from multivariable-adjusted Cox proportional hazards models, with treatment and covariates included as previously described. We assessed agreement between predicted and observed SGLT2-inhibitor benefit by decile of predicted benefit using calibration plots. We calculated numbers-needed-to-treat (NNT) as the inverse of the absolute benefit. We explored the potential of the CKD-PC risk score to stratify those at lower risk, using a 0.66% predicted ARR as cut-off. This would correspond to a NNT ≤150, with higher numbers less likely to be clinically relevant.

For the secondary outcomes, we estimated HRs with SGLT2-inhibitors versus DPP4-inhibitors/sulfonylureas using Cox proportional hazards models with multivariable adjustments as described above. We estimated the 3-year observed benefits as described above.

### Secondary cohort

To evaluate how many people in the general diabetes population would have clinically relevant SGLT2-inhibitor absolute benefit, we compiled a separate cohort from CPRD including all people with type 2 diabetes on glucose-lowering treatment on the 1st of July 2020 with non-missing eGFR and uACR. We categorised this population into the same subgroups defined by eGFR and albuminuria as the primary study cohort. For subgroups stratified by predicted risk, we assumed a similar distribution of CKD-PC risk scores to the primary cohort. By applying the median SGLT2-inhibitor benefits observed in the primary study cohort to these subgroups, we generalised the probable benefit distribution in the broader diabetes population.

All analyses were conducted with R version 4.3.3 (R Foundation for Statistical Computing, Vienna, Austria). We followed the TRIPOD prediction model guidance.

### Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

## RESULTS

We identified 266,384 adults with type 2 diabetes initiating treatment with SGLT2-inhibitors, DPP4-inhibitors, or sulfonylureas during the study period (356,143 unique drug episodes). Following exclusion of drug episodes where people had established atherosclerotic vascular disease (n=98,668), heart failure (n=6,722), ESKD or eGFR <20mL/min/1.73m2 (n=1,927), concurrent GLP1-receptor agonist prescriptions (n=4,833), BMI <20 kg/m2 or age <20 or age >80 years (n=9,715), or treatment initiation of DPP4-inhibitors/sulfonylureas after initiation of any of the other study drugs (n=22,792), the final cohort included 157,393 unique drug episodes of 134,420 adults with type 2 diabetes initiating SGLT2-inhibitors (55,976 drug episodes), DPP4-inhibitors, and sulfonylureas (101,417 drug episodes). The mean age was 59 ±11 years; 91,721 (58%) were men. Furthermore, 9,153 (6%) had reduced Egfr (<60mL/min/1.73m2), 32,106 (20%) had low-level albuminuria (3-30mg/mmol), and 3,919 (2%) had severely increased albuminuria (≥30mg/mmol). Baseline characteristics by initiated drug class are reported in Table 1, and by eGFR and albuminuria subgroup in Supplemental Table 1.

View this table:
[Table 1.](http://medrxiv.org/content/early/2024/09/02/2024.09.01.24312905/T1)

Table 1. Baseline characteristics by initiated drug class.

### Clinical trial estimates of the relative benefit of SGLT2-inhibitors for kidney protection are generalisable in people with type 2 diabetes

The risk of kidney disease progression was lower with SGLT2-inhibitors compared with DPP4-inhibitors/sulfonylureas (2.6 vs 6.0 per 1000 patient-years), with a HR in our cohort that was numerically similar to previous trial meta-analysis (0.60, 95% CI 0.52-0.70 vs 0.62, 95% CI 0.56-0.68)19, and consistent across analytical approaches (Figure 1). This relative risk reduction was also consistent across eGFR and albuminuria subgroups (p=0.36 for trend, Supplemental Figure 2), and by continuous CKD-PC risk score (p=0.20 for non-linear interaction term, Supplementary Figure 3).

