Premature morbidity and mortality associated with potentially undiagnosed familial hypercholesterolemia in the general population

American Journal of Preventive Cardiology 15 (2023) 100580 Available online 6 September 2023 2666-6677/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Premature morbidity and mortality associated with potentially undiagnosed familial hypercholesterolemia in the general population Kausik K. Ray a , * , Demetris Pillas b , Savvas Hadjiphilippou a , Kamlesh Khunti c , Sreenivasa Rao Kondapally Seshasai d , Antonio J. Vallejo-Vaz a , e , f , David Neasham b , Janet Addison g a Imperial Centre for Cardiovascular Disease Prevention (ICCP), Dept. of Primary Care and Public Health, School of Public Health, Imperial College London, Charing Cross Campus, The Reynolds Building, St Dunstan’s Road, London W6 8RP, United Kingdom b Amgen Ltd, Uxbridge, United Kingdom c University of Leicester, Leicester, United Kingdom d Cardiovascular Clinical Academic Group, Molecular and Clinical Sciences Research Institute, St George’s, University of London and St George’s University Hospitals NHS Foundation Trust, London, United Kingdom e Department of Medicine, Faculty of Medicine, University of Seville, Seville, Spain f Clinical Epidemiology and Vascular Risk, Instituto de Biomedicina de Sevilla, IBiS/Hospital Universitario Virgen del Rocío/Universidad de Sevilla/CSIC. Seville, Spain g Biogen Idec Ltd, Maidenhead, United Kingdom ARTICLE INFO Keywords: Familial hypercholesterolemia Cardiovascular disease prevention LDL-C Screening ABSTRACT Background: Familial hypercholesterolemia (FH) is common, but underdiagnosed, and few systematic early screening programs exist. Objective: To assess health outcomes among those with a recorded diagnosis of FH and potential cases of FH with no recorded diagnosis. Methods: Retrospective cohort study using the UK Clinical Practice Research Datalink. Records of adults were classified as diagnosed FH (FH Coded ), or via accepted algorithms using LDL-C and clinical characteristics as potential FH (FH Potential ) or unlikely FH (FH Unlikely ) using the DLCN or EUROASPIRE criteria (but no record of FH). Outcomes assessed were premature cardiovascular (CV) events, premature deaths and life expectancy. Results: Among 1,729,046 individuals free from CV events, a record of FH Coded before the age of 40 was 0.3/1000 (IQR 0.3–0.4) and increased with age. Where LDL-C levels were available, 1.8/1000 (IQR 1.6–2.0) could be classified as FH Potential . LDL-C was higher for both FH Coded and FH Potential vs FH Unlikely (185.6 and 216.6 vs 116 mg/dL, respectively, p<0.001). Compared to FH Unlikely both FH Coded and FH Potential cohorts had a higher risk of premature cardiovascular events (both p<0.001) with highest rates among FH Coded . Risk of premature deaths did not differ between FH Coded and FH Unlikely, but was 1.88 (95% CI 1.27–2.78, p =0.002) for FH Potential vs FH Coded and 2.40 (95% CI 1.57–3.67, p<0.001) for FH Potential vs FH Unlikely . At age 18, the FH Potential cohort had a life expectancy 16 years lower than the FH Coded cohort (p<0.001). Conclusions: Potential cases of FH had a doubling in risk of premature death and a large reduction in life expectancy compared to individuals with a recorded diagnosis of FH. These findings strengthen the critical importance of identifying potential cases of FH early and early treatment. 1. Introduction FH is an autosomal co-dominant genetic disease mostly due to mutations in the LDL-R, ApoB or PCSK9 genes. It has a prevalence of 1:311 in general populations without founder effects [1] and is associated with a cumulative exposure to elevated low density lipoprotein Abbreviations: FH, Familial Hypercholesterolemia; GP, General Practitioner; CV, Cardiovascular; LDL-C, low density lipoprotein-C; UK, United Kingdom; CPRD, Clinical Practice Research Datalink; HES, Hospital Episodes Statistics; ONS, Office of National Statistics; NICE, National Institute for Health and Care Excellence; DLCN, Dutch Lipid Clinic Network; ICD, International Classification of Diseases; IRRs, Incidence Rate Ratios; HRs, Hazard Ratios; PAR, Population Attributable Risk. * Corresponding author. E-mail address: [email protected] (K.K. Ray). Contents lists available at ScienceDirect American Journal of Preventive Cardiology journal homepage: www.journals.elsevier.com/american-journal-of-preventive-cardiology https://doi.org/10.1016/j.ajpc.2023.100580 Received 17 March 2023; Received in revised form 1 August 2023; Accepted 29 August 2023 American Journal of Preventive Cardiology 15 (2023) 100580 2 cholesterol (LDL-C) levels from birth, resulting in an increased risk of premature cardiovascular disease [2]. Despite these data [3], FH remains under-diagnosed thus representing a major public health challenge [2]. As the cardiovascular consequences of FH are largely silent until its major clinical manifestation with premature cardiovascular mortality and morbidity [4,5], altering its natural history relies upon early detection and initiation of lipid-lowering therapies and behavioral/lifestyle changes [2,6]. That said, there is currently considerable debate about the merits of screening and how best to achieve it, with some advocating universal screening while others proposing cascade screening of index cases [7–11]. Beyond currently available data, the case for screening for FH could be further supported if the natural history of undiagnosed cases of FH (and, by inference, delayed diagnosis/treatment), were shown to adversely impact mortality and morbidity. Hence, we compared health outcomes between adults undiagnosed but potentially having FH, versus those with a recorded diagnosis of FH and those unlikely to have FH. We hypothesized that potential cases of FH in whom there was no recorded diagnosis would have worse health outcomes compared to those in whom a diagnosis of FH was unlikely or those where a diagnosis of FH was present. 