Population health researchers, insurers, and officials have used the subset of the CDC’s Health-related Quality of Life (HRQOL) measures, the “Healthy Days” (HD) measures, to quantify the mental and physical health of Americans for decades.(2) The HD measures consist of several questions about a person’s health in the past 30 days:
Would you say that in general your health is excellent, very good, good, fair, or poor?
Now thinking about your physical health, which includes physical illness and injury, how many days during the past 30 days was your physical health not good?
Now thinking about your mental health, which includes stress, depression, and problems with emotions, how many days during the past 30 days was your mental health not good?
During the past 30 days, approximately how many days did poor physical or mental health keep you from doing your usual activities, such as self-care, work, or recreation?
Developed in response to the U.S. DHHS Healthy People 2000 Strategy, the HD measures aim to improve the population’s overall quality of life.(2,3) They have been assessed in more than 1.2 million U.S. adults through the Behavioral Risk Factor Surveillance System (BRFSS) interviews and validated as effective indicators of self-perceived health and predictors of mortality, hospitalization, and healthcare utilization.(2,4)
The population health field has adopted new goals that may not be fully captured by the HD measures. The measures are straightforward but disconnected from the field’s emerging strategy in two key ways. Healthy People 2030 has broadened its vision to include achieving “full potential for health and well-being across the lifespan,” but the HD questions haven’t been updated to reflect this.(5) Additionally, the initiative now prioritizes social aspects of health, aiming to improve health equity and address social determinants of health.(6,7) The HD measures don’t account for these expanded objectives.
Harvard University’s Human Flourishing Program developed the Flourishing Questions (FQ), a 12-question measure of human flourishing, or complete well-being, that aligns with the population health field’s current ambitions.(1,8) It covers six domains: happiness and life satisfaction, physical and mental health, meaning and purpose, character and virtue, close social relationships, and financial and material stability.
To be concise, we refer to this battery as the Flourishing Questions, although technically speaking, the first 10 questions are known as the Flourishing Questions, and when the two Financial and Material Stability Questions are added, the measure is referred to as the Secure Flourishing Questions.
Each domain has two questions. The questions have also been assessed through the Global Flourishing Study, which was launched to track participants’ flourishing longitudinally.(9) The FQ align with the field’s current vision of helping people achieve their “full potential for health and well-being.”(5) They also assess social determinants of health, such as close social relationships and financial stability, which are a priority area not captured by the HD questions.
Considering the converging trends in population health and the development of the FQ, Blue Cross and Blue Shield of Florida (BCBSF) has broadened its definition of well-being — as “an overall positive and hopeful mental state fueled by a sense of purpose and satisfaction with life, work, relationships, physical health and functioning, and supported by the capacity to adapt to life stressors along with a feeling of control over one’s future” — and adopted the FQ in addition to the HD measures. This has created a unique dataset with responses to both measures from the same individuals over time. This paper presents an initial analysis of the relationship between these measures.
This analysis raises the question of whether the FQ can replace HD as a population health metric. A conclusive answer requires comprehensive research extending beyond this analysis. This initial study aims to spark further investigation into this question in other contexts, encouraging others to explore the potential of the FQ in population health research.
Methods
This study focused on the mental and physical health (MH and PH) questions from the FQ and HD. The FQ, which are measured on a 0–10 scale, are: “In general, how would you rate your physical health?” and “How would you rate your overall mental health?” The corresponding HD questions are, “Now thinking about your physical health, which includes physical illness and injury, how many days during the past 30 days was your physical health not good?” and “Now thinking about your mental health, which includes stress, depression, and problems with emotions, how many days during the past 30 days was your mental health not good?”
The analysis had two parts: 1) examining whether the analogous HD and FQ are correlated and 2) comparing the correlations between post-survey hospitalization (PSH90) and the FQ and HD measures, respectively.
Hypotheses
The first analysis tested the following null hypothesis:
H0: There is no correlation between an individual’s HD response and their analogous FQ response for the PH and MH questions.
The second analysis tested the following null hypothesis:
H0: There is no difference between the Phi correlation coefficient of PSH90 with HD and PSH90 with FQ for the PH and MH questions.
