Relationship Between Socioeconomic Deprivation and Healthy Days Among Commercially Insured Adults Using a Digital Health Application: An Observational Study

Abstract

Population health research has long demonstrated that where someone lives is highly correlated with health outcomes and quality of life. This study explored if this relationship remained between zip code socioeconomic deprivation index (SDI) and member-reported healthy days among commercially insured adults interacting with virtual care and navigation services between May 1, 2023 and May 31, 2024, offered by Included Health. Members received an SMS-based survey that included the 4 Centers for Disease Control and Prevention Healthy Days questions after interacting with the digital health application. The proportion of members who reported frequent (14 or more) physically unhealthy, mentally unhealthy, and total unhealthy days during the past 30 days was calculated. The adjusted odds ratio for reporting frequent total unhealthy days was estimated by members’ zip code SDI quartile, accounting for member demographic characteristics. Of the 6692 survey respondents, 13.7% reported frequent physically unhealthy days, 20.8% reported frequent mentally unhealthy days, and 29.2% reported frequent total unhealthy days. After adjusting for covariates, members in the highest SDI quartile were 1.2 times more likely to report frequent unhealthy days (P = 0.047) than those in the lowest SDI quartile. The results demonstrate the importance of geographic indices, in the absence of other data, to assist employers in identifying members with potentially higher need of digital health services. It also highlights the feasibility of collecting quality of life measures to identify members who could benefit from timely intervention.

Introduction

Research in population health has long demonstrated the intricate relationship between community socioeconomic deprivation and health outcomes.^1

–4 One study estimated that state-level socioeconomic factors contributed to approximately 47% of health outcomes.⁵ Community socioeconomic status includes factors such as income levels, educational attainment, housing quality, and employment opportunities within a defined geographical area. These indicators provide a multidimensional assessment of neighborhood-level disadvantage, shedding light on disparities in health outcomes across different communities.

Concurrently, population health researchers have long prioritized self-reported health-related quality of life (HRQoL) measures as a critical means of monitoring population health. In fact, many organizations, including the Commonwealth Fund, have advocated for increased use of patient-reported outcome measures, including HRQoL, to improve care quality.^6,7 HRQoL measures capture individuals’ subjective evaluations of their own health over a specified period, offering insights into overall well-being that extend beyond traditional clinical markers. One visible example of the importance of HRQoL measures is their inclusion within the US Department of Health and Human Services Healthy People 2020 and 2030 initiatives.^8,9 The Healthy People 2020 initiative captured HRQoL using the Centers for Disease Control and Prevention’s (CDC) HRQoL-4 (“Healthy Days”) survey.¹⁰ Healthy Days is a validated instrument that has been used for over 30 years, including as a component within the Behavioral Risk Factor Surveillance System (BRFSS) survey, to collect self-reported HRQoL through 4 questions: overall health, and the number of physically unhealthy days, mentally unhealthy days, and activity-limited days during the past 30 days.^10,11 The CDC created this questionnaire as a concise alternative to longer measures of HRQoL for extensive population surveys. It is designed to be easily comprehensible to both the general public and policymakers. Aside from its simplicity, Healthy Days is distinguished by its ability to separately assess physical and mental health domains, providing a comprehensive yet straightforward evaluation. The strengths and utility of the CDC’s Healthy Days measures are likely why some health care companies are capturing these data from their members.^12
–14

Population health research has shown an association between community socioeconomic deprivation and healthy days specifically. One study found that individuals residing in counties with greater socioeconomic deprivation reported higher unhealthy days.¹² This relationship remained, even after accounting for individual SDOH.^15,16 In fact, it has been estimated that county-level SDOH factors explain up to 22% of the variance in physically healthy days and up to 10% of the variance in mentally healthy days between counties.¹⁵ Additional opportunities remain to comprehensively explore this correlation to inform effective health interventions. First, previous studies have not addressed the association between community socioeconomic deprivation and healthy days among specific subgroups, including those with access to virtual health care services that may mitigate health disparities. Second, the studies defined socioeconomic deprivation at the state and county levels, both of which are heterogeneous geographic areas. Additional research using more homogenous geographic areas, such as zip codes, may clarify the relationship between geographic socioeconomic deprivation and healthy days. Finally, existing studies have adjusted for community and individual-level SDOH but have not incorporated other characteristics linked to healthy days, such as sexual orientation and gender identity.^17
–19

Included Health, a health care company offering virtual care and navigation services through employer-provided benefits, uses the Healthy Days survey to identify and engage individuals most in need of services and measure the impact of these services on quality of life. Using this data, this study aims to determine to what extent zip code-level socioeconomic deprivation is associated with healthy days among adults who are engaging with virtual care and navigation services while controlling for individual characteristics not readily available in many research datasets.

