United States Dietary Trends Since 1800: Lack of Association Between Saturated Fatty Acid Consumption and Non-communicable Diseases

We reviewed data on the American diet from 1800 to 2019. Methods: We examined food availability and estimated consumption data from 1800 to 2019 using historical sources from the federal government and additional public data sources. Results: Processed and ultra-processed foods increased from <5 to >60% of foods. Large increases occurred for sugar, white and whole wheat flour, rice, poultry, eggs, vegetable oils, dairy products, and fresh vegetables. Saturated fats from animal sources declined while polyunsaturated fats from vegetable oils rose. Non-communicable diseases (NCDs) rose over the twentieth century in parallel with increased consumption of processed foods, including sugar, refined flour and rice, and vegetable oils. Saturated fats from animal sources were inversely correlated with the prevalence of NCDs. Conclusions: As observed from the food availability data, processed and ultra-processed foods dramatically increased over the past two centuries, especially sugar, white flour, white rice, vegetable oils, and ready-to-eat meals. These changes paralleled the rising incidence of NCDs, while animal fat consumption was inversely correlated.


INTRODUCTION
The American diet changed radically since 1800 due to industrial and technological advances, geographic spread, urbanization, wars, cultural changes, as well as food industry conglomerates and globalization. Foods became progressively more processed, associated with a parallel but delayed rise in non-communicable diseases (NCDs) in the United States, other western nations, and more recently, in developing nations, as their diets and lifestyles westernized. Chronic NCDs include metabolic syndrome [abdominal obesity, hypertension, non-alcoholic fatty liver disease, insulin resistance, hyperinsulinemia, elevated triglycerides, low high-density lipoprotein (HDL)], type 2 diabetes (T2D), gout, heart disease, stroke, cancer, polycystic ovarian syndrome, and Alzheimer's disease (1,2). Ancel Keys' Diet-Heart Hypothesis posited that the midnineteenth century heart disease epidemic resulted from "a changing American diet": increased consumption of fats, especially saturated fatty acids (SFAs), and decreased grain consumption (2,3). Supporting evidence included (1) hypercholesterolemia in heart disease patients, (2) rabbits fed high fat diets developed hypercholesterolemia and atherosclerotic-like lesions, (3) cholesterol comprised 40-70% of atherosclerotic plaques, (4) SFAs increased serum cholesterol in short-term feeding studies, (5) high fat diets were associated with higher cholesterol levels in some populations, (6) fat consumption was correlated with heart disease deaths over time in the United States and six other countries, and (7) familial hypercholesterolemia patients had increased heart disease rates (4).
In 1961, the American Heart Association first recommended that men at high-risk for cardiac disease should reduce total fat consumption to 25-35% of calories and substitute polyunsaturated fatty acids (PUFAs) for SFAs (5). Neither prospective studies nor randomized trials supported that recommendation (6,7). In 1977, Senator George McGovern's Dietary Goals transformed this hypothesis to national policy and extended SFA's effects to obesity and cancer and recommended reductions for everyone over age 2 years. The Surgeon General, National Research Council, and American Cancer Society also recommended low-fat or low-SFA diets to reduce coronary heart disease (CHD) and cancer (8)(9)(10)(11). However, some studies correlated low cholesterol levels with higher cancer rates and low-fat diets or substituting PUFAs for SFAs correlated with increased mortality, cardiovascular disease, and cancer (12)(13)(14)(15)(16)(17). Prospective studies and randomized trials found that dietary fat, SFAs, or elevated cholesterol levels were not associated with increased cancer risk (18,19).
The belief that fat and SFAs drive obesity and heart disease has persisted more than a century (20). In 1960, the Framingham Heart Study found no link between fat or SFA consumption and heart disease, but this data was never published (21,22). In 1967, Fredrickson (NIH Director 1975-1981, Levy (NHLBI Director 1975-1981, and Lees (Rockefeller University) wrote a five-part New England Journal of Medicine series that identified elevated very-low-density lipoprotein (VLDL) cholesterol as the most common lipoprotein disorder. A "sizable fraction of the population suffering from coronary heart disease" with "carbohydrate-induced hyperlipidemia" should be treated with "weight control, avoidance of excessive dietary carbohydrates and hypolipemic agents (23)." The identification of metabolic syndrome and its association in animal models and humans with diets rich in refined carbohydrates further supported that fats and SFA were not the primary dietary drivers of obesity and heart disease (24).
A foundation of Keys' Diet-Heart Hypothesis and McGovern's Dietary Goals was that heart disease resulted from America's declining grain and increasing fat and SFA consumption (8,25). Short-term studies do not replicate long-term metabolic and hormonal changes, and population-based studies over time are limited by reliability and validity (26)(27)(28). We reviewed the available data to examine the American diet from 1800 to 2019.