![Figure 1.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/09/02/2024.09.01.24312905/F1.medium.gif)

[Figure 1.](http://medrxiv.org/content/early/2024/09/02/2024.09.01.24312905/F1)

Figure 1. 
Hazard ratios for kidney disease progression (sustained ≥50% decline in eGFR, ESKD, or death due to kidney-related causes) with SGLT2-inhibitors. This forest plot includes hazard ratios from a meta-analysis of randomised controlled trials (Nuffield Group, 2022) and this study cohort (from Clinical Practice Research Datalink) with numerically similar hazard ratios across 3 analytical approaches (multivariable adjustment, overlap weighting, and inverse probability of treatment weighting). The comparator group were sulfonylureas in the meta-analysis of randomised controlled trials and both DPP4-inhibitors and sulfonylureas in this study cohort. Abbreviations: RCT randomised controlled trial; SGLT2i SGLT2-inhibitor; DPP4i/SU DPP4-inhibitor/sulfonylurea; CI confidence interval; CPRD Clinical Practice Research Datalink; IPTW inverse probability of treatment weighting.

### Existing risk scores can predict risk of kidney disease progression with reasonable accuracy in people with type 2 diabetes and preserved eGFR

The CKD-PC risk score for absolute risk of kidney disease progression showed fair discrimination in people with preserved eGFR (≥60mL/min/1.73m2) (C-statistic 0.653, 95%CI 0.648-0.659) but poorer discrimination in people with reduced eGFR (<60mL/min/1.73m2) (C-statistic 0.625, 95%CI 0.615-0.634). As the predicted benefit range was narrow in people with reduced eGFR (<60mL/min/1.73m2) (Supplemental Figure 4), we recalibrated the risk score in people with preserved eGFR only (Supplemental Table 2), after which the risk score showed improved calibration (Supplemental Figure 5).

### SGLT2-inhibitor benefit for kidney disease progression varies markedly across eGFR, albuminuria, and risk score-derived subgroups

To estimate predicted SGLT2-inhibitor benefit, we integrated the validated relative risk reduction with predicted individual-level absolute risk in individuals with preserved eGFR (≥60mL/min/1.73m2). Predicted benefits showed good agreement with observed benefits (Supplemental Figure 6).

As can be seen from Figure 2A and C, SGLT2-inhibitor benefit was limited in individuals with preserved eGFR and normal albuminuria (predicted ARR 0.37%, adjusted observed 0.35%) or low-level albuminuria (predicted ARR 0.36%, adjusted observed 0.38%), except for those with low-level albuminuria and a high CKD-PC risk score (Figure 2B). These individuals, with a predicted benefit above the predefined cut-off of 0.66%, had a CKD-PC risk score of ≥1.75%, corresponding to ≥80th percentile, and had clinically relevant benefit (predicted ARR 0.86%, adjusted observed 0.85%). In contrast, SGLT2-inhibitor benefit was substantial in all individuals with reduced eGFR or severely increased albuminuria (adjusted observed ARR ≥1.21%) who were at high risk of kidney disease progression (Figure 2D).

![Figure 2.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/09/02/2024.09.01.24312905/F2.medium.gif)

[Figure 2.](http://medrxiv.org/content/early/2024/09/02/2024.09.01.24312905/F2)

Figure 2. 
Cumulative incidence curves of risk of kidney disease progression (defined as a sustained ≥50% decline in eGFR, end-stage kidney disease [ESKD], or death due to kidney-related causes) in people with type 2 diabetes by eGFR/albuminuria subgroup. The blue lines represent DPP4-inhibitors/sulfonylureas, the yellow lines represent SGLT2-inhibitors, and the shaded areas represent the 95% confidence intervals. Panel A shows people with eGFR ≥60mL/min/1.73m2, uACR 3-30mg/mmol, and CKD-PC risk score 0-80th percentile, who have limited SGLT2-inhibitor benefit. Panel B shows people with eGFR ≥60mL/min/1.73m2, uACR 3-30mg/mmol, and CKD-PC risk score ≥80th percentile, who have SGLT2-inhibitor benefit. Panel C shows people with eGFR ≥60mL/min/1.73m2 and uACR <3mg/mmol, whom guidelines do not recommend treating with SGLT2-inhibitors for kidney protection. Panel D shows people with eGFR <60mL/min/1.73m2 on the left (not including those with uACR ≥30mg/mmol) and people with uACR ≥30mg/mmol on the right (including those with eGFR <60mL/min/1.73m2), both of whom guidelines recommend treating with SGLT2-inhibitors for kidney protection. Abbreviations: CKD-PC Chronic Kidney Disease Prognosis Consortium.