Methods

This cross-sectional study utilized de-identified data from Included Health’s digital health application. The authors did not have access to any identifiable information about the study participants. As part of the Included Health application registration process, individuals agree to the data privacy policy, which includes a provision that their de-identified data may be used for research purposes. The WIRB-Copernicus Group Institutional Review Board considered this study exempt in accordance with 45 CFR §46.102.

Study setting and participants

This study included commercially insured adults, aged 18 or older, who accessed the Included Health digital application between May 1, 2023 and May 31, 2024, and were eligible to subsequently receive an SMS-based “Health check-in” survey. Included Health provides health care navigation and virtual care services as an employer-provided benefit to employees and their dependents (“members”). Services include claims advocacy, virtual second opinions, high-quality provider recommendations, specialized navigation for members who identify as Lesbian, Gay, Bisexual, Transgender, and Queer or within the Black population, and virtual care (including behavioral health therapy, psychiatry, primary care, and urgent care) delivered via video by a team of clinicians.

Within 1 hour of accessing the digital health application, the member received the 4 questions from the Healthy Days survey via SMS (Table 1). This information was used by Included Health to prioritize which members received an invitation for clinician outreach via phone to help them identify and access services that could ultimately improve their health.

Table 1.

SMS-Based “Health Check-In” Survey, May 2023 Through May 2024

1. You’re invited to check in with your care team about your health via text. Every 3 months, we’ll text you 4 questions about your health. Your responses will help us understand if you’re getting the care you need so you always feel your best. [Your responses are private, and processed in accordance with our privacy policy. You can opt out of these check-ins at any time by texting “STOP.”] Is it ok if we check in with you every 3 months?

Reply…

1-for Yes

2-for No

2. How would you describe your general health?

Reply…

1-Excellent

2-Very Good

3-Good

4-Fair

5-Poor

3. Now thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 days was your physical health not good?

Reply with a number “0” through “30” or “not sure”

4. Now thinking about your mental health, which includes stress, depressions, and problems with emotions, for how many days during the past 30 days was your mental health not good?

Reply with a number “0” through “30” or “not sure”

5. During the past 30 days, for about how many days did poor physical or mental health keep you from doing your usual activities, such as self-care, work, or recreation?

Reply with a number “0” through “30” or “not sure”

6. ^*Thanks for sharing. We’d love to help you feel better. A medical provider can call you in 2 to 3 days at no cost to you. Is that something you’d like?

Reply…

1-for Yes

2-for No

Asked only of members with fair or poor health (question 2), 15+ physically unhealthy days (question 3), 15+ mentally unhealthy days (question 4), or 15+ physically and mentally unhealthy days combined (questions 3 and 4).

Members were invited to opt in to a recurring “Health check-in” survey every 90 days. This study was limited to a member’s first survey response only. Members with missing age, sex, or zip code data, or with less than 6 months continuous enrollment prior to the survey send date, were excluded.

Outcome variable: Frequent unhealthy days

The primary outcome for this study was a binary variable that categorized members with or without frequent unhealthy days. A member was defined as having frequent unhealthy days if the sum of physically unhealthy days and mentally unhealthy days was 14 days or greater, similar to prior studies.^12,20 The proportion of respondents with frequent physically unhealthy days (14+ days) or mentally unhealthy days (14+ days) was also reported, as well as the mean number of physically unhealthy, mentally unhealthy, and total unhealthy days.