1.1. Subjects and methods 1.1.1. Study design and data sources This retrospective cohort study utilized data from UK CPRD - a nationwide, longitudinal, electronic database initiated in 1987 and shown to be representative of the UK population [12]. CPRD houses data for over 10 million individuals, using Read codes as standard terminology, and, where linkages allow, supplemented with additional data from Hospital Episodes Statistics (HES; Supplemental Methods) and mortality data from the Office of National Statistics (ONS; Supplemental Methods). Scientific approval was granted by the CPRD Independent Scientific Advisory Committee. 1.2. Study patients Adults (≥18 years of age) registered in CPRD within the 5-year period from August 1, 2008 (the date of introduction of National Institute for Health and Care Excellence (NICE) guidance on FH diagnosis) [13] to July 31, 2013 were eligible for inclusion. Utilizing all available CPRD data (from date of first CPRD record to July 31, 2013), three cohorts were defined. Supplemental Methods Tables 1 and 2 provide details of variables that were available and allowed for the cohorts to be created. The three cohorts were: (i) FH Coded : individuals with FH diagnosis identified by READ codes (Supplemental Methods Table 1) [14]; (ii) FH Potential : individuals with at least one LDL-C value and no Read-coded FH diagnosis, but who could potentially be considered to have FH based on achieving a ‘Definite’ or ‘Probable’ score calculated according to either the Dutch Lipid Clinic Network (DLCN) (Supplemental Methods Table 2) [2] or EUROASPIRE modified DLCN criteria which accounts for lipid lowering therapy at baseline (Supplemental Methods Table 3) [15], or both; hence individuals who are not on lipid lowering therapy would be captured via the DLCN criteria and those on lipid lowering treatment by EUROASPIRE (iii) FH Unlikely : individuals with at least one LDL-C record and not meeting criteria (i) or (ii). Thus, every person in FH Potential or FH Unlikley category had at least one LDL-C measurement. If a variable contributing to DLCN or EUROASPIRE data was not recorded in CPRD, then we assumed the characteristic to be absent. Potential secondary causes were not specifically excluded. No READ codes specific for homozygous FH were included. We excluded individuals with a prior history of cardiovascular disease at baseline in each of the three cohorts (FH Coded , FH Potential , and FH Unlikely ) to mitigate reverse association bias. Hence, to be defined as FH Potential an individual would require a higher LDL-C level to cross the threshold of being defined as FH Potential . Without this restriction, an individual might cross the threshold for FH Potential with a lower LDL-C level and a positive history of premature cardiovascular disease. Using this approach, we used the FH Coded and FH Potential cohorts to act as proxies for diagnosed and undiagnosed FH, respectively. For the FH Coded cohort, study baseline was defined as the date of the first diagnostic FH Read code record in CPRD. For FH Potential , baseline was set as the date at which a patient attained the threshold classification score according to the algorithm. For FH Unlikely , baseline was set at August 1, 2008. Baseline demographics were extracted from CPRD and HES. 1.3. Outcomes Outcomes of interest included age-specific cardiovascular event rate, premature cardiovascular events, life expectancy, premature mortality and premature cardiovascular mortality occurring during the individual follow-up period from patient entry into CPRD to July 31, 2014. A cardiovascular event was defined as any of the following atherosclerosisrelated outcomes (fatal or non-fatal): unstable angina, acute myocardial infarction, ischemic stroke, transient ischemic attack, coronary revascularization, and peripheral arterial disease, identified by CPRD Read Codes (Supplemental Methods Table 4) and HES ICD-10 codes (HES; Supplemental Methods). Premature cardiovascular events were defined as those occurring prior to 55 and 60 years of age in males and females, respectively, and premature death occurring below the age of 60 in either gender [2]. Cause-specific mortality and date of death were identified in ONS, or in CPRD for patients not linked to ONS. 1.4. Statistical analysis The proportion of patients in CPRD meeting the definitions of FH Coded and FH Potential cohorts was estimated. Age-specific cardiovascular event rates were calculated for the three cohorts after excluding individuals with prior cardiovascular disease at baseline, as described above. Differences in incidence and risk of premature cardiovascular events were estimated through Incidence Rate Ratios (IRRs) and Hazard Ratios (HRs). The IRRs of premature cardiovascular events were calculated for the FH Coded and FH Potential cohorts vs FH Unlikely and for FH Potential vs FH Coded . HRs of experiencing a cardiovascular or a premature cardiovascular event were estimated for FH Coded and FH Potential vs FH Unlikely , for all patients and by age and sex. HRs were estimated using the PHREG command in SAS v.9.4, that fits a superset of the Cox model, known as the multiplicative hazards model. Student t-tests were performed to evaluate between-group differences for continuous variables. Mortality rates for all deaths were ageand sex-standardized to the European Standard Population, and life-expectancy estimates were calculated using the Chiang II method (Supplemental Methods) [16]. The contribution of circulatory-related deaths (ICD-10 Chapter 9) to the mortality rates of the FH Coded and FH Potential cohorts was estimated as a percentage of the mortality rate difference between the two groups. The proportion of the risk of premature death in the FH Coded and FH Potential groups attributable to being classified as FH Potential was calculated (Population Attributable Risk Percentage (PAR%); Supplemental Methods). A complete list of causes of death included under “mortality attributable to circulatory causes” are listed in the supplemental data (Supplemental Methods Table 6) of which the vast majority were related to atherosclerotic cardiovascular disease. To evaluate the extent to which differences in outcomes might be due to imbalances in baseline