This article, titled “New Forecasting Methodology Indicates More Disease And Earlier Mortality Ahead For Today’s Younger Americans,” authored by Eric N. Reither, S. Jay Olshansky, and Yang Yang, and published in Health Affairs in 2011, introduces a groundbreaking perspective on how population health statistics should be projected to inform public policy. The authors argue that traditional methods of forecasting health and longevity are fundamentally flawed, leading to potentially “inaccurate results and lead to inferior or even poor policy decisions”.
The core issue with these traditional “two-dimensional” forecasting methods is their reliance on the problematic assumption that “the future will be like the past”.
- They typically involve extrapolating past trends in death rates and life expectancy over various periods of observation.
- Such approaches fail to account for the crucial “time lag between behavioral risk factors now present among groups of younger people and those risk factors’ likely influence on the health status of these cohorts as they age”.
- Essentially, these methods provide only a “limited, ‘two-dimensional’ vision of the future” because they do not consider the potentially different risk factors accumulated by people who are still alive.
- The article highlights instances where two-dimensional forecasts have been inaccurate, such as:
- Company pension plans becoming strained or bankrupt due to underestimation of increased longevity.
- A trillion-dollar gap in state pension funds caused by inaccurate assumptions about future life expectancy.
- The Social Security Administration predicting a decline in diabetes mortality despite increasing prevalence of diabetes and obesity.
To address these limitations, the authors propose and demonstrate a more robust “three-dimensional” forecasting methodology. This approach is designed to provide a “more accurate gauge of the future of health and longevity” by uniquely incorporating:
- Age
- Period of observation
- Birth cohort
This “three-dimensional perspective” is critical because it explicitly accounts for the “time lag (latency period) between the accumulation of risk factors and their eventual manifestation in health outcomes,” such as cardiovascular disease or premature mortality. By focusing on the “accumulated life experiences of people now alive, not by the past life, health, and mortality experiences of the recently deceased,” this method offers a more reliable picture of future health.
The article applies this advanced technique to the US obesity epidemic, presenting a concerning outlook:
- The analysis suggests that “future death rates and health care expenditures could be far worse than currently anticipated” by conventional methods.
- Evidence cited indicates that recent cohorts of younger Americans (born since 1960) are at an increased risk of obesity and are becoming obese “earlier in life” compared to predecessors. This means they will carry “excess weight—and the elevated morbidity and mortality risks associated with it—for a larger proportion of their lives”.
- Furthermore, these trends were found to be “consistently stronger among blacks, particularly black females,” suggesting that existing racial/ethnic disparities are “deeply rooted in the past and therefore likely to persist into the future”.
- Supporting evidence for a worsening health outlook includes:
- Cardiovascular risk profiles having worsened since the late 1980s due to unfavorable trends in type 2 diabetes, blood pressure, and body mass index.
- Declines in the severity of coronary heart disease among younger people ending in the mid-1990s and possibly reversing after 2000.
- Significantly less improvement in coronary heart disease mortality among Americans ages 35–54 compared to those age 55 or older.
- Statistically significant declines in life expectancy observed between 1983 and 1999 in a “substantial number of US counties,” particularly among females in disadvantaged counties in the Southeast where obesity is concentrated.
To demonstrate the utility of their three-dimensional approach, the authors performed a concrete illustration using data from 1960-1999 to predict coronary heart disease mortality among US males in 2002 and 2007.
- They compared linear extrapolation (two-dimensional) with their age-period-cohort model (three-dimensional).
- Their hypothesis was that two-dimensional projections would “systematically underestimate observed coronary heart disease mortality rates for males age fifty and younger”, while the three-dimensional approach, by accounting for birth cohort variation, would be “less susceptible to errors in prediction”.
- The results confirmed their hypotheses: the two-dimensional method “underestimated observed rates of coronary heart disease mortality” for cohorts born after 1950. In contrast, three-dimensional projections “more closely approximated observed mortality”. Specifically, the observed decline in coronary heart disease mortality slowed for men aged 45-49 and reversed for men aged 25-29 after 2000, a trend not captured by linear extrapolation [26, Exhibit 1, Exhibit 2].
The article also emphasizes the Implications for Morbidity Projections:
- Scientists generally agree that the high prevalence of obesity and other adverse health conditions among children today will “increase the burden of various chronic diseases that children will face in the future”.
- Obese children are susceptible to metabolic syndrome, and excess weight gain in childhood is linked to its presence in young adulthood.
- The duration and degree of obesity influence the latency period for type 2 diabetes, suggesting that younger cohorts will develop diabetes earlier, accelerating negative health outcomes like end-stage renal disease.
- Early onset of type 2 diabetes from childhood obesity will lead to “several decades of exposure that will greatly elevate risks of not just coronary heart disease and stroke but also microvascular complications that can eventually lead to amputations, blindness, and kidney failure” by midlife.
- Two-dimensional projections are likely to “underestimate the future prevalence and severity of morbidity” because they don’t account for earlier onset of obesity in recent cohorts.
The research further delves into intergenerational processes, highlighting that obesity risks may be determined even before birth through:
- Fetal programming: Developmental changes in a fetus due to maternal environmental conditions, leading to harmful health effects later in life. For example, fetal overnutrition can permanently increase appetite and preferences for junk food in offspring.
- Transgenerational epigenetic inheritance: The inheritance of acquired factors from one’s forebears that alter genetic expression. Studies show maternal and paternal health behaviors, like diet and smoking, affect offspring’s obesity probability through metabolic programming.
- The “profound” implication is that the “high prevalence of obesity observed among younger people today is likely to be transmitted to future generations—regardless of the health behaviors of children yet to be born,” a factor not accounted for by two-dimensional models.
Ultimately, the authors strongly advocate for public policy makers to adopt this more robust forecasting tool and to “redouble efforts to develop and implement effective obesity-related prevention programs and interventions,” especially those targeting childhood obesity. This perspective, by focusing on the “observed health status of younger cohorts alive today,” offers a more realistic and comprehensive picture of future health and longevity, enabling better-informed public health strategies and targeted actions to improve population well-being. They note that existing initiatives, such as the Affordable Care Act of 2010’s Childhood Obesity Demonstration Project and First Lady Michelle Obama’s “Let’s Move” campaign, are consistent with the policies supported by this three-dimensional view.
Reference: Reither, E. N., Olshansky, S. J., & Yang, Y. (2011). New forecasting methodology indicates more disease and earlier mortality ahead for today’s younger Americans. Health Affairs, 30(8), 1562–1568. https://doi.org/10.1377/hlthaff.2011.0092