### Secondary outcomes: SGLT2-inhibitor benefit is similar across different kidney endpoints with greater benefit in high-risk groups, whereas SGLT2-inhibitors are consistently associated with increased risk of diabetic ketoacidosis and mycotic genital infection

SGLT2-inhibitors were also associated with relative risk reductions in 40% decline in eGFR or ESKD (5.8 vs 11.1 per 1000 patient-years), progression to severely increased albuminuria (≥30mg/mmol) (10.3 vs 12.9 per 1000 patient-years), whereas they were associated with relative risk increases in diabetic keto-acidosis (2.4 vs 1.1 per 1000 patient-years) and mycotic genital infection (41.2 vs 12.5 per 1000 patient-years; Supplemental Figure 7). Risk of amputation was not significantly increased with SGLT2-inhibitors (0.8 vs 0.6 per 1000 patient-years). Similar to SGLT2-inhibitor benefit for the primary outcome, the benefit for severely increased albuminuria prevention was less in individuals with preserved eGFR (≥60mL/min/1.73m2) and low-level albuminuria (3-30mg/mmol) and a CKD-PC risk score 0-80th percentile (ARR 0.58%) than in those with a CKD-PC risk score ≥80th percentile (ARR 0.83%). In contrast, absolute risk increases for diabetic ketoacidosis and mycotic genital infections were consistent across subgroups (Supplemental Table 3).

### Population-level impact of targeted prescribing of SGLT2-inhibitors for kidney protection

Using a contemporary population-representative UK primary care cohort of people with type 2 diabetes (n=388,115), we estimate that 20% of all people with type 2 diabetes (n=75,774) would be recommended SGLT2-inhibitor treatment for kidney protection alone under current KDIGO/ADA/EASD guidelines (eGFR ≥20mL/min/1.73m2 and either reduced eGFR <60mL/min/1.73m2 or albuminuria ≥3mg/mmol, without a history of heart failure or atherosclerotic vascular disease)2,3. However, 46% of those recommended treatment represent people with preserved eGFR (≥60mL/min/1.73m2), low-level albuminuria (3-30mg/mmol), and a CKD-PC risk score 0-80th percentile, who would have limited SGLT2-inhibitor kidney protection benefit (NNT 264; Figure 3).

![Figure 3.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/09/02/2024.09.01.24312905/F3.medium.gif)

[Figure 3.](http://medrxiv.org/content/early/2024/09/02/2024.09.01.24312905/F3)

Figure 3. 
Proportion of people with type 2 diabetes recommended for SGLT2-inhibitor treatment for kidney protection and the proportion among these who benefit from it. Panel A shows, in grey on the left, a UK-representative cross-section of people with type 2 diabetes on glucose-lowering medication with data on eGFR and albuminuria as of 01/07/2020, compiled using the Clinical Practice Research Datalink (n=388,115). On the right, in yellow, is the proportion of people that current KDIGO/ADA/EASD guidelines recommend treating with SGLT2-inhibitors for kidney protection alone (eGFR ≥20mL/min/1.73m2 and either reduced eGFR [<60mL/min/1.73m2] or albuminuria [≥3mg/mmol], without a history of heart failure or atherosclerotic vascular disease). In orange are people recommended for SGLT2-inhibitor treatment due to a history of heart failure or atherosclerotic vascular disease, and in grey are people without chronic kidney disease, heart failure, or atherosclerotic vascular disease, whom guidelines do not recommend treating for kidney protection. Panel B shows, in yellow on the left, the subset of people that current KDIGO/ADA/EASD guidelines recommend treating with SGLT2-inhibitors for kidney protection alone (n=75,774). On the right, in grey, is the proportion of people who derive limited benefit (eGFR ≥60mL/min/1.73m2, uACR 3-30mg/mmol, CKD-PC risk score 0-80th percentile), and in orange are people who derive kidney protection benefit from SGLT2-inhibitors (eGFR ≥60mL/min/1.73m2, uACR 3-30mg/mmol, CKD-PC risk score ≥80th percentile; eGFR <60mL/min/1.73m2; and uACR ≥30mg/mmol). As an illustration, the 3-year NNT for kidney disease progression derived from the primary study cohort are displayed. Abbreviations: CKD-PC Chronic Kidney Disease Prognosis Consortium.