MATERIALS AND METHODS
We analyzed data from 1800 to 2019 on food availability for categories (e.g., red meat, poultry, fruits, vegetables, fats, and oils, sugar, etc.), total energy consumed (kcal), and macronutrient intake from the US Department of Agriculture (USDA) Economic Research Service (ERS), USDA Department Circular 241 (1929), US Department of Commerce (USDC) Historical Statistics of the US 1789-1945 (1949), USDA Food Consumption 1909-1952(1953, USDA The National Food Situation NFS- 74 (1956), and the National Bureau of Economic Research's The Changing Body (2011) (29)(30)(31)(32)(33)(34). The resources with their respective data availability are summarized in Table 1. Additional sources were used when estimating availability of processed and ultra-processed foods.
For years with multiple data sources, an average was calculated to estimate food availability. Semi-centennial averages are summarized in Table 2. All numbers are average food availability per capita per year, unless otherwise stated.
For estimates from 1909 to the present, the USDA ERS was our primary source, with data gathered directly from producers and distributors, tracking annual commodity production to end products. Per capita food availability estimates were first published in 1941 to assess WWII resources and a historical series for 1909-1940 was retrospectively created. Since 1941, yearly per capita food availability estimates were published.
Per the USDA ERS, the Total annual food supply of a commodity = Available commodity supply (production + imports + beginning stocks) -Measurable nonfood use (farm inputs + exports + ending stocks, etc.) and Per capita availability = Total annual food supply of a commodity / U.S. population for that year (from the US Census Bureau) (29).
Classification of food processing is based on NOVA criteria (44,45). We used per capita food availability estimates interchangeably with per capita food consumption, as the USDA used these data as proxies for actual consumption at the national level.
These data sources provide per capita estimates on consumption of energy, nutrients, and non-nutrient food components from foods and beverages from 1909 to 2019. These nutrient intakes do not include dietary supplements or medications. We graphed data trends over time and calculated percent changes.
We also included USDA loss-adjusted food availability data from 1970 to 2019, calculated by estimating food loss at primary, retail, and consumer levels. Primary losses occurred from farm to retail weight. Retail losses occurred at supermarkets, convenience stores, and small grocery stores. Consumer losses included food discarded at home or at restaurants, including expired foods, non-edible portions (e.g., apple cores), loss from cooking and   1789-1945 (1949), USDA Food Consumption 1909-1952(1953, USDA The National Food Situation NFS- 74 (1956), and The Changing Body (2011).
plate waste (29). Loss-adjusted food availability data from 1970 to 2019 were analyzed separately.

