We conducted a retrospective cohort study to enable longitudinal analysis of all LDL-C measurements among eligible KPNC adult members after a qualifying index nonfatal MACE event between 01/01/2012 and 12/31/2022. KPNC is an integrated health care delivery system that provides care to over 4.6 million people. The KPNC coverage area includes 23 counties in the San Francisco Bay Area and the Central Valley of California, from Sacramento to Fresno. The KPNC membership is ethnically and socio-economically diverse and is broadly a representative of the Northern California population except the extremes of income distribution. A series of sequential exclusions were applied to maximize accurate identification of patients with MACE and robust follow-up for LDL-C monitoring and clinical events (see Figure 1 A). The initial cohort of 84,898 patients included many individuals with type 2 MI (demand ischemia) and other nonobstructive CAD phenotypes such as Takotsubo syndrome, necessitating exclusions to narrow the cohort to the correct patient population. The final sample for analysis comprised 47,416 individuals. Follow-up for clinical outcomes after the initial qualifying event extended through 12/31/2024 (see Figure 1 B). The study was approved by the Kaiser Foundation Research Institute Institutional Review Board, and the requirement of informed consent was waived as this was a data-only study.

Flow diagram of cohort derivation (A); Study design (B).

KPNC’s Virtual Data Warehouse and KPNC electronic medical record (HealthConnect®) served as the primary data sources for patient identification, characterization and follow-up for clinical outcomes. ^, The Virtual Data Warehouse is a standardized data resource comprised of electronic datasets that is populated with linked demographic, administrative, outpatient pharmacy, laboratory test results, and health care utilization (ie, ambulatory visits and network and non-network hospitalizations with diagnoses and procedures) and vital status (ie, mortality) data for all KPNC members.

Baseline covariates were ascertained at the time of the index MACE going back up to 5 years. These included age, sex, race/ethnicity, educational attainment, household income, neighborhood deprivation index (NDI), smoking status, body mass index, nine common comorbidities, engagement with the Cardiac Rehabilitation Program, limited English proficiency (as a proxy to language barrier) and distance to the closest KP facility. Educational attainment, household income and the NDI were obtained from the 2020 US Census Demographic and Housing Characteristics File at the Census Block Group level. LDL-C (estimated using the Friedewald equation or with direct method when triglycerides exceed 400 mg/dL) was measured at a Clinical Laboratory Improvement Amendments (CLIA)-certified KP Regional Laboratory (Berkeley, CA). The KPNC Cardiac Rehabilitation Program is offered to all members after confirming eligibility and consists of 6 visits, typically virtual. Prescription of cardiac medications at index event discharge (including statins, ezetimibe, PCSK9 inhibitors, other lipid lowering drugs, beta blockers, ACE inhibitors, calcium channel blockers, diuretics, hypoglycemics, angiotensin-II receptor blockers, antiarrhythmics, and anticoagulants/antiplatelets) were ascertained using the Pharmacy Information Management System that captures prescription orders, timing, and dose in real time. As measures of compliance with lipid lowering therapy (LLT), we derived medication possession ratios for cholesterol lowering drugs at 12 months after index date and for the entire follow-up period as days-supply divided by follow-up days with right censoring for death or discontinuation of health plan membership.

The diagnostic and procedure codes that defined the study cohort, comorbidities and clinical outcomes are provided in Supplemental Table I. Prevalent diabetes was ascertained by cross referencing the Kaiser Permanente Division of Research (KP-DOR) Diabetes Registry ^, and/or use of hypoglycemics, and prevalent cancer using the KP-DOR Cancer Registry which contributes data to the National Cancer Data Base . The Virtual Data Warehouse has complimentary information on deaths, including in-hospital deaths and out-of-hospital deaths through the State of California and Social Security Administration files.

For Aim 1, we generated descriptive statistics (mean and standard deviation for continuous variables and frequency distributions for categorical variables) and estimated the proportions attaining LDL-C target of 70 mg/dL or not by year of follow-up from 1 to 10 years after the index event. This was done among those with LDL-C measured available and excluding from the denominator those who died or discontinued KP membership in the corresponding year. When there were multiple LDL-C measures in one year, we selected the lowest LDL-C value. Next, we described the time trajectories over 10 years of attaining LDL-C target goal by sex, race/ethnicity and degree of engagement with the Cardiac Rehabilitation Program. Also, to assess temporal trend, we estimated target LDL-C attainment in the year following the index event by calendar time.

For Aim 2, we performed statistical modeling to identify multivariable predictors of being above LDL-C target of 70 mg/dL. Odds ratios of being above LDL-C target were estimated using a multivariable logistic model with simultaneous consideration of selected baseline characteristics.

We did not include education and income because they are already included in the NDI. The same multivariable logistic approach (without the lipid lowering compliance measure) was followed to generate the individual independent correlates of lacking LDL-C testing during the first year post index date.

For Aim 3, we implemented two approaches. The first focused on average LDL-C measures between index date and right censoring (this was done separately for ASCVD and composite CVD adverse outcomes) and then categorizing cohort members into <70, 70-99, 100-129, 130-159 and ≥160 mg/dL, respectively. Sensitivity analysis was done with alternative cut-points of <55, 55-69, 70-99, 100-129, 130-159 and ≥160 mg/dL, respectively. The second approach consisted of dichotomizing average LDL-C levels between index date and right censoring into achieving target of <70 mg/dL or not. Sensitivity analysis was done with alternative cut-point of 55 mg/dL. We first estimate age-adjusted rates of all adverse clinical outcomes (ASCVD, composite CVD) per 10,000 person-years in each of the LDL-C or target achievement groups. The period of observation for events was index event though event of interest, discontinuation of KP membership or death, whichever occurred first. Time-to event modeling was done with Cox proportional hazards models yielding hazard ratios (HR) and 95% confidence intervals adjusting for age, sex and race/ethnicity, NDI, smoking, BMI, nine comorbidities, degree of engagement with the Cardiac Rehabilitation Program, limited English proficiency and distance to the nearest Kaiser Permanente facility. Finally, to assess clinical utility and the potential benefit of a theoretical intervention to bring the cohort into achieving LDL-C target of <70 mg/dL, we calculated the number needed to treat (NNT) to prevent one ASCVD and one composite CVD adverse event. All data management and statistical analyses were performed using SAS 9.4.