risk factors, sensitivity analyses excluding patients with a history of diabetes and/or hypertension, and current smokers were conducted. No imputation for missing data was performed. Analyses were conducted using SAS software, version 9.4 (SAS Institute). K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 3 2. Results 2.1. Study population There were 1,729,046 individuals without cardiovascular disease at baseline who fulfilled our inclusion criteria (Supplemental Figure 1). The proportion of individuals in the FH Coded and FH Potential cohorts were 1.3 (95% CI: 1.2–1.3) and 7.8 (95% CI: 7.7–7.9) per thousand, and were lowest below the age of 40 (p<0.001) (Supplemental Table 1). Coding for FH was lowest in Northern Ireland, 0.6 per 1000 (95% CI 0.5–0.8), and highest in Wales at 3.0 per 1000 (95% CI 2.8–3.1). Approximately 45% of the FH Coded and 42% of the FH Potential cohorts were male; the FH Coded cohort were on average 10 years younger and less likely to be smokers, have diabetes or hypertension (p<0.001 for all) vs other cohorts (Table 1). Lipids are shown as mean and standard deviation (SD), or median and interquartile range (IQR). The total number for each baseline characteristic reflects recording of variables in the CPRD database. BMI=body-mass index, ESC=European Society of Cardiology, FH= familial hypercholesterolemia, HDL=high density lipoprotein, LDL= low-density lipoprotein. Of the 1708,744 in the FH Unlikely cohort 5.6% (95,689) could be classified as possible FH, thus the majority could be considered as true FH Unlikely. The FH Potential cohorts were derived from 2873 individuals who fulfilled the DLNC for probable or definite FH and 10 586 individuals using the equivalent EUROASPIRE criteria. Of those considered as FH Potential through the DLCN criteria, the major drivers were high LDL-C (71.8%) followed by family history (26.2%) (Supplemental Table 2). The corresponding figures for those conforming to the EUROASPIRE criteria were high LDL-C in 95.5% and family history in 4.5%. (Supplemental Table 2). The proportion of individuals who could be considered definite by DLCN criteria were 6% and by EUROASPIRE criteria 7.7% respectively, thus the majority of individuals in the FH Potential category would be considered as consisting of probable FH by either criteria (Supplemental Table 3). 2.2. Lipids On average, the FH Potential cohort were 4.3 years (95% CI: 4.0–4.7, p<0.001) older than the FH Coded cohort at time of first recorded LDL-C measurement (Fig. 1A). Among the FH Coded cohort, 25% had a LDL-C record by the age of 45 and 50% by age 55 (Fig. 1B). Among the FH Potential cohort the corresponding figures were 50 and 58 years, respectively. At baseline, recorded LDL-C levels were 185.6 ±58 mg/dL and 216.6 ±50.3 mg/dL for FH Coded and FH Potential , respectively (Table 1.) with FH Unlikely having lower levels (116 ±38.7 mg/dL) (p<0.001). Recorded statin use at baseline varied between groups with 33.3% of FH Coded recording statin use (of which 11.3% was high intensity) versus 67.5% (of which 20.5% was high intensity) among the FH Potential . 2.3. FH and cardiovascular events The incidence of cardiovascular events increased with age for all three cohorts and was highest for FH Coded and lowest for FH Unlikely within each age category (Fig. 2A and Supplemental Table 4). Compared to FH Unlikely , the risk of cardiovascular events was significantly higher for both FH Coded and FH Potential from age 40 onwards (Table 2). Compared with FH Unlikely , the risk of premature cardiovascular events was significantly higher for both FH Coded and FH Potential (HR 4.21, 95% CI 3.69–4.81 and HR 3.46, 95% CI 3.07–3.89 respectively). Of first cardiovascular events, 40.6% occurred by the age of 60 in the FH Coded cohort compared with 29.7% and 18.6% in the FH Potential and FH Unlikely cohorts, respectively (p<0.001) (Fig. 2B). In sensitivity analyses excluding individuals with diabetes, hypertension and current smokers, the risk of premature cardiovascular events was lower among FH Potential 0.62 (95% CI 0.45–0.86) than for FH Coded . However, the risk of premature cardiovascular events among the FH Potential cohort were higher than for FH Unlikely (RR 1.95, 95% CI 1.53–2.48) after excluding individuals with the same risk factors (Supplemental Table 5). Compared to FH Unlikely premature cardiovascular events increased in a graded fashion among the FH Potential cohort with probable FH having a HR of 3.08 (95% CI 2.70–3.52) and definite FH having a HR of 6.50 (95% CI 5.03–8.39), p <0.001 for both (Supplemental Table 6). 2.4. FH and premature mortality The proportion of premature deaths attributable to circulatory causes was similar for FH Coded (30.0%) and FH Potential (31.9%) and higher than in the FH Unlikely cohort (9.5%) (Table 3). Standardized mortality rates were highest among FH Potential and lowest among FH Unlikely , but were not significantly different for FH Coded vs FH Unlikely (p =0.689). In contrast, the risk of premature deaths among the FH Potential was sevenfold higher vs FH Unlikely (RR 7.30, 95% CI 5.26–10.11, p<0.001) and sixfold higher vs FH Coded (RR 6.69, 95% CI 4.89–9.15, p<0.001). Although the elevated risk of premature deaths among FH Potential vs FH Coded was in part related to comorbidities and attenuated considerably in sensitivity analyses which excluded individuals with diabetes, hypertension and smoking at baseline, it remained significant (RR 1.88, 95% CI 1.27–2.78, p =0.002) (Table 3). Corresponding figures for FH Potential vs FH Unlikely were RR 2.40, 95% CI 1.57–3.67, p<0.001. Compared to FH Coded premature deaths increased in a graded fashion among the FH Potential cohort with probable FH having a HR of 2.99 (95% CI 1.73–5.16) and definite FH having a HR of 3.71 (95% CI 1.68–8.18), p Table 1 Baseline Characteristics of Patients Stratified by FH Cohort. FH Coded FH Potential FH Unlikely n =6843 (0.4%) n =13,459 (0.8%) n =1708,744 (98.8%) Demographic Characteristics Age, years, mean (SD) 50.9 (13.7) 60.8 (11.9) 62.2 (15.8) Gender, n (%) Male 3090 (45.2) 5641 (41.9) 818,190 (47.9) Female 