Data Limitations
Data from the USDA, USDC, and other sources used different methodologies which varied over time. For example, some values gathered from different time periods were reported in different units, and conversions are not easily applied as food and beverage weight and volume measurements differ greatly. We reviewed all identified USDA and USDC, US Government Printing Offices and other federal websites, and historical sources from the nineteenth and early twentieth centuries. Despite the historical sources, we lacked data from 1800 to 1908 for fish and shellfish, added fats and oils, grains, and vegetables. We also lacked data on sugars and sweeteners from 1800 to 1874. USDA availability data from 2019 were also not available at the time of analysis for fresh fruits and vegetables and for fish and shellfish. USDA ERS availability data from 2010 and onward for durum flour (embedded in total wheat flour) and added fats and oils were missing due to the termination of select Current Industrial Reports (CIR) by the Census Bureau. Data for rice are unavailable after 2010 due to a large unexplained decline in the implied total domestic and residual use (unreported losses in the milling, transporting, and marketing of rice) estimates. Thus, per capita daily amounts of calories and food pattern equivalents could not be calculated beyond 2010 for the added fats & oils group, as well as the summary estimates or totals across all food groups (29).
USDA ERS 1909-1941 food availability data used retrospective estimates without referenced sources. Data were likely more accurate for products primarily imported and taxed like sugar, or those entering the US during specific times: e.g., industrial seed (vegetable) oils and margarines after 1910 and high fructose corn syrup (HFCS) in the 1970s. Ready-to-eat cereal data were not available from 1989 to 2012. The USDA ERS acknowledged inaccuracies due to "incomplete reporting, inaccurate conversion factors, and inappropriate estimation techniques, " as well as information on retailers' and wholesalers' inventories. Food availability data do not include information on processing before sale, where the foods were sold, how they were prepared and eaten, or consumer profiles (29).
Changing inclusion criteria for food availability data limited some comparisons across different periods. The USDA included the military in average food availability estimates until 1941 but excluded the military afterwards; this accounted for the dramatic decrease in certain foods from 1941-1945, together with civilian quotas (29).
The 1909-1941 USDA ERS estimates did not accurately track foods consumed before transport, underrepresenting local produce and meats from farms at a time when many Americans lived on or near farms. For example, in 1920, 30.2% of Americans lived on farms and many others lived nearby (46). Consumption of offal (organ) meats, marrow, feet, snouts, "spam, " and bloods was rarely tracked and varied over time and socioeconomic groups (47). Farm to retail and consumer-level food loss estimates were imprecise (48). The USDA ERS data estimated commercial vegetables and home-grown vegetables from the early 1970s and the present, although home-garden data are unreliable (49,50).
USDA ERS loss-adjusted food availability data were published after 1970 but were unreliable. Consumer and household surveys suffered from recall bias and variations in food loss over time due to increased fat trimming from meats or greater shelf-life due to preservatives were not accurately assessed (29,48). We only included Nationwide Food Consumption Surveys (NFCS) and National Health and Nutrition Examination Survey (NHANES) data for non-nutritive sweeteners. The surveys only started reporting data in 1965 (NFCS) and 1999 (NHANES) and were limited by biases, inaccurate recall, mis-estimates of energy consumption, and internal inconsistencies (51).
Whole milk and cream availability decreased 56% from 1800 to 2019 (323-142 lbs.). After adjusting for food loss, between 1970 and 2017, milk and cream availability decreased 49% from 188 to 96.6 lbs.

Added Fats and Oils
Estimated total added fat and oil availability increased 118% from 1909 to 2010 (38.5-83.8 lbs.), with a striking decline in animal fat and increase in industrial seed oil and vegetable shortening from 1909 to 1970. From 1909 to 2010, total availability per capita for animalbased fats (including butter, lard, edible tallow) decreased 58% (21.2-8.8 lbs.) and vegetable-based fats and oils (margarine, shortening, salad, and cooking oils (included only after 1965), and edible fats and oils found in confectionery products and non-dairy creamers) increased 159% (31.7-82.2 lbs.) (Figure 5).
Additional data from the NFCS and NHANES estimated a 306% increase from 1965 to 2002 in SSB caloric consumption per capita per day (50-203 kcal) among the general population. In the same time frame, the percentage of people reported to have had consumed SSB increased from 29.2 to 59.6%. And, among SSB consumers only, the average SSB calories consumed per day increased 86% (173-321 kcal) (56).

Non-nutritive Sweeteners
In estimates gathered from the NFCS and NHANES, total non-nutritive sweetener consumption per capita increased 1,227% from 1965 to 2004 (11-146 g). In the same time frame, the percentage of the US population who reported to have had consumed foods and beverages containing nonnutritive sweetener increased from 3.3 to 15.1%. When considering the average consumption of non-nutritive sweeteners among only those who reported to consume them, the average consumption increased 118% (304-663 g) per consumer (58).

Processed and Ultra Processed Foods
The 2014 consensus classification of unprocessed/minimally processed, processed ingredients, ready-to-eat food products, and ultra-processed foods cannot be applied systematically to earlier periods (49). There was a dramatic increase in processed and ultra-processed food consumption in the United States over the twentieth century, including ingredients (e.g., seed oils, white flour, rice, sugars, and syrups), food products (e.g., canned or bottled vegetables and legumes in brine; peeled or sliced fruits in syrup; tinned fish in oil; salted nuts; ham, bacon, smoked fish, cheese), and ultra-processed products (e.g., chips, pretzels, ice cream, hot dogs and hamburgers, chicken fingers, cereals, cakes, energy bars, pizza, sodas and sports drinks, fruit yogurts) (35).