## DISCUSSION

Our study shows that while SGLT2-inhibitors consistently reduce the relative risk of kidney disease progression across different subgroups of people with type 2 diabetes, absolute benefits differ markedly across clinically actionable subgroups. Absolute benefit is most pronounced in individuals with reduced eGFR (<60mL/min/1.73m2) or severely increased albuminuria (≥30mg/mmol). In contrast, SGLT2-inhibitor benefit is limited and near-identical to those not recommended SGLT2-inhibitors for kidney protection in 80% of individuals with preserved eGFR (≥60mL/min/1.73m2) and low-level albuminuria (3-30mg/mmol), while making up nearly half of the population recommended treatment by current guidelines. This study demonstrates that those likely to benefit can be identified using routine clinical data, highlighting the potential for stratifying treatment recommendations based on predicted benefit.

The relative risk reductions with SGLT2-inhibitors we found in real-world data are consistent with previous trial data. The overall relative risk reduction for kidney disease progression we observed is comparable to that reported in SGLT2-inhibitor trial meta-analysis26. Analyses of previous trials also support the consistency of relative risk reduction across eGFR and albuminuria subgroups11,12,27. The agreement in findings is likely aided by our use of similar outcome definitions and follow-up durations. On the other hand, the absolute risk reductions in individuals with reduced eGFR or severely increased albuminuria were lower in our data than reported in key kidney outcome trials5-7. This discrepancy is likely due to overrepresentation of individuals with more severe albuminuria and lower eGFR in these trials28. Our cohort, derived from routine general practice data, is more representative of the general diabetes population, which supports the generalisability of our findings.

There is a lack of previous literature on the absolute benefit with SGLT2-inhibitors for people with preserved eGFR and low-level albuminuria. Kidney outcome trials typically did not include these patients5-7, and most cardiovascular outcome trials did not report specifically on eGFR and albuminuria subgroups. DECLARE-TIMI included patients with type 2 diabetes at high risk of or with established atherosclerotic vascular disease, reporting an absolute risk reduction of kidney disease progression of 1.3% after median 4.1 years in those with low-level albuminuria12. However, this group also included individuals with reduced eGFR (<60mL/min/1.73m2). VERTIS-CV included patients with type 2 diabetes, established atherosclerotic vascular disease, and eGFR >30mL/min/1.73m2, reporting an absolute risk reduction of kidney disease progression of 0.6 per 1000 person-years in individuals in the moderate KDIGO risk category (eGFR ≥60mL/min/1.73m2 and low-level albuminuria or eGFR 45-60mL/min/1.73m2 without albuminuria) 111. These heterogeneous findings likely arise from a lack of specific subgroup analysis, highlighting the importance of taking subgroups and individual risk into account to estimate absolute benefit.

In addition to evaluating the relative risk reduction, our study assessed the CKD-PC risk score for kidney disease progression to predict absolute risk. Unlike the Kidney Failure Risk Equation29, it can be calculated for both individuals with preserved and reduced eGFR by using stratified models21. However, our findings indicate poorer discrimination of the CKD-PC risk score for individuals with reduced eGFR. This could reflect demographic and clinical differences between the predominantly US-based cohorts used to develop the score and the UK population in our study. The lack of external validation in non-North American populations suggests that the risk score for people with reduced eGFR may require further recalibration for accurate use in different settings. In these high-risk populations, alternative tools such as the internationally validated Kidney Failure Risk Equation may warrant consideration for predicting adverse kidney outcomes. In our study, the CKD-PC risk score for individuals with preserved eGFR showed fair discrimination (C-statistic 0.653), which is less than reported in the original study (0.781)21. Nevertheless, despite this, predicted benefits using this score were well-calibrated, underscoring its clinical utility within a European population.

Despite broad guideline recommendations for SGLT2 inhibitors in type 2 diabetes, our data suggest that only 20% of individuals with preserved eGFR (>60mL/min/1.73m2) and low-level albuminuria are likely to have a meaningful benefit, reflected in a modest 3-year ARR of 0.85% for preventing kidney disease progression. In comparison, this is lower than the estimated 3-year ARR for statins to prevent major adverse cardiovascular events (1.12%) in individuals with the lowest QRISK2 score (10%) at which treatment is recommended30.

Although the exact threshold at which SGLT2-inhibitor kidney protection treatment would be most cost effective may vary depending on specific circumstances of different countries or healthcare systems, our findings demonstrate that SGLT2-inhibitor benefit can be predicted using the CKD-PC risk score, which could easily be employed in clinical settings to risk-stratify and guide treatment. Employing this stratified treatment approach based on predicted benefit has the potential to reduce unnecessary treatments and optimise use of healthcare resources.