Independent variable: Social deprivation index

The primary independent variable was the social deprivation index (SDI) assigned to the zip code tabulation area. The SDI is a composite score of 7 variables from the 2015–2020 American Community Survey, which was predictive of health care access and health outcomes.²¹ The variables from which the SDI is calculated include percent (%) living below the 100% federal poverty level, % with <12 years of education over 25 years of age, % non-employed between ages 16 and 64 years, % households living in renter-occupied housing units, % households living in crowded housing units, % single parent families with dependents <18 years, and % households with no vehicle. The SDI ranges from 0 to 100, with a higher value representing greater deprivation. Similar to other studies, the SDI was converted into quartiles.^22,23

Covariates: Member characteristics

To account for differences in member characteristics that may be associated with frequent unhealthy days, additional variables were included as covariates in the risk-adjusted model. These included age in years, US region, rural versus urban residency, employer industry, and an indicator of any chronic condition in the 6 months prior to survey completion. The chronic condition indicator was defined as a nonzero case mix index using the Department of Health and Human Services Hierarchical Condition Categories (HHS-HCC) methodology.^24,25

In addition, member-reported race and ethnicity, sexual orientation, and gender identity collected during the application registration process were included. Despite a large proportion of missing values, the variables were included given the uniqueness of these data and the importance of understanding their correlation with frequent unhealthy days. The large proportion of members with missing values was due to 2 reasons: (1) after registering with the application, members had to opt-in to “personalize their care” and may have chosen to not continue; without opting in, members did not see the optional race, sexual orientation, and gender identity questions; and (2) members chose to continue to “personalize their care” but did not respond to the optional 3 questions or chose “prefer not to answer.”

Race and ethnicity and sexual orientation were multiple select fields; members who selected more than 1 option were categorized as “multiple races” or “multiple sexual orientations” to enable statistical testing. Details on the collection of these member-reported data and the distribution of responses are available in a prior study.²⁶

Statistical analysis

Means, standard deviations (SDs), counts, and proportions were used to understand the study population and to report survey responses. Chi-square tests were used to compare unadjusted differences between survey respondents and nonrespondents. A logistic regression that included all variables of interest defined above examined the relationship between SDI quartile and the odds of having frequent unhealthy days, after accounting for confounders. Adjusted odds ratio (aOR) with 95% confidence intervals were calculated.

Results

Sample characteristics

A total of 69,610 members received an SMS-based survey during the study period (Table 2). Members lived in zip codes across all 4 SDI quartiles: 24,815 (35.6%) in the first quartile (lowest deprivation), 23,644 (34.0%) in the second quartile, 13,680 (19.7%) in the third quartile, and 7471 (10.7%) in the fourth quartile (highest deprivation). The majority of members were aged from 26 to 45 years (68.2%). Over half of the sample lived in the West region of the United States (50.4%); the remaining 22.1% were in the South, 17.4% in the Northeast, and 10.2% in the Midwest. Approximately 97.9% lived in urban areas, 17.5% had a positive HHS-HCC case mix score, indicating evidence of chronic conditions, and the most common employer industry was professional, scientific, and technical services (69.3%). The most common race and ethnicity, sexual orientation, and gender identity were null (49.6%, 53.5%, and 53.1%, respectively). Approximately 23.4% of members were White, 16.4% Asian, 4.5% Hispanic, and 3.4% Black. Approximately 41.9% of members were heterosexual, 2.6% were gay or bisexual, and the remaining 1.9% were lesbian, other, queer, or multiple sexual orientations. Over 25% of members were cisgender men, followed by 20.8% cisgender women and 0.9% as nonbinary, other, or transgender.

Table 2.

Study Sample, Survey Respondents and Nonrespondents, Demographic and Clinical Characteristics, May 2023 Through May 2024