Processed Meat
Processed meat includes both red meat and poultry that has gone through salting, curing, fermentation, smoking, or had chemical preservatives added to it. Data from NHANES revealed that the average processed meat consumption per week per capita among adults (≥20 years old) increased 3% from the 1999-2000 survey period to the 2015-2016 time period (182-187 g). The most common processed meat consumed in the 2015-2016 survey period was luncheon meat (73.3 g/week), followed by sausage (45.5 g/week), hot dog (17.5 g/week), ham (17.5 g/week), and bacon (8.6 g/week) (60).
Additional data from NHANES include the prevalence of RTE cereal consumption among adults and children, rather than average consumption per capita. Prevalence of RTE cereal consumption among US adults aged ≥18 years was 20% in the 2003-2004 cycle before increasing to 24% in the 2009-2019 cycle and subsequently decreasing to 19% in the 2015-2016 cycle. In the 2015-2016 cycle, RTE cereal contributed to 10% of total energy intake (214 kcal/day/capita out of 2,135 kcal/day/capita) in adults aged ≥18 years who are RTE cereal eaters. When combining both RTE cereal eaters and non-eaters as the whole population, the number was 2% (42 kcal/day/capita out of 2,102 kcal/day/capita) (67).
Among all children from 0.5 to 17 years old, prevalence of RTE cereal consumption was 41% in the 2003-2004 cycle before decreasing to 36% in the 2015-2016 cycle. In the 2015-2016 cycle, among children 0.5-17 years old who are RTE cereal eaters, RTE cereals make up 9% of total daily energy intake (161 kcal/day/capita out of 1,788 kcal/day/capita). When combining both RTE cereal eaters and non-eaters as the whole population, the number decreased to 3% (54 kcal/day/capita out of 1,807 kcal/day/capita) (68).

DISCUSSION
The American diet has changed radically in the past two centuries, with the most marked changes including increased consumption of processed and ultra-processed food (e.g., sugar, white flour, white rice, and industrial seed/vegetable oils) and poultry and reduced consumption of unprocessed foods (e.g., fresh fruits and vegetables) and animal fats (e.g., whole milk, butter, and lard). Changes in food availability over the past two centuries included (1) increased processed and ultra-processed foods, sugar, industrial seed oils, and poultry; and (2) decreased butter/lard/shortening, dairy (mainly whole fat), fresh fruits, fresh vegetables, and red meat (beef/pork). Ultra-processed foods were rare before 1900 but increased to more than 50% of the current American diet (44). SFA consumption remained relatively stable, as lard, butter, whole milk, and red meat decreased while margarine, shortening, and other vegetablebased saturated fats increased. Meanwhile, PUFA and MUFA consumption increased dramatically with the introduction of ultra-processed foods and industrial seed and vegetable oils.
The unprocessed elements of our nineteenth century dietanimal fats, whole fat dairy, fresh vegetables, and fresh fruitswere progressively replaced with more processed elements, including industrial seed oils, HFCS, and ready-to-eat snacks and meals. The data do not support the widely publicized "changing American diet" of increasing animal-derived SFAs over the first 60 years of the twentieth century (8,25,69,70). Rather, polyunsaturated fats and partially hydrogenated fats from vegetable oils progressively replaced lard, butter, and other animal-derived fats. Across the twentieth century, rising rates of obesity, diabetes, heart disease, and cancer were associated with stable SFA consumption. Yet, large increases in sugar and refined carbohydrate consumption and more modest increases in total calories make refined carbohydrates and total calories more likely factors than SFA in NCD pathogenesis.
Data from the USDA and other sources have multiple and significant confounds. The more recent National Health and Nutrition Examination Surveys (NHANES) data we used to estimate processed and ultra-processed foods are considered the gold standard but their validity remains controversial, with major shortcomings (48,(71)(72)(73)(74)(75). Retrospective USDA estimates from 1909 to 1940 were inaccurate and unreliable, to an unknown degree. As one moves back in the nineteenth century, data are progressively scant and imprecise. Data on commodities such as fruits, vegetables, and grains are limited before 1940 by poor documentation of local sources. Historical accounts and records identify marked seasonal, geographic, and socioeconomic differences. Further, local consumption was extensive as most Americans lived on or near farms, but the data were not accurately measured in national estimates.