Our study has several strengths. To our knowledge, this is the first study to validate the trial meta-analysis relative risk reduction with SGLT2-inhibitors for kidney disease progression in population data. These results were robust across multiple comprehensive statistical approaches, enhancing their reliability. By integrating these validated summary relative risk estimates with established risk prediction models, our approach provides a higher level of evidence than observational data alone. This makes our absolute risk reduction predictions more robust and generalisable. Finally, our approach to predicting kidney protection benefit relies on generally readily available clinical data, making it easy to implement in routine clinical practice.

Our study has some limitations. First, this study focuses on kidney outcomes only. While further research from our group addresses heart failure outcomes, future studies should aim to integrate kidney, heart failure, and other relevant outcomes to assess overall benefit comprehensively. This may be particularly pertinent for individuals over 80 years old, who were not included in the present analyses. Future research in this population should evaluate their overall benefit while accounting for competing risks. Additionally, follow-up in this study was limited to 3 years, which is consistent with most randomised clinical trials. Nevertheless, interpreting longer follow-up studies could be challenging, particularly due to the difficulty in accounting for changes in patients’ risk profiles over time. This underscores the importance of regularly assessing patients’ risk profiles in routine diabetes care to ensure treatment remains appropriate over time.

In conclusion, our approach enables the targeted use of SGLT2-inhibitors to those likely to have clinically relevant kidney protection benefit. Currently, SGLT2-inhibitor benefit is limited for nearly half of people with type 2 diabetes eligible for kidney protection treatment under existing guidelines. Therefore, guidelines should consider a more stratified approach to treatment recommendations based on predicted absolute SGLT2-inhibitor benefit, which could reduce treatment burden and allow more efficient use of healthcare resources.

## Supporting information

Supplemental Tables [[supplements/312905_file02.docx]](pending:yes)

Supplemental Figures [[supplements/312905_file03.docx]](pending:yes)

## Data Availability

CPRD data are available by application to the CPRD Independent Scientific Advisory Committee. All R code to reproduce the analyses in this paper is available at [https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts](https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts)

[https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts](https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts) 

[https://github.com/Exeter-Diabetes/CPRD-Codelists](https://github.com/Exeter-Diabetes/CPRD-Codelists) 

[https://github.com/Exeter-Diabetes/CPRD-Cohort-scripts](https://github.com/Exeter-Diabetes/CPRD-Cohort-scripts) 

## FUNDING

This research was funded by the Medical Research Council (UK) (MR/N00633X/1), and supported by EFSD/Novo Nordisk. JMD is supported by a Wellcome Trust Early Career award (227070/Z/23/Z). ATH and BMS are supported by the NIHR Exeter Clinical Research Facility; the views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. AGJ was supported by an NIHR Clinician Scientist fellowship (CS-2015-15-018).

## DISCLOSURES

AGJ declares research funding to his university from the UK Medical Research Council, Diabetes UK, JDRF, and the European Foundation for the Study of Diabetes. ERP has received honoraria for speaking from Lilly, Novo Nordisk and Illumina. No industry representatives were involved in the writing of the manuscript or analysis of data. For all authors these are outside the submitted work; there are no other relationships or activities that might bias, or be perceived to bias, their work.

## DATA AVAILABILITY

CPRD data are available by application to the CPRD Independent Scientific Advisory Committee. All R code to reproduce the analyses in this paper are available on [https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts](https://github.com/Exeter-Diabetes/CPRD-Thijs-SGLT2-KF-scripts).

## ACKNOWLEDGMENTS

This article is based in part on data from the Clinical Practice Research Datalink obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. CPRD data are provided by patients and collected by the NHS as part of their care and support.

*   Received September 1, 2024.
*   Revision received September 1, 2024.
*   Accepted September 2, 2024.


*   © 2024, Posted by Cold Spring Harbor Laboratory

This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/)

## REFERENCES

1.  1.McGuire DK, Shih WJ, Cosentino F, et al. Association of SGLT2 Inhibitors With Cardiovascular and Kidney Outcomes in Patients With Type 2 Diabetes: A Meta-analysis. JAMA Cardiol 2021; 6(2): 148–58.
    