Population
The FQ and HD data came from two surveys conducted by Blue Cross Blue Shield of Florida. In 2019, BCBSF incorporated the HD measures into a monthly survey to monitor members’ MH and PH. In 2022, BCBSF began separately distributing the FQ — containing all 12 questions from the Secure Flourish measure — as part of the emerging emphasis on overall well-being in population health. This analysis utilized BCBSF’s member database and measurement of both the HD and FQ to examine whether the Flourishing responses correlate with HD.
The HD are assessed as part of a monthly survey known as the Relationship Survey. Eligible members can receive this survey via email every four months and can thus report HD a maximum of three times per year. Not all questions are required, so HD are not always reported. The average response rate for the Relationship Survey is 4.33% (n ≈ 14,000 responses/month), and of those who begin the survey, 61.61% (n ≈ 8,650) complete at least one HD question.
The FQ were first distributed to BCBSF members in September 2022 and are sent to Relationship Survey respondents from the prior month’s distribution. It was distributed monthly in 2022 and semi-quarterly thereafter. While a FQ respondent also responded to the Relationship Survey in the previous month, they might not have answered the HD questions. The response rate for the FQ is higher (19%), which is expected; FQ respondents also responded to the Relationship Survey in the prior month and are presumably more likely to respond to surveys. BCBSF collected 34,709 FQ responses from unique members between September 2022 and November 2024.
Measures
Healthy Days
The focus of this study was the PH and MH questions from the CDC’s HD concept because these are analogous to the PH and MH domain from the FQ. These HD questions are integer variables (0-30 days). A score of 0 indicates that, within the past 30 days, the respondent reported no physically or mentally unhealthy days, while a score of 30 indicates that the respondent reported the maximum amount of mentally or physically unhealthy days. All responses of “Don’t Know” or “Not Sure” were coded as missing.
Flourishing Questions
The FQ includes 12 questions spread across six domains: happiness and life satisfaction, physical and mental health, meaning and purpose, character and virtue, close social relationships, and financial and material stability. This paper focused on the PH and MH domain (Domain 2). The two questions in this domain are: 1) “In general, how would you rate your physical health?” (0 = Poor, 10 = Excellent) and 2) “How would you rate your overall mental health?” (0 = Poor, 10 = Excellent).
Additional Measurements
Age, gender, and insurance segment — Individual Under 65 (IU65), Group, or Medicare — were gathered from BCBSF enrollment tables at the time of each member’s FQ response. IU65 members obtained a non-Medicare health policy either through the Health Insurance Marketplace or directly from BCBSF. Group members have coverage sponsored by their employer, and Medicare members are eligible with the federal government upon turning 65 or having a qualifying health condition or disability.
Additionally, each member’s geographic location was classified as rural or non-rural (or out-of-state for non-Florida members) based on the Florida Department of Health’s designation of their county.(10) Race and ethnicity were self-reported.
To evaluate the correlations between FQ or HD with PSH90, BCBSF claims data was leveraged to determine the member’s hospitalization status within 90 days post-survey. This measure was only calculated for members continuously enrolled for 90 days post-survey. Inpatient hospitalizations, psychiatric units, and comprehensive rehabilitation facilities were included.
Correlations
First Analysis: Flourishing Correlation with Healthy Days. For this first analysis, the most recent response from each member was selected. Of the 34,709 responses, 22,851 (65.8%) completed the Physically Healthy Days (PHD) question, and 23,890 (68.8%) completed the Mentally Healthy Days (MHD) question (Table 1).
The similarity of the PHD and MHD questions and the Physical Flourishing Question (PFQ) and Mental Flourishing Question (MFQ) was evaluated by calculating the Spearman correlation between the PHD and PFQ, and the MHD and MFQ. The Spearman correlation was chosen because of its robustness for non-linear relationships, as the relationship between the HD and FQ appeared possibly non-linear (Figure 1).
Figure 1. The Relationship Between Flourishing Question Score (Physical or Mental Health) and Mean Number of Unhealthy Days (Physical or Mental Health), Stratified by Health Insurance Segment, with 90% Confidence Interval Error Bars.