3753 (54.8) 7818 (58.1) 890,549 (52.1) Ethnicity, n (%) White 3550 (51.9) 4284 (31.8) 540,432 (31.6) Region, n (%) England 4947 (72.3) 10,250 (76.2) 1,393,334 (81.5) Scotland 480 (7.0) 902 (6.7) 85,854 (5.0) Wales 1325 (19.4) 1587 (11.8) 166,681 (9.8) Northern Ireland 91 (1.3) 720 (5.3) 62,875 (3.7) Clinical Characteristics BMI, n (%) Normal weight (BMI: <25) 1045 (15.3) 2391 (17.8) 391,508 (22.9) Overweight (BMI: 25–29.9) 1479 (21.6) 4532 (33.7) 480,077 (28.1) Obese (BMI: 30+) 964 (14.1) 4031 (30.0) 431,895 (25.3) Smoking, n (%) Never smoked 2185 (31.9) 5781 (43.0) 786,728 (46.0) Previous smoker 983 (14.4) 3547 (26.4) 456,120 (26.7) Current smoker 914 (13.4) 3001 (22.3) 239,269 (14.0) Diabetes, n (%) 120 (1.8) 1890 (14.0) 157,237 (9.2) Hypertension, n (%) 829 (12.1) 4064 (30.2) 299,518 (17.5) Family History 567 (8.3) 1226 (9.1) 54,680 (3.2) Lipids (mg/dL) LDL cholesterol level 185.6 (58) 216.6 (50.3) 116 (38.7) HDL cholesterol 58 (65.7) 61.9 (123.7) 58 (61.9) Triglycerides 186 (IQR115.2–310) 212.6 (IQR 141.7–487.2) 115.2 (IQR) 97.4–132.9) Statin prescription, n (%) 2280 (33.3) 9084 (67.5) 413,776 (23.9) Highintensity statin prescription, n (%) 771 (11.3) 2756 (20.5) 48,428 (2.8) K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 4 <0.001 for both (Supplemental Table 7).The population attributable risk of premature death among the combined cohort with elevated LDLC (FH Coded and FH Potential ) attributed to the FH Potential cohort was 67.53% overall and 55.58% after excluding people with diabetes, hypertension and current smoking (Supplemental Table 8). 2.5. FH and life expectancy Compared with the FH Coded cohort, life expectancy of individuals in the FH Potential cohort was one and a half decades lower (15.9 years, 95% CI:13.4–18.5 years) for those aged 18–19 years, one decade lower (10.1 years, 95% CI:7.9–12.2) for those aged 40–44 years, and approximately half a decade lower (4.5 years, 95% CI:3.0–5.9) for those aged 60–64 years (Fig. 3A and B). There were no significant differences in life expectancy between the FH Coded and FH Unlikely cohorts (Supplemental Figure 2). 3. Discussion A key premise to making a diagnosis of FH is that an LDL-C measurement is required as a minimum. This is supported by the observation the FH Coded group were more likely to have their first LDL-C recorded before the age of 40 compared to the other two cohorts. That said, the proportion of individuals coded as FH before the age of 40 was low and increased with age, possibly reflecting a greater propensity for opportunistic diagnoses during interaction with the healthcare system or Fig. 1. The Timing of First Cholesterol Measurements by Age Across Cohorts. Panel A: Age at first LDL Cholesterol Measurement. Panel B: LDL Cholesterol Measurement by age. On average, the FH Potential cohort were 4.3 years (95% CI: 4.0–4.7, p<0.001) older than the FH Coded cohort at time of first recorded LDL-C measurement. Among the FH Coded cohort, 25% had a LDL-C record by the age of 45 and 50% by age 55. FH=Familial Hypercholesterolemia. K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 5 arising through cascade screening. Furthermore, there was considerable variation in recorded coding of FH across the United Kingdom, with Wales, which has an established screening program recording the highest coding for FH, and Northern Ireland, which lacks a formal strategy, recording the lowest. Taken together, these observations reinforce the role of public health policies on FH detection rates as well as the fact that most members of the general population are unlikely to have their first measurement of LDL-C before the age of 40. In this study, mean baseline LDL-C values in the FH Coded and FH Potential cohorts were considerably higher than the FH Unlikely cohort. Agespecific rates of fatal and non-fatal cardiovascular events increased cumulatively in all three studied cohorts but were generally highest at a given age among the FH Coded then FH Potential cohorts versus FH Unlikely. This is despite the older age and overall higher risk profile of the FH Potential group suggesting that the FH Coded were correctly identified as having the highest prevalence of patients with genetic variants causing FH and is consistent with higher, long-term LDL-C exposure [17]. Moreover the FH Coded group appeared to be somewhat “undertreated” at baseline and had very high event rates reinforcing the need to both diagnosis and better treatment. The contribution of circulatory deaths to overall premature all-cause mortality was similar between FH Potential and FH Coded at around one third of all deaths; in contrast, circulatory deaths contributed to <10% of premature deaths in the FH Unlikely cohort, consistent with exposure to lower life-long levels of LDL-C. Though the FH Potential and FH Coded cohorts were more similar when compared to the FH Unlikely cohort, they differed in that age standardized risk of premature deaths were almost two fold higher among the FH Potential cohort versus FH Coded with differences in survival between FH Potential and FH Coded already apparent by the age of 18. Although the FH Potential group were more likely to be older and have additional cardiovascular risk factors, these risk factors did not appear to explain the excess age standardized premature mortality. The paradoxical findings of less statin use, less intensive statin prescription among FH Coded Vs FH Potential merit discussion. The most likely Fig. 2. The Occurrence of Cardiovascular Events by Age Across Cohorts. Panel A: Age-specific Cardiovascular Event Incidence Rate. Panel B: Age at First Cardiovascular Event. The age specific incidence of fatal and non-fatal cardiovascular events increased with age for all three cohorts and was highest for FH Coded and lowest for FH Unlikely within each age category. Of first cardiovascular events, 40.6% occurred by the age of 60 in the FH Coded cohort compared with 29.7% and 18.6% in the FH Potential and FH Unlikely cohorts, respectively (p<0.001). FH=Familial Hypercholesterolemia. K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 6 reason is