The Changing American Diet: History and Influence
The increased consumption of red meat and SFAs as the cause of the heart disease epidemic was one foundation for Keys' Diet-Heart Hypothesis, strengthened by authoritative repetition, including McGovern's Senate Select Committee's Dietary Goals for America (1977), Science in the Public Interest's (1978) monograph The Changing American Diet, the New York Times columnist Jane Brody's (1985) Good Food Book, Surgeon General Koop's Report on Nutrition and Health (1988), and the World Health Organization's Diet, Nutrition, and the Prevention of Chronic Diseases (1990) (8,69,70,75,76). However, neither the USDA nor other data supported this narrative (77). From 1800 to 2000, red meat consumption declined by 44%, fluid and cream dairy consumption declined by 48%, and egg consumption increased by 241%. From 1909 to 2010, lard consumption declined 78% and butter declined 68%, while margarine increased 192%, shortening increased 91%, and salad and cooking oils increase 329%. Americans consumed up to 70% fewer SFAs from animal sources by the end of the century, as obesity and diabetes epidemics emerged, alongside an increased incidence of NCDs such as cancer and heart disease (78).
The alleged increase in American SFA consumption in the twentieth century was considered the cause of the dramatic rise of non-communicable diseases (NCDs). Fats, especially SFAs, were considered uniquely toxic due to their caloric density or role in atherogenesis. Disorders linked to high fat/SFA diets included (1) overweight and obesity (too many calories with fat as main driver, insufficient exercise), (2) elevated cholesterol (from SFA), (3) hypertension (high salt and obesity), (4) colon and breast cancer (fat and SFA), and (5) diabetes (obesity and fats) (8). Yet, the rate of in NCDs continued to increase even after CDC guidelines encouraged Americans to reduce SFAs (79). Total SFA consumption increased slightly for total grams consumed while the percentage of all calories was stable (∼13.2%). From 1909 to present day, SFA from animal sources declined significantly but SFA from partially hydrogenated vegetable oils (contained in shortening and processed/ultra-processed foods) increased greatly. By contrast, the average American consumed >10-fold more "heart-healthy" PUFAs and MUFAs, and added caloric sweeteners tripled across the twentieth century. Our findings suggest that SFAs are unlikely to drive obesity, diabetes, or other NCDs, although this belief is held by many leading public health organizations (76). The early data that led to the belief that SFAs were dangerous deserve scrutiny.
The 1961 Framingham Heart Study (FHS) initially reported that high cholesterol correlated with heart disease and dietary SFA was the nutrient most strongly related to elevated total cholesterol in short-term feeding studies (80). However, by 1961, the relationship between dietary fats, carbohydrates, and lipoproteins was more complex. The effects of short-term and long-term feeding studies often differ and nutrients such as sugar and SFAs affect lipoprotein fractions differently. SFAs raise highdensity lipoproteins (HDL), which carry HDL-cholesterol, and high HDL levels have been shown to be potent predictors of heart disease risk than low-density lipoproteins (LDL) or total cholesterol (81). Additionally, diets rich in sugar and refined carbohydrates elevate triglycerides and inflammation (82,83). Longer follow-ups with more patient-years from the FHS found that total cholesterol, after accounting for factors such as blood pressure and smoking, was only a risk factor in heart disease or total mortality for men under age 65 years; it was far less significant for women under age 50 years and insignificant for those older than 50 years old (84,85). Further into the study, the FHS dietary data found that neither fat nor SFA consumption were related to cholesterol levels, coronary heart disease, or mortality (80). Subsequent studies, with larger and more diverse samples, failed to confirm the Seven Countries Study association of SFAs or fats with heart disease (19,(86)(87)(88)(89).
McGovern's Senate Select Committee's Dietary Goals for America (1977) was pivotal in definitively linking dietary SFAs as a major cause of heart disease, obesity, and cancer (8). Yet, three of eight senators dissented because many experts testified that neither total fat nor SFAs caused heart disease; rather, they interpreted the evidence as implicating sugar and refined carbohydrates in causing obesity, diabetes, and heart disease in animals and humans (90). A decade before the McGovern report, the future NIH and NHLBI directors found that the most common hyperlipidemia in cardiac patients primarily resulted from excess carbohydrates (23). Further, converging evidence revealed that metabolic syndrome results from refined carbohydrates in animals and humans.