	All members who received survey(N = 69,610)		Respondents(N = 6682)		Nonrespondents(N = 62,928)		Respondents vs. nonrespondents P-value^a
	N	%	N	%	N	%	Respondents vs. nonrespondents P-value^a
Social deprivation index quartile
Quartile 1 (lowest deprivation)	24,815	35.6	2504	37.5	22,311	35.5	0.012
Quartile 2	23,644	34.0	2101	31.4	21,543	34.2	<0.000
Quartile 3	13,680	19.7	1328	19.9	12,352	19.6	0.679
Quartile 4 (highest deprivation)	7471	10.7	749	11.2	6722	10.7	0.232
Age
18–25	3026	4.3	228	3.4	2798	4.4	<0.000
26–35	24,237	34.8	1655	24.8	22,582	35.9	<0.000
36–45	23,231	33.4	1905	28.5	21,326	33.9	<0.000
46–55	12,189	17.5	1600	23.9	10,589	16.8	<0.000
56–64	5223	7.5	940	14.1	4283	6.8	<0.000
64+	1704	2.4	354	5.3	1350	2.1	<0.000
Self-reported race and ethnicity
None reported	34,497	49.6	2568	38.4	31,929	50.7	<0.000
Asian	11,406	16.4	926	13.9	10,480	16.7	<0.000
Black	2369	3.4	332	5.0	2037	3.2	<0.000
Hispanic	3123	4.5	390	5.8	2733	4.3	<0.000
Multiple races	1310	1.9	162	2.4	1148	1.8	0.001
Native Hawaiian, American Indian/Native American	137	0.2	17	0.3	120	0.2	<0.000
Other	509	0.7	54	0.8	455	0.7	<0.000
White	16,259	23.4	2233	33.4	14,026	22.3	<0.000
Self-reported sexual orientation
None reported	37,220	53.5	2788	41.7	34,432	54.7	<0.000
Bisexual	788	1.1	113	1.7	675	1.1	<0.000
Gay	1049	1.5	144	2.2	905	1.4	<0.000
Heterosexual	29,200	41.9	3397	50.8	25,803	41.0	<0.000
Lesbian	263	0.4	50	0.7	213	0.3	<0.000
Multiple	541	0.8	97	1.5	444	0.7	<0.000
Other	312	0.4	51	0.8	261	0.4	<0.000
Queer	237	0.3	42	0.6	195	0.3	<0.000
Self-reported gender identity
None	36,940	53.1	2883	43.1	34,057	54.1	<0.000
Cisgender man	17,583	25.3	2043	30.6	15,540	24.7	<0.000
Cisgender woman	14,473	20.8	1633	24.4	12,840	20.4	<0.000
Nonbinary	229	0.3	45	0.7	184	0.3	<0.000
Other	198	0.3	36	0.5	162	0.3	<0.000
Transgender	187	0.3	42	0.6	145	0.2	<0.000
Employer industry
Accommodation and food services, construction, mining, quarrying, and oil and gas extraction, and utilities	401	0.6	49	0.7	352	0.6	0.074
Finance and insurance	13,214	19.0	2008	30.1	11,206	17.8	<0.000
Information	5447	7.8	612	9.2	4835	7.7	<0.000
Manufacturing, retail trade, transportation and warehousing, wholesale trade	2336	3.4	298	4.5	2038	3.2	<0.000
Professional, scientific, and technical services	48,212	69.3	3715	55.6	44,497	70.7	<0.000
Region
Midwest	7067	10.2	907	13.6	6160	9.8	<0.000
Northeast	12,137	17.4	1005	15.0	11,132	17.7	<0.000
South	15,354	22.1	1888	28.3	13,466	21.4	<0.000
West	35,052	50.4	2882	43.1	32,170	51.1	<0.000
Urban/Rural		0.0
Rural	1442	2.1	222	3.3	1220	1.9	<0.000
Urban	68,168	97.9	6460	96.7	61,708	98.1	0.299
HHS-HCC chronic condition indicator^b
No chronic conditions identified	57,430	82.5	5473	81.9	51,957	82.6	0.590
1 or more chronic conditions	12,180	17.5	1209	18.1	10,971	17.4	0.242

P-value from unadjusted chi-square tests comparing actual with expected distribution of respondents.

Department of Health and Human Services Hierarchical Condition Categories (HHS-HCC) assigns a score to chronic conditions identified in medical claims. Members with any chronic conditions during 6 months prior to receiving the survey were categorized as “1 or more chronic conditions”; members with no chronic conditions or incomplete data during that time were categorized as “no chronic conditions identified.”