that it is a function of the study design and the limitation of capturing changes in care that occurred after what was considered the baseline observation period. The design of this analysis required either a coded diagnosis of FH to create the FH Coded cohort or an LDL-C measurement enabling us to create the FH unlikely or FH Potential groups. This means that the FH Potential patients were older, more likely to have higher LDL-C levels as often the physical signs contributing to the DLCN criteria were not recorded. As they were older they were also more likely to have comorbidities and as a result may have been more likely to receive statins and possibly at a higher dose (bias by indication). Moreover, the baseline observation period may not account for treatment initiation or intensification, as part of shared care with specialist hospital clinics for those with a recorded diagnosis. In the UK, most patients interact with their GP and treatment may not have been initiated by the GP who may have suspected FH before referring to secondary care. If the diagnosis of FH were to be confirmed in secondary care then upon return to the GP, medication would have been initiated post the baseline observation period. Reliable evidence supports the notion that LDL-C exposure is not only a causal risk factor for CVD but its effects are cumulative [18]. As LDL-C cumulative exposure tracks with atherosclerosis burden [19,20] and events [21], it follows that among those in whom treatment is initiated later there is already likely to be a greater underlying burden of atherosclerosis. Therefore, even if treatment is initiated, but occurs later, it may not be enough to mitigate part of the excess risk from the “missed cholesterol years [22].” Moreover, patients with low adherence to high-intensity statins have less LDL-C reduction and higher risk than patients with optimal adherence to low intensity statins [23], reinforcing the potential impact of behavior on outcomes beyond prescriptions alone. Those diagnosed as having FH (FH Coded ) had LDL-C measurements earlier and were thus more likely to have treatment initiated earlier and potentially the first cardiovascular events which occurred, though more frequent, could have been less severe. In contrast, those who met criteria to be considered as potentially having FH, tended to be older and more likely had treatment initiated when older and thus have the potential for a greater burden of underlying atherosclerosis. It follows therefore that though apparently treated more intensively, the “missed cholesterol years” of exposure could contribute to worse survival if the first event that occurs were more likely to be fatal than a non-fatal event. This is consistent with our observation that among the FH Potential group, a graded relationship was observed after separating this group into Probable FH and Definite FH. Despite the significantly higher rates of premature fatal and non-fatal cardiovascular events and higher LDL-C levels, premature mortality rates were not statistically higher in the FH Coded (diagnosed) cohort compared with FH Unlikely . The underlying reasons for this are unclear and could in part reflect survival bias with the sickest individuals having died prior to being eligible for our study as well as those with preexisting cardiovascular disease being excluded (by study design). There was greater use of statins at baseline and fewer cardiovascular risk Table 2 Hazard Ratios of Cardiovascular Events by Age for Coded and Potential FH vs Unlikely, and Overall Cardiovascular events and Premature Cardiovascular Event stratified by Gender. Hazard Ratio Hazard Ratio (FH Coded vs FH Unlikely ) (FH Potential vs FH Unlikely ) Age (years) HR (95% CIs) P-value HR (95% CIs) P-value 18–24 N/A N/A N/A N/A 25–29 N/A N/A N/A N/A 30–34 N/A N/A N/A N/A 35–39 0.80 (0.11, 5.73) 0.827 1.87 (0.26, 13.35) 0.531 40–44 2.99 (1.60, 5.59) <0.001 3.29 (1.56, 6.92) 0.002 45–49 3.02 (2.02, 4.52) <0.001 3.27 (2.19, 4.90) <0.001 50–54 2.64 (1.94, 3.61) <0.001 3.78 (3.00, 4.77) <0.001 55–59 3.01 (2.35, 3.86) <0.001 2.91 (2.45, 3.46) <0.001 60–64 2.53 (2.02, 3.17) <0.001 2.11 (1.79, 2.49) <0.001 65–69 1.97 (1.60, 2.43) <0.001 1.80 (1.56, 2.08) <0.001 70–74 2.12 (1.72, 2.60) <0.001 1.69 (1.47, 1.94) <0.001 75–79 2.17 (1.78, 2.64) <0.001 1.57 (1.36, 1.81) <0.001 80–84 2.31 (1.90, 2.81) <0.001 1.43 (1.23, 1.66) <0.001 85–89 1.90 (1.46, 2.46) <0.001 1.34 (1.11, 1.62) 0.002 All Cardiovascular events All patients 1.97 (1.82, 2.12) <0.001 1.89 (1.79, 1.99) <0.001 Male 2.04 (1.84, 2.26) <0.001 1.84 (1.70, 1.99) <0.001 Female 1.95 (1.74, 2.17) <0.001 2.04 (1.89, 2.19) <0.001 Premature Cardiovascular event All patients 4.21 (3.69, 4.81) <0.001 3.46 (3.07, 3.89) <0.001 Male 4.35 (3.66, 5.18) <0.001 3.37 (2.86, 3.96) <0.001 Female 3.94 (3.20, 4.85) <0.001 3.61 (3.04, 4.28) <0.001 CI=confidence interval, CV=cardiovascular, FH=familial hypercholesterolemia, HR=hazard ratio. Table 3 Standardized Mortality Rates and Relative Risks of Premature Deaths (<60 years). Original Result Sensitivity Analysis I Sensitivity Analysis II Excluding Diabetes and Hypertension Excluding Diabetes, Hypertension, and Smoking Standardized Mortality Rate per 100,000 (95% Cis) Standardized Mortality Rate per 100,000 (95% Cis) Standardized Mortality Rate per 100,000 (95% Cis) Premature Mortality FH Coded 44.6 (26.4, 62.8) 40.1 (16.4, 63.7) 38.3 (13.3, 63.4) FH Potential 298.4 (240.8, 356.0) 125.4 (79.8, 171.1) 72.6 (34.6, 110.6) FH Unlikely 40.9 (40.0, 41.8) 38.2 (37.1, 39.4) 30.2 (29.1, 31.3) Relative Risk (FH Potential vs FH Coded ) 6.69 (4.88, 9.16), p<0.001 3.13 (2.19, 4.46), p<0.001 1.88 (1.27, 2.78), p =0.002 Relative Risk (FH Potential vs FH Unlikely ) 7.30 (5.26, 10.11), p<0.001 3.28 (2.29, 4.72), p<0.001 2.40 (1.57, 3.67), p<0.001 Premature Mortality attributable to circulatory causes Proportion Attributable to Circulatory Causes FH Coded 13.4 (3.5, 23.4) 30.0% FH Potential 95.1 (58.5, 131.6) 31.9% FH Unlikely 3.9 (3.7, 4.0) 9.5% Rates are per 100,000 population, and are ageand gender-standardized to the European Standard Population. Sensitivity