    

2.  2.Kidney Disease: Improving Global Outcomes, Diabetes Work Group,. KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int 2022; 102(5S): S1–S127.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.kint.2022.06.008&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=36272764&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

3.  3.Davies MJ, Aroda VR, Collins BS, et al. Management of Hyperglycemia in Type 2 Diabetes, 2022. A Consensus Report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2022; 45(11): 2753–86.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.2337/DCI22-0034&link_type=DOI) 

4.  4.de Boer IH, Rue TC, Hall YN, Heagerty PJ, Weiss NS, Himmelfarb J. Temporal trends in the prevalence of diabetic kidney disease in the United States. JAMA 2011; 305(24): 2532–9.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2011.861&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21693741&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000291860400020&link_type=ISI) 

5.  5.Perkovic V, Jardine MJ, Neal B, et al. Canagliflozin and Renal Outcomes in Type 2 Diabetes and Nephropathy. N Engl J Med 2019; 380(24): 2295–306.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa1811744&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

6.  6.Heerspink HJL, Stefansson BV, Correa-Rotter R, et al. Dapagliflozin in Patients with Chronic Kidney Disease. N Engl J Med 2020; 383(15): 1436–46.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa2024816&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=32970396&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

7.  7.Empa-Kidney Collaborative Group, Herrington WG, Staplin N, et al. Empagliflozin in Patients with Chronic Kidney Disease. N Engl J Med 2023; 388(2): 117–27.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa2204233&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=36331190&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

8.  8.McEwan P, Darlington O, Miller R, et al. Cost-Effectiveness of Dapagliflozin as a Treatment for Chronic Kidney Disease: A Health-Economic Analysis of DAPA-CKD. Clin J Am Soc Nephrol 2022; 17(12): 1730–41.
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoiY2xpbmphc24iO3M6NToicmVzaWQiO3M6MTA6IjE3LzEyLzE3MzAiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wOS8wMi8yMDI0LjA5LjAxLjI0MzEyOTA1LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

9.  9.McEwan P, Foos V, Martin B, Chen J, Evans M. Estimating the value of sodium-glucose cotransporter-2 inhibitors within the context of contemporary guidelines and the totality of evidence. Diabetes Obes Metab 2023; 25(7): 1830–8.
    
    

10. 10.Li X, Hoogenveen R, El Alili M, et al. Cost-Effectiveness of SGLT2 Inhibitors in a Real-World Population: A MICADO Model-Based Analysis Using Routine Data from a GP Registry. Pharmacoeconomics 2023; 41(10): 1249–62.
    
    

11. 11.Cherney DZI, Charbonnel B, Cosentino F, et al. Effects of ertugliflozin on kidney composite outcomes, renal function and albuminuria in patients with type 2 diabetes mellitus: an analysis from the randomised VERTIS CV trial. Diabetologia 2021; 64(6): 1256–67.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s00125-021-05407-5&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=33665685&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

12. 12.Mosenzon O, Wiviott SD, Heerspink HJL, et al. The Effect of Dapagliflozin on Albuminuria in DECLARE-TIMI 58. Diabetes Care 2021; 44(8): 1805–15.
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiZGlhY2FyZSI7czo1OiJyZXNpZCI7czo5OiI0NC84LzE4MDUiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wOS8wMi8yMDI0LjA5LjAxLjI0MzEyOTA1LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

13. 13.Wolf A, Dedman D, Campbell J, et al. Data resource profile: Clinical Practice Research Datalink (CPRD) Aurum. Int J Epidemiol 2019; 48(6): 1740–g.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ije/dyz034&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

14. 14.Rodgers LR, Weedon MN, Henley WE, Hattersley AT, Shields BM. Cohort profile for the MASTERMIND study: using the Clinical Practice Research Datalink (CPRD) to investigate stratification of response to treatment in patients with type 2 diabetes. BMJ Open 2017; 7(10): e017989.
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiYm1qb3BlbiI7czo1OiJyZXNpZCI7czoxMjoiNy8xMC9lMDE3OTg5IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDkvMDIvMjAyNC4wOS4wMS4yNDMxMjkwNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