Respondents differed from nonrespondents

Approximately 10% (N = 6682) of members completed all 4 questions (“respondents”) (Table 2). Respondents differed from nonrespondents. Respondents were more likely to live in the South (28.3% vs. 21.4%) and Midwest (13.6% vs. 9.8%) regions of the United States, in rural areas (3.3% vs. 1.9%), and in zip codes within the 1st quartile of SDI (lowest deprivation) (37.5% vs. 35.5%) and less likely to live in the second quartile (31.4% vs. 34.2%), Ps < 0.001. Respondents were more likely to be over age 45 (43.3% vs. 25.8%), less likely to be employed in the professional, scientific, and technical services industry (55.6% vs. 70.7%), and more likely to have reported their race and ethnicity (61.6% vs. 49.3%), sexual orientation (59.2% vs. 45.3%), and gender identity (56.1% vs. 45.9%), Ps < 0.001. There was no difference in the proportion of members with a positive HHS-HCC case mix score, indicating evidence of chronic conditions, between respondents and nonrespondents.

Unadjusted healthy days and SDI quartile

Almost one-third (29.2%) of the 6682 respondents reported frequent unhealthy days (14 or more physically and mentally unhealthy days in the past 30 days), with a mean (SD) of 10.6 (10.5) days. This was primarily driven by mentally unhealthy days; 20.8% reported frequent mentally unhealthy days (mean 6.9 days, SD = 8.6). Approximately 13.7% of members reported frequent physically unhealthy days (mean 5.3 days, SD = 8.1).

The proportion of members who reported frequent total unhealthy days and mentally unhealthy days increased with higher social deprivation (Fig. 1). Approximately 26.5% of members in the lowest SDI quartile reported frequent total unhealthy days compared with 30.0%, 30.9%, and 32.4% of members in the second, third, and fourth SDI quartiles. A similar pattern was found between SDI quartile and frequent mentally unhealthy days; 18.1% of members in the lowest SDI quartile reported frequent mentally unhealthy days compared with 21.5%, 23.4%, and 23.2% in the second, third, and fourth quartiles, respectively. While the relationship between SDI quartile and frequent physically unhealthy days was similar, differences between the first and fourth quartiles were not statistically significant.

FIG. 1.

Proportion of respondents who reported frequent unhealthy physical, mental, and total days, by social deprivation index quartile and overall, May 2023 through May 2024. (1) SDI = social deprivation index of zip code where member resides. (2) Bars represent the proportion of respondents who reported frequent unhealthy days: 14 or more physically unhealthy days, 14 or more mentally unhealthy days, and 14 total unhealthy days. (3) Overall includes all members, regardless of zip code SDI quartile. (4) Lines indicate standard deviation.

Adjusted odds ratio

After accounting for confounders, members living in zip codes with higher SDI were more likely to report frequent unhealthy days (Table 3). Members who lived in the highest SDI quartile were 1.21 times more likely to experience frequent unhealthy days than members in the lowest SDI quartile (P = 0.047). The higher likelihood of having frequent unhealthy days was also found for members living in zip codes with an SDI in the second and third quartile, relative to the 1st quartile, but were not statistically significant (aOR = 1.12, P = 0.09; aOR = 1.16, P = 0.054).

Table 3.

Adjusted Odds Ratio of Frequent Total Unhealthy Days, by Demographic and Clinical Characteristics, May 2023 Through May 2024

	aOR^a (95% CI)
Social deprivation index quartile
Quartile 1 (lowest deprivation)	1 (reference)
Quartile 2	1.12 (0.98–1.28)
Quartile 3	1.16 (0.997–1.35)
Quartile 4 (highest deprivation)	1.21 (1.002–1.46)
Age (in years)	0.98 (0.978–0.988)
Self-reported race and ethnicity
White	1 (reference)
None reported	0.86 (0.67–1.09)
Asian	0.67 (0.55–0.8)
Black	0.71 (0.54–0.92)
Hispanic	0.78 (0.61–0.996)
Multiple races	0.84 (0.59–1.2)
Native Hawaiian, American Indian/Native American	0.42 (0.12–1.48)
Other	0.7 (0.38–1.3)
Self-reported sexual orientation
Heterosexual	1 (reference)
None reported	0.84 (0.65–1.08)
Bisexual	1.49 (1.01–2.21)
Gay	1.1 (0.76–1.61)
Lesbian	0.84 (0.45–1.59)
Multiple	1.91 (1.22–2.98)
Other	1.42 (0.77–2.61)
Queer	2.16 (1.13–4.11)
Self-reported gender identity
Cisgender man	1 (reference)