Cohort characteristics at baseline (ie, index MACE event) are provided in Table I . Mean (SD) age was 66 (12) years, and 72% of the cohort was male. The race/ethnic distribution was as follows: 58% white, 6% African American, 19% Asian/Pacific Islander, 13% Hispanic/Latino and 3.5% other/mixed/unknown race. In terms of socioeconomic characteristics, about 29% (assigning the % education attainment of the census block to the person) had less than college education and 39% lived on census blocks where the average annual income level was below 75K. Whereas 12% were current smokers, 38% had a body mass index in the obesity range. In the 5 years preindex event (when multiple LDL-C measures were available, the closest to the index event was selected), 22% had LDL-C below 70 mg/dL, 14% had LDL-C in the 130-159 mg/dL range and 9% an LDL-C ≥ 160 mg/dL. The most common comorbidities were hypertension (70%), chronic kidney disease (22%), peripheral vascular disease (17%), iron deficiency anemia (15%), diabetes mellitus and atrial fibrillation (both 10%). About 66% enrolled in the Cardiac Rehabilitation Program, with 30% attending less than 6 visits, and 35% completing 6 visits or more. Seven percent of study participants reported limited English proficiency. While 9% of cohort members lived less than 10 miles from the nearest KP facility, 67% lived 10 to 50 miles away and 20% lived 50 or more miles away. About 84% were prescribed high intensity statin treatment, 14% moderate intensity statin treatment, and 2% low intensity statin treatment. By contrast, very few received Ezetimibe (2%), PCSK9 inhibitors (0.2%) or other cholesterol lowering drugs (3%). Beta blockers were prescribed to 94% and anticoagulants/antiplatelets to 95% (some aspirin prescriptions may have been filled over-the-counter and potentially not captured in our database). Other cardiac medications ranged from antiarrhythmics (15%) to ACE inhibitors (56%).

For the entire cohort, the medication possession ratio for lipid lowering therapy was 82% in Year 1 and 65% for the full follow-up period (Supplemental Table IIA). In Year 1 the medication possession ratio tended to be greater in men (83%) than in women (80%) ( P <.001), and in white (84%) and Asian/Pacific Islander (82%) subjects than in African American (76%) or Hispanic/Latino subjects (79%) ( P <.001) and the same pattern was discernable for the full follow-up period. Also in Year 1, the medication possession ratio was higher among those who completed the Cardiac Rehabilitation Program (86%) than among those with partial completion (81%) or no attendance (78%). The proportion attaining LDL-C target goal of 70 mg/dL increased with increasing compliance with lipid lowering therapy, from 43% when the medication possession ratio was 0% to 25% to 71% when the medication possession ratio was ≥90% (Supplemental Table IIB).

The proportions lacking LDL-C testing by follow-up years (overall and in sub-groups) are summarized in Supplemental Table III. In the entire cohort, 12% lacked LDL-C measures in the first year and then this proportion increased gradually from 42% in year 2 until reaching 61% in year 10. There were no significant differences in proportions lacking LDL-C testing by sex. Conversely, there were significant differences in lacking LDL-C measures by race/ethnicity which were most marked in year 1 (47% in African Americans vs 10% to 13% in other ethnic groups). It is also remarkable that African-Americans showed the highest proportion of lacking LDL-C testing every year thereafter, up to year 10. There were also prominent differences in lacking LDL-C testing according to degree of engagement with the Cardiac Rehabilitation Program, specifically in year 1: it was 1% among those completing the program (6+ sessions), 12% among those with partial completion (1-5 sessions), and 23% among those not enrolling in the program. Differences across cardiac rehabilitation groups were not obvious in subsequent years.

The following factors (ascertained at baseline) were (significantly and independently) positively associated with not having any LDL-C measure(s) in Year 1 after the index event: current smoking (OR = 1.18; 95% CI, 1.08-1.29; P <.001), BMI < 18 Kg/m ² (OR = 1.69; 95% CI, 1.26-2.25; P <.001), hypertension (OR = 1.14; 95% CI, 1.07-1.23; P <.001), iron deficiency anemia (OR = 1.10; 95% CI, 1.02-1.18; P =.02), heart failure (OR = 1.48; 95% CI, 1.36-1.61; P <.001) and living 10 to 50 miles from the closest KP facility (OR = 1.09; 95% CI, 1.02-1.17; P =.01) (Supplemental Table IV). In turn, the following factors were (significantly and independently) negatively associated with not having any LDL-C measure(s) in Year 1: age 65-74 (OR = 0.83; 95% CI, 0.75-0.91; P <.001), female sex (OR = 0.93; 95% CI, 0.87-0.99; P =.02), Asian & Pacific Islander race (OR = 0.82; 95% CI, 0.76-0.89; P <.001), quintiles 2-4 of the NDI (OR = 0.91;95% CI, 0.85-0.98; P =.01), 1-5 Cardiac Rehabilitation Program visits (OR = 0.49; 95% CI, 0.46-0.52; P <.001), and 6 visits to the Cardiac Rehabilitation Program (OR = 0.06; 95% CI, 0.05-0.09; P <.001).