15. 15.Rosenstock J, Perkovic V, Johansen OE, et al. Effect of Linagliptin vs Placebo on Major Cardiovascular Events in Adults With Type 2 Diabetes and High Cardiovascular and Renal Risk: The CARMELINA Randomized Clinical Trial. JAMA 2019; 321(1): 69–79.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2018.18269&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30418475&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

16. 16.Rosenstock J, Kahn SE, Johansen OE, et al. Effect of Linagliptin vs Glimepiride on Major Adverse Cardiovascular Outcomes in Patients With Type 2 Diabetes: The CAROLINA Randomized Clinical Trial. JAMA 2019; 322(12): 1155–66.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2019.13772&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=31536101&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

17. 17.Perkovic V, Tuttle KR, Rossing P, et al. Effects of Semaglutide on Chronic Kidney Disease in Patients with Type 2 Diabetes. N Engl J Med 2024; 391(2): 109–21.
    
    

18. 18.Zelniker TA, Wiviott SD, Raz I, et al. SGLT2 inhibitors for primary and secondary prevention of cardiovascular and renal outcomes in type 2 diabetes: a systematic review and meta-analysis of cardiovascular outcome trials. Lancet 2019; 393(10166): 31–9.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(18)32590-X&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30424892&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

19. 19.Nuffield Department of Population Health Renal Studies Group, SGLT2 inhibitor Meta-Analysis Cardio-Renal Trialists’ Consortium Impact of diabetes on the effects of sodium glucose co-transporter-2 inhibitors on kidney outcomes: collaborative meta-analysis of large placebo-controlled trials. Lancet 2022; 400(10365): 1788–801.
    
    

20. 20.Inker LA, Eneanya ND, Coresh J, et al. New Creatinine- and Cystatin C-Based Equations to Estimate GFR without Race. N Engl J Med 2021; 385(19): 1737–49.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access\_num=10.1056/NEJMOA2102953/SUPPL_FILE/NEJMOA2102953_DISCLOSURES.PDF&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

21. 21.Grams ME, Brunskill NJ, Ballew SH, et al. Development and Validation of Prediction Models of Adverse Kidney Outcomes in the Population With and Without Diabetes. Diabetes Care 2022; 45(9): 2055–63.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.2337/dc22-0698&link_type=DOI) 

22. 22.Rubin DB, Schenker N. Multiple Imputation for Interval Estimation From Simple Random Samples With Ignorable Nonresponse. Journal of the American Statistical Association 1986; 81(394): 366–74.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.2307/2289225&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=WOS:A1986C64&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1986C648000012&link_type=ISI) 

23. 23.Thomas LE, Li F, Pencina MJ. Overlap Weighting: A Propensity Score Method That Mimics Attributes of a Randomized Clinical Trial. JAMA 2020; 323(23): 2417–8.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

24. 24.Chesnaye NC, Stel VS, Tripepi G, et al. An introduction to inverse probability of treatment weighting in observational research. Clin Kidney J 2022; 15(1): 14–20.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ckj/sfab158&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

25. 25.Steyerberg E. Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating; 2009.
    
    

26. 26.Natale P, Tunnicliffe DJ, Toyama T, et al. Sodium-glucose co-transporter protein 2 (SGLT2) inhibitors for people with chronic kidney disease and diabetes. Cochrane Database Syst Rev 2024; 5(5): CD015588.
    
    

27. 27.Zannad F, Ferreira JP, Pocock SJ, et al. Cardiac and Kidney Benefits of Empagliflozin in Heart Failure Across the Spectrum of Kidney Function: Insights From EMPEROR-Reduced. Circulation 2021; 143(4): 310–21.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

28. 28.Forbes AK, Hinton W, Feher MD, et al. A Comparison of Sodium-Glucose Co-Transporter 2 Inhibitor Kidney Outcome Trial Participants with a Real-World Chronic Kidney Disease Primary Care Population. Nephrol Dial Transplant 2024.
    
    

29. 29.Tangri N, Grams ME, Levey AS, et al. Multinational Assessment of Accuracy of Equations for Predicting Risk of Kidney Failure: A Meta-analysis. JAMA 2016; 315(2): 164–74.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2015.18202&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26757465&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F09%2F02%2F2024.09.01.24312905.atom) 

30. 30.National Institute for Health and Care Excellence (NICE) Cardiovascular disease: risk assessment and reduction, including lipid modification (NG238). 2023.