analysis excluded patients with a history of diabetes and/or hypertension, and then current smoking. FH=familial hypercholesterolemia, RR=relative risk. K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 7 factors for the FH Coded cohort compared with FH Unlikely . Even though these individuals had higher LDL-C, receiving a diagnosis of FH could result in favorable behavioral and lifestyle changes [24] which may contribute towards attenuating some of the risks related to FH, particularly if the diagnosis is made early and treatment initiated early as suggested by recent studies [22]. In this study, the proportion of individuals with a recorded diagnosis of FH and who were deemed to potentially have FH using an algorithm that utilized LDL-C measurement and other parameters was higher after the age of 40. This is contrary to what is desirable when screening for a condition which exposes an individual to high cholesterol levels from birth. Population attributable risk is used to estimate the proportion of events that could be reduced if that condition were eliminated. Among individuals in our study with elevated LDL-C, over half of premature mortality was attributed to the FH Potential cohort. Within the limitations of study design, our observations could have other practical implications. For example, identification of potential cases of FH through the systematic application of a routine diagnostic algorithm may offer a simple tool to screen individuals prior to more formal assessment such as genetic testing. The DLCN algorithm we applied to classify FH Potential individuals in this study has been shown to predict the presence of FH mutations, with odds ratios for a mutation of 439 and 90 for Definite and Probable scores, respectively [25]. The potential yield from this approach would further augment cascade testing from index cases, currently used in the UK, and help to increase the overall low detection rates for this condition. Furthermore, irrespective of whether a pathogenic mutation was present among the FH Potential group, these individuals represent a high-risk group that may also benefit from early lifestyle changes and risk factor modification [26,27]. Finally, given our findings and the declining costs of biochemical tests, universal cholesterol screening (for instance, by the age of 20 years) may yield greater diagnostic returns in FH detection than current approaches and potentially reduce the premature burden of cardiovascular disease in this population [11] (Fig. 4). The strengths and limitations of our study merit careful consideration. A major strength is the large size and granularity of the CPRD database including risk factor and prescription codes [12], its geographical coverage and thus generalizability, and linkage to mortality and hospital episode statistics. The widely accepted validity of the diagnostic algorithms [28,29] implemented to categorize the FH Potential cohort, and the reliance on elevated LDL-C to achieve the ’Potential’ threshold, gives credence to our classification. Our diagnostic classification was further validated via the restriction to the post mid-2008 timeline (introduction date of NICE guidance on FH-diagnosis). Laboratory test results may fluctuate and tend to regress to the mean; hence, had we opted to rely on limited information pre-2008 (for example, a Fig. 3. Life Expectancy at a Given Age by Cohort. Panel A: Estimated Life Expectancy. Panel B: Estimated Reduction in Life Expectancy. Compared with the FH Coded cohort, life expectancy of individuals in the FH Potential cohort was one and a half decades lower (15.9 years, 95% CI:13.4–18.5 years) for those aged 18–19 years, one decade lower (10.1 years, 95% CI:7.9–12.2) for those aged 40–44 years, and approximately half a decade lower (4.5 years, 95% CI:3.0–5.9) for those aged 60–64 years. There were no significant differences in life expectancy between the FH Coded and FH Unlikely cohorts. FH=Familial Hypercholesterolemia. K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 8 single LDL-C test result), we would likely have introduced bias. For mortality, we only included patients following an LDL-C measurement, regardless of when that happened (pre-2008 or post-2008). We included time after the LDL-C measurement because individuals could not have died before they had their LDL-C test. Immortal time bias was not a confounder, since for all mortality analyses we included time following the first LDL-C measure, regardless of whether it occurred preor post-2008 (this also ensured that person-time would not be included for a period when patients could not have died; the period prior to having their LDL-C record). 3.1. Limitations As with all studies using electronic medical records, the data captured may not be systematic and thus prone to bias. Specifically, information on parameters which make up the Dutch Lipid Network Criteria score, of which LDL-C is a key component, may be absent. A key limitation of our study was the potential under-underestimation of cardiovascular events and mortality, since the diagnosis of FH requires a blood test, and in the absence of universal screening many patients lacked LDL-C measurements and, hence, were not captured in this analysis. Furthermore the gold standard genetic diagnosis was not available in GP records. Our design has practical limitations and though many of the FH Potential cases are likely to truly have had FH some may not. In that regard, we sacrificed specificity for sensitivity and we should acknowledge that perhaps beyond our initial aims these findings show the risk from high LDL-C per se. Moreover, a survival bias may exist for both the FH Potential and FH Coded groups as the sickest patients may have died before cholesterol levels were recorded or a diagnosis was achieved. Furthermore, patients in the FH Coded group may have had their diagnosis, initial LDL-C measurements and initiation/changes in therapy made in tertiary care. Information from this encounter may not have been transcribed into the CPRD record, and thus be missing in our analysis, which may account