For each PFQ and MFQ response, the correlation with the analogous PHD and MHD response from the prior month was calculated (based on the assumption that responses closest in time should be the most similar). Correlations were stratified by insurance segment. For each correlation, a 95% confidence interval was constructed using 10,000 bootstrap iterations.
Second Analysis: Correlations with Post-Survey Hospitalization. To further evaluate the relationship between FQ and HD, the correlation between each of the four PH and MH questions and PSH90 was measured using Phi correlation coefficients.
PH and MH responses were separated into two samples. Each sample was constructed by identifying pairs of FQ and HD responses that both included PSH90 data. To ensure that we were not double-counting outcomes and to prevent a large gap of time between the paired responses, the paired responses were required to be at least 90 days apart and no more than 1.5 years apart. If a member had multiple FQ responses that met this criteria, one response was randomly selected. It was ensured that the FQ response came first in half of the pairs via a randomization method designed to ensure equally sized groups.
We transformed the HD and FQ into categorical measures (PHDgood, MHDgood, PFQgood, and MHQgood) where the binary value of “True” indicated good health. For PFQgood and MHQgood, “Good Health” was indicated by having a score of 6 or higher, and a score of 13 or fewer unhealthy days was considered “Good Health” for PHDgood and MHDgood. The Phi correlations were stratified by insurance segment. A 95% confidence interval was created for every Phi correlation using 10,000 bootstrap iterations.(11)
Results
First Analysis — Sample
Table 1 provides a profile of the members who completed the PH (or MH) responses on both the FQ and preceding HD. In both the PH and MH samples, there are roughly 23,000 members. There are similar distributions across segments: The majority are Medicare, nearly one-third are IU65, and the smallest share are Group members. The sample is also predominantly female, from non-rural counties, and older, with the majority of Group and IU65 members between ages 55 and 65. Also, although most members did not report their race or ethnicity, most self-reporters indicated White Non-Hispanic.
Table 1 also gives the mean FQ and HD scores. Among PH and MH respondents, on average, Group members have the highest FQ response and lowest HD response, whereas the pattern is reversed among Medicare respondents, with IU65 members in the middle.
First Analysis – Correlations
Figure 1 and Table 3 provide insight into the correlations between FQ and HD responses. For PH and MH, those who report higher levels of Flourishing tend to report fewer unhealthy days, except for a slight increase in unhealthy days when moving from a FQ score of 0 to 1. The PH and MH Spearman correlations were moderately negative and statistically significant at conventional levels for each segment. For Group, Medicare, and IU65, the respective MH correlations were -0.540, -0.430, and -0.514, and the PH correlations were -0.403, -0.398, and -0.425.
Second Analysis – Sample
As seen in Table 2, the PH portion of the second analysis includes 9,345 members, with the following distribution by segment: 740 from Group, 6,458 from Medicare, and 2,147 from IU65. The MH sample includes 9,676 members with a similar distribution by segment. The distributions of age, gender, race/ethnicity, and rural/non-rural closely follows those of the analogous populations in the first analysis (Table 1). Table 2 also gives the PSH90 status for each of the study’s four subsamples (PFQ, MFQ, PHD, and MHD respondents). Across all subsamples, few respondents experienced PSH90, with Medicare respondents having the highest proportion.
Second Analysis – Correlations
Figure 2 and Table 3 present the Phi correlations between PSH90 and PFQgood, MFQgood, PHDgood, and MHDgood by segment, along with 95% confidence intervals. All Phi correlation coefficients are negative, indicating members with “good” FQ or HD responses were less likely to be hospitalized after the survey than those with “bad” responses. The coefficients are relatively weak, and the FQ confidence intervals overlap with their HD counterparts, indicating that the FQ-HD measure-pairs were not statistically different.
Figure 2. Correlation (Phi) Coefficients and Corresponding 95% Confidence Intervals by Health Insurance Segment
Between Binary Measures of Healthy Days or Flourishing and Post-Survey Hospitalization as defined by any
Hospitalization with 90 Days Post-Survey.
Discussion
This is the first paper to examine the relationship between the FQ and HD measures of population health. We compared the PH and MH questions of both measures. The questions are moderately correlated at statistically significant levels, suggesting that they may be measuring similar phenomena. Correlations between PSH90 and the HD and FQ measures were calculated, based on evidence that HD measures have been shown to predict health outcomes.
