Demographics and health status
The study population consisted of 729,531 US adult respondents, aged 40 years and older at the time of the survey. The study population was predominately white (85.2%) and approximately half were women (53.4%) (Table 1). Health status ranged from 24.9% excellent, 28% very good, 28.6% good, 12.7% fair, to 5.4% poor (Table 1). The distribution of self-rated health status differed by age at the interview, with excellent health reported proportionately lower in the older participants, while fair to poor health status was higher in the older participants (Fig. 1). Health ratings were similar in men and women, though men were slightly more likely to rate their health as “excellent” relative to women in younger participants: for example, the proportion of those who rated their health as excellent among men and women at age 40 were: 39.2 and 36.4% in whites and 30.9 and 23.5% in blacks, respectively.
Impact of health status on survival
Individuals reporting poor and fair health had a much higher risk of death compared to individuals reporting excellent health, and the difference in risk by health status was substantially higher in younger than older people (Fig. 2a). For example, among white men aged 40 years, those reporting poor health had 8.5 times (95% CI: 7.0, 10.3) greater risk of death compared with those reporting excellent health; whereas among white men aged 80 years, risk of death was 2.4 times higher (95% CI: 2.1, 2.8) for poor health compared to excellent health. In white women, comparisons of poor and excellent health correspond to 10.9 times (95% CI: 9.0, 13.2) greater risk of death at age 40 years and 2.1 times (95% CI: 1.9, 2.4) at age 80 years. Risk of death for those with very good and good health, compared to those in excellent health at the same chronological age, were similar across age groups. Figure 2b shows the effect of race on risk of death. Relative to white participants, black participants had a greater risk of death at younger ages but had similar or lower risk at older ages. At younger ages, the impact of health status on survival and life expectancy was greater in black compared to whites. While other races had a lower risk of death compared to whites overall.
Estimated survival varied greatly by age and health status among white participants (Fig. 3), while the results were similar for blacks and other races (data not shown). There is almost no or little difference in survival among health status at a younger age. However, the small change in the survival probability will result in a large difference in life expectancy. The effect of health status on survival is strong for middle to old age. At older age, the effect of health status becomes smaller again. In other words, life expectancy in higher ages has less variability.
The survival experiences of individuals with favorable health status were generally similar (i.e., excellent, very good, and good health in younger participants; excellent and very good health in the elderly). As expected, survival probabilities were dramatically lower in older participants. Survival was close to 100% for individuals aged 40 to 55 with excellent to good health. Indeed, the survival estimated from the US life tables is generally close to the survival for people in good health. Individuals of all ages who reported fair and poor health had worse survival estimates relative to the US general population.
Notably, older individuals’ survival was higher in the NHIS study population compared to the US general population. This is likely because the NHIS sample did not include institutionalized individuals, who likely have a higher mortality risk. For example, 5-year (and 10-year) survival of the average white woman conditional on surviving at age 80 is 78% (50%) in the NHIS population, while it is 72% (41%) in the US general population. As such, at age 80, the survival experience of the US general population was close to individuals who reported fair health in the NHIS.
The sensitivity analysis used different models and found that the estimated survival outcomes were similar regardless of model selection (Supplemental Fig. 2). Compared to self-report, family reported (proxy-report) health status showed lower mortality risk (risk of death) in reported excellent health at all ages and higher mortality risk in individuals with poor health at older ages. However, the magnitudes were minor and the effects were not significant in some age groups (Supplemental Fig. 3).
Life expectancy adjusted by health status
Life expectancy varied considerably by self-rated health status. When compared to the US general population matched by age, sex, and race, the life expectancies of those in excellent and very good health were longer, while of individuals with fair and poor health were shorter (Table 2). The remaining life expectancy at 60 males in the US general population was 20 for whites, 17 for blacks, 24 for others. In contrast, when considering health status, the remaining life expectancies varied substantially; those were 27 for excellent, 24 for very good, 21 for good, 17 for fair, 14 for poor health status in whites; 28, 25, 22, 17, 12 for blacks; 31, 28, 25, 20, 16 for other races. Variabilities between life expectancy by health status and general population average were more substantial in blacks than those in other races. The results were similar in the females; however, there existed a larger variability by health status. Discrepancies in life expectancy by health status were greater in younger ages. For example, life expectancies of white men in excellent health versus poor health is 45 versus 22 years (i.e., 23-year difference) at age 40, whereas it is 10 versus 5 years (i.e., 5-year difference) at age 80. In other words, at age 40, life expectancy (the differences in life expectancy relative to the US population average) for excellent, very good, good, fair, poor were 45 (8),42 (5), 37 (0), 29 (− 8), 22 (− 15); their remaining life vary greatly from 22 to 45, and the difference with US average was greater up to 15 years. In contrast, those were 10 (3), 10 (2), 8 (1), 7 (0), 5 (− 2) at age 80, showing, at older ages, less variability in life expectancy. Life expectancy highly depends on age, and in older ages, the effect of other factors (such as self-rated health), will become less distinctive.
The life expectancies of the NHIS study population were closer to reported good health status rates in all ages (Table 2). Life expectancies of the average US general population in the elderly are shorter, particularly in blacks, than those estimated from good health status in the NHIS population. For example, in blacks, the average life expectancy of the US general population is 4 years shorter at age 65 and 3 years shorter at age 70 (Table 2).