for the low baseline use of statin. The observational nature of our study cannot entirely exclude the possibility of residual confounding, as a possible alternative explanation for our findings. Although we observed worse outcomes among individuals potentially with FH and no recorded diagnosis vs those with a recorded diagnosis, by design our study cannot elucidate the reasons behind this, nor directly infer causality that making a diagnosis would translate into better outcomes; such conclusions would require specific trials. Finally, our study cannot provide estimates of prevalence of FH. By definition in order to be included, individuals needed a measurement of LDL-C or a coded diagnosis of FH out of all available records (5,677,532). Hence combining FH Coded and FH Potential appears to have a prevalence exceeding the 1:311 reported in systematic reviews as the denominator reflects inclusion criteria for this study rather than the whole population (CPRD dataset) from which this cohort is derived (1,729,046). Similarly, LDL-C measurements contribute to the denominator making up FH Potential or FH Unlikely , which is different to the whole CPRD dataset which includes those without LDL-C measurements. Moreover, those coded as FH may have been coded incorrectly as many also had higher triglyceride levels than might be expected. Further validation of the findings of the present study in other large real world datasets are desirable but may not be easily feasible as ICD codes for FH are not standard practice in many parts of the world or have only recently been implemented.. 4 Conclusion Recorded diagnoses of FH before the age of 40 are low and in the general population LDL-C is largely measured after the age of 40. Compared to individuals with a recorded diagnosis of FH, those individuals potentially having FH but in whom a diagnosis had not been recorded appear to have worse health outcomes, including loss of up to 16 life years by the age of 18. Such individuals account for more than half of all premature deaths among those with hypercholesterolemia. Limitations of the study design notwithstanding, our findings further strengthen the case for screening for FH earlier in life. Sources of funding Amgen-sponsored study. Amgen Ltd provided epidemiological and statistical support analysing the CPRD data in line with the design of the analysis plan by the first and last authors (KKR and JA). Authors employed by the sponsor contributed to the interpretation of the data. The final decision to submit the manuscript lay with the first author KKR. CRediT authorship contribution statement Kausik K. Ray: Conceptualization, Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review & editing. Demetris Pillas: Data curation, Formal analysis, Software. Savvas Hadjiphilippou: Writing – original draft, Writing – review & editing. Kamlesh Khunti: Writing – original draft, Writing – review & editing. Sreenivasa Rao Kondapally Seshasai: Writing – review & editing. Fig. 4. Central Illustration: Familial Hypercholesterolemia screening and diagnosis. Although familial hypercholesterolemia (FH) occurs in ~1:300 individuals, few systematic screening programs exist. As a result, most cases of FH remain undiagnosed resulting in significant missed years of LDL exposure, and atherosclerotic burden. This increases the likelihood of subsequent premature cardiovascular disease and related morbidity and mortality. Patients with potentially undiagnosed FH have a two-fold increased risk of premature deaths with 16 years reduced life expectancy at age 18 compared to those with a confirmed diagnosis of FH. Early diagnosis and treatment is key to reducing the cardiovascular consequences of FH. CVD=cardiovascular disease, FH=Familial Hypercholesterolemia, He=Heterozygous, Ho=Homozygous, LDL=low density lipoprotein. K.K. Ray et al. American Journal of Preventive Cardiology 15 (2023) 100580 9 Antonio J. Vallejo-Vaz: Writing – review & editing. David Neasham: Data curation, Formal analysis, Software. Janet Addison: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Declaration of Competing Interest Professor Kausik K. Ray reports the following; Unrestricted research grants to Imperial College London from Amgen, Daiichi Sankyo, Regeneron, Sanofi, SC, EC or advisory boards honoraria from Novartis, Esperion, Daiichi Sankyo, Abbott, Bayer, Eli Lilly, Silence Therapeutics, CSL Behring, New Amsterdam Pharma, Sanofi, Amgen, Novo Nordisk, BI, Scribe, Vaxxinity, CRISPR, AZ, Kowa, Cargene, Honoria for CME and non CME from Novartis, Novo Nordisk, BI, AZ, Viatris, Daiichi Sankyo, Amgen, Sanofi and stock options PEMI-31. Dr. Demetris Pillas provided consultancy services to Amgen Ltd. Dr. Savvas Hadjiphilippou has no disclosures. Professor Kamlesh Khunti has received research grants from Lilly, Sanofi-Aventis, Boehringer Ingelheim, Merck, Sharpe & Dohme, and Novo Nordisk, has provided consultancy services to Amgen, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly, Servier, and Merck, Sharpe & Dohme, has served in non-remunerative positions of influence at Lilly, Sanofi-Aventis, Merck, Sharpe & Dohme, and Novo Nordisk and has participated in Speakers Bureau for Lilly, Sanofi-Aventis, Merck, Sharpe & Dohme, and Novo Nordisk. Dr. Sreenivasa Rao Kondapally Seshasai has provided consultancy services to Amgen. Dr. Antonio J. Vallejo-Vaz reports current or past participation in research grants to Imperial College London from Pfizer, Amgen, MSD, Sanofi-Aventis, Daiichi Sankyo, and Regeneron; personal fees for consulting from Bayer and Regeneron; and honoraria for lectures from Amgen, Mylan, Akcea, and Ferrer; all outside the submitted work. Dr David Neasham is employed at Amgen Ltd. Ms. Janet Addison was previously employed at Amgen Ltd. Acknowledgments We thank the Amgen CfOR Data & Analytics Center for providing statistical programming support for this study. KKR acknowledges support from the NIHR Imperial BRC. KK acknowledges support from the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care East Midland (CLAHRC East Midlands) and the NIHR Leicester Biomedical Research center. AJVV acknowledges support from the Programme “Beatriz Galindo” from the Ministry of Universities, Spain, and University of Seville, Spain. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ajpc.2023.100580. References [1] Hu P, Dharmayat KI, Stevens CAT, et al. Prevalence of familial hypercholesterolemia among the general population and patients with atherosclerotic cardiovascular disease: a systematic review and meta-analysis. Circulation 2020;141(22):1742–59. [2] Nordestgaard BG, Chapman MJ, Humphries SE, et al. Familial hypercholesterolemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society. Eur Heart J 2013;34:3478–90. a. [3] Wald DS, Bangash FA, Bestwick JP. Prevalence of DNA-confirmed familial hypercholesterolemia in young patients with myocardial infarction. Eur J Intern Med 2015;26:127–30. [4] Benn M, Watts GF, Tybjaerg-Hansen A, Nordestgaard BG. Familial hypercholesterolemia in the Danish general population: prevalence, coronary artery disease, and cholesterol-lowering medication. J Clin Endocrinol Metab 2012; 97:3956–64. [5] Hopkins P. Encouraging appropriate treatment for familial hypercholesterolemia. Clin Lipidol 2010;5:339–54. [6] Versmissen J, Oosterveer DM, Yazdanpanah M, et al. Efficacy of statins in familial hypercholesterolemia: a long term cohort study. BMJ 2008;337:a2423. [7] Goldberg AC, Hopkins PN, Toth PP, et al. Familial hypercholesterolemia: screening, diagnosis and management of pediatric and adult patients: clinical guidance from the National Lipid Association Expert Panel on Familial Hypercholesterolemia. J Clin Lipidol 2011;5:133–40. [8] Kerr M, Pears R, Miedzybrodzka Z, et al. Cost effectiveness of cascade testing for familial hypercholesterolemia, based on data from familial hypercholesterolemia services in the UK. Eur Heart J 2017;38:1832–9. [9] Khera AV, Won HH, Peloso GM, et al. Diagnostic yield and clinical utility of sequencing familial hypercholesterolemia genes in patients with severe hypercholesterolemia. J Am Coll Cardiol 2016;67:2578–89. [10] Umans-Eckenhausen MA, Defesche JC, Sijbrands EJ, Scheerder RL, Kastelein JJ. Review of first 5 years of screening for familial hypercholesterolemia in the Netherlands. Lancet 2001;357:165–8. [11] Wald DS, Bestwick JP, Morris JK, Whyte K, Jenkins L, Wald NJ. Child-parent familial hypercholesterolemia screening in primary care. N Engl J Med 2016;375: 1628–37. [12] Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource profile: clinical practice research datalink (CPRD). Int J Epidemiol 2015;44:827–36. [13] DeMott K, Nherera L, Shaw EJ, et al. Clinical guidelines and evidence review for familial hypercholesterolaemia: the identification and management of adults and children with familial hypercholesterolaemia. London: National Collaborating Centre for Primary Care and Royal College of General Practitioners; 2008. [14] Health and Social Care Information Centre (HSCIC). National health service (NHS). England: Dataset and Business Rules - Cardiovascular Disease Primary Prevention (CVD-PP) Indicator Set (Retired Indicators) (v29.1); 2014. p. 1–10. [15] De Backer G, Besseling J, Chapman J, et al. Prevalence and management of familial hypercholesterolemia in coronary patients: an analysis of EUROASPIRE IV, a study of the European Society of Cardiology. Atherosclerosis 2015;241:169–75. [16] Toson B, Baker A. Life expectancy at birth: methodological options for small populations. Natl Stat Methodol Ser 2003;33. Office for National Statistics. [17] Raal FJ, Santos RD. Homozygous familial hypercholesterolemia: current perspectives on diagnosis and treatment. Atherosclerosis 2012;223:262–8. [18] Ference BA, Ginsberg HN, Graham I, et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel. Eur Heart J 2017 Aug 21;38(32):2459–72. [19] Fern´ andez-Friera L, Fuster V, L´ opez-Melgar B, et al. Normal LDL-cholesterol levels are associated with subclinical atherosclerosis in the absence of risk factors. J Am Coll Cardiol 2017 Dec 19;70(24):2979–91. [20] L´ opez-Melgar B, Fern´ andez-Friera L, Oliva B, García-Ruiz JM, et al. Subclinical atherosclerosis burden by 3D ultrasound in mid-life: the PESA study. J Am Coll Cardiol 2017 Jul 18;70(3):301–13. [21] Domanski MJ, Tian X, Wu CO, et al. Time course of LDL cholesterol exposure and cardiovascular disease event risk. J Am Coll Cardiol 2020 Sep 29;76(13):1507–16. [22] Luirink IK, Wiegman A, Kusters DM, et al. 20-year follow-up of statins in children with familial hypercholesterolemia. N Engl J Med 2019 Oct 17;381(16):1547–56. [23] Khunti K, Danese MD, Kutikova L, et al. Association of a combined measure of adherence and treatment intensity with cardiovascular outcomes in patients with atherosclerosis or other cardiovascular risk factors treated with statins and/or ezetimibe. JAMA Netw Open 2018 Dec 7;1(8):e185554. [24] Khera AV, Emdin CA, Drake I, et al. Genetic risk, adherence to a healthy lifestyle, and coronary disease. N Engl J Med 2016;375:2349–58. [25] Benn M, Watts GF, Tybjaerg-Hansen A, Nordestgaard BG. Mutations causative of familial hypercholesterolemia: screening of 98 098 individuals from the Copenhagen General Population Study estimated a prevalence of 1 in 217. Eur Heart J 2016;37:1384–94. [26] Knowles JW, Rader DJ, Khoury MJ. Cascade screening for familial hypercholesterolemia and the use of genetic testing. JAMA 2017;318:381–2. [27] Gidding SS, Champagne MA, de Ferranti SD, et al. The agenda for familial hypercholesterolemia: a scientific statement from the American Heart Association. Circulation 2015;132:2167–92. [28] National Institute for Health and Care Excellence (NICE). Familial hypercholesterolemia: identification and management. Clin Guideline [CG71] 2017:1–35. [29] Catapano AL, Graham I, De Backer G, et al. 2016 ESC/EAS guidelines for the management of dyslipidaemias: the task force for the management of dyslipidaemias of the European Society of Cardiology (ESC) and European Atherosclerosis Society (EAS) developed with the special contribution of the European Assocciation for Cardiovascular Prevention & Rehabilitation (EACPR). Atherosclerosis 2016;253:281–344. K.K. Ray et al.