In this analysis, there were no statistically significant differences between the correlations, indicating that neither measurement group correlated more nor less with hospitalization. Additional modeling exercises are required to determine if FQ measures perform similarly in predicting other relevant health outcomes.
There are several interesting patterns in the data that could be examined in future work. Although Medicare members likely struggle with more health issues than Group or IU65 members, those in this sample appear to have a relatively better perception of their own health (Table 1). Further, the weakest correlation is in the Medicare segment, indicating that the PH and MH of this older population may be less related to HD when compared to younger respondents.
Additionally, the FQ and HD MH questions are more strongly correlated with each other than the PH questions (Table 3). Finally, the relationship of PH and MH questions between FQ and HD may not be linear (Figure 1).
This study had several limitations. The sample is not representative of the population at large. It consisted of respondents with BCBSF insurance who differed from the American population in measured and unmeasured ways. For instance, they are older than average, and their racial/ethnic composition is largely unknown. Additionally, this study only examined the relationship between PFQ and MFQ with PHD and MHD questions. Future research should also consider how the other two questions from the HRQOL-4 HD measures (limited activity days and general health status) relate to the entire FQ.
The analyses presented suggest the PFQ and MFQ are empirically similar to their HD counterparts, but the advantages of FQ lie beyond these measurements. The FQ’s other domains capture social determinants of health not encompassed by the HRQOL-4 measures, such as close relationships, financial security, and life purpose, ultimately aligning with the population health field’s goal of measuring complete well-being.
Conclusion
Significant, moderately negative correlations were found between PFQ with PHD and MFQ with MHD across all health insurance segments, ranging from -0.398 to -0.540 (Table 3). There were no significant differences in correlations between PFQ/MFQ and PSH90 compared to PHD/MHD, suggesting that FQ measures may behave similarly to HD measures. However, further testing using models is needed to confirm this finding and explore the potential of the FQ in population health research.
References
Harvard University. Our flourishing measure. The Human Flourishing Program at Harvard’s Institute for Quantitative Social Science. https://hfh.fas.harvard.edu/measuring-flourishing .
Hennessy CH, Moriarty DG, Zack MM, Scherr PA, Brackbill R. Measuring health-related quality of life for public health surveillance. Public Health Rep. 1994;109(5):665–672. https://pubmed.ncbi.nlm.nih.gov/7938388 .
Centers for Disease Control and Prevention. Healthy People 2000. CDC National Center for Health Statistics. November 6, 2015. https://www.cdc.gov/Nchs/healthy_people/hp2000.htm .
Dominick KL, Ahern FM, Gold CH, Heller DA. Relationship of health-related quality of life to health care utilization and mortality among older adults. Aging Clinical and Experimental Research. 2002;14(6):499–508. https://doi.org/10.1007/bf03327351 .
Healthy People 2030 Objectives and Measures. Healthy People 2030. https://odphp.health.gov/healthypeople/objectives-and-data/about-objectives/healthy-people-2030-objectives-and-measures#:~:text=Healthy%20People%202030%20sets%20data,achieving%20the%20Healthy%20People%20vision .
Health Equity in Healthy People 2030. Healthy People 2030. https://odphp.health.gov/healthypeople/priority-areas .
Social Determinants of Health. Healthy People 2030. https://odphp.health.gov/healthypeople/priority-areas/social-determinants-health
VanderWeele TJ. On the promotion of human flourishing. Proc. Natl. Acad. Sci. U.S.A. 2017;114(31):8148–8156. https://doi.org/10.1073/pnas.1702996114
Johnson BR, VanderWeele TJ. The global flourishing study: a new era for the study of well-being. International Bulletin of Mission Research. 2022;46(2):272–275. https://doi.org/10.1177/23969393211068096
Community Health. Florida Department of Health. https://www.floridahealth.gov/programs-and-services/community-health/
Efron B, Tibshirani RJ. An introduction to the Bootstrap. New York: Chapman and Hall; 1994. https://doi.org/10.1201/9780429246593