US and international agencies and medical associations strongly supported a low-fat/low-SFA, high-carbohydrate diet for everyone over age 2 years, and through 2008, advocated sugar as healthy for diabetics and the general population (91). The strongest evidence implicating SFA remains in studies in which SFAs are replaced with MUFAs or PUFAs, and heart disease, and less often, overall mortality, were reduced, although some observational studies and randomized controlled trials challenge these findings (19,88,92,93). These studies cannot assess the harmful effects of SFAs or how increased MUFAs and PUFAs may be beneficial and SFAs neutral, as suggested by population-based prospective studies (94)(95)(96).
Untangling the causes of NCDs is complex, multifactorial, and controversially unresolved. The profound dietary changes were accompanied by other lifestyle and demographic changes, including (1) increased urbanization and population density, (2) reduced physical activity commuting to and at work, (3) longer commutes, (4) higher stress, (5) less sleep, (6) more machine and less human time, (7) higher rates of mental health disorders, (8) increased prescription and over-the-counter drug use, many of which increase appetite, and (9) higher salt intake (94). Increased obesity is a common precursor and risk factor for many NCDs (e.g., metabolic syndrome, T2D, heart disease, cancer, and gout) (97).
The energy balance hypothesis of obesity is supported by the 22% increase in available calories from 1970 to 2010 (Figure 1). There was a >30% increase in overweight Americans from 1976-1980 (25.4%) to 1988-1991 (33.3%), associated with an 11% decrease in percent of fat calories (41.0-36.6%), a 4% decrease in daily calories (1,854-1,785 kcal), and a 9.8-fold increase in high fructose corn syrup (78). During this period, Americans consuming low-calorie products rose from 19 to 76% while physical activity was stable (78). However, in the Women's Health Initiative study, three years after the intervention group consumed an average of 100 fewer calories per day and exercised more than the control group, the controls weighed 1.3 kg more, yet the energy balance predicted a difference of > 16 kg (88). Many impoverished populations underwent a dietary transition followed by rising obesity without any obesogenic environmental factors such as abundant dietary SFAs or labor-saving devices (Pima Native Americans in 1890-1920, Sioux Native Americans in 1920s, Jamaicans in 1970s, Zulus in Durbin, South Africa in 1960) (104,105,107,111). This rising obesity in adults, mostly women, while their children were malnourished, refutes the energy balance hypothesis as adults reduce their basal metabolic rate rapidly with decreased caloric intake, while children only do so after losing 20-30% of body weight (112)(113)(114)(115).
NCDs such as obesity, T2DM, heart disease, and cancer are rare in indigenous populations consuming native diets, even among elderly individuals (116)(117)(118)(119). These populations consumed diverse diets, some very high in SFAs from animals (e.g., Inuit, Maasai, Plains Native Americans) or plants (e.g., Polynesians, Tokelauns), while many others consumed diets high in complex carbohydrates and very low in fats (e.g., Pueblo Native Americans, Japanese, and Chinese farmers) (120)(121)(122)(123)(124). Native, minimally processed diets included minimal sugar or refined carbohydrates; honey being a major exception in some populations such as the Hadza (125). When populations adopted Western diets and lifestyles, NCDs emerged and increased (117,(126)(127)(128)(129)(130). Commensurate with these dietary transitions in indigenous populations, our findings suggest that increased sugar and refined carbohydrate consumptions during the twentieth century in America may have played a larger role than total calories or physical activity, although this remains a speculation without accurate data on all variables.

Future Direction
Understanding the pathogenic changes in American and other diets that drove the dramatic rise in NCDs remains one of the greatest challenges in public health. Given the challenges in obtaining accurate caloric estimates in national data, humility is needed to assess the diets of populations more than a century ago. Only well-defined changes (e.g., increased caloric sweeteners and PUFA and decreased SFA from lard and butter) can be identified. A more complete understanding of dietary and lifestyle factors in NCDs may emerge from an unbiased synthesis of the diverse evidentiary lines.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article. Further inquires can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
JHL, MD, and OD contributed to the study conception and design. JHL, MD, TR, and OD assisted with data and material collection. JHL and MD performed the data analysis and wrote sections of the manuscript. OD wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

FUNDING
This work was supported by the Finding A Cure for Epilepsy and Seizures (FACES) organization affiliated with NYU Langone Health and the NYU Comprehensive Epilepsy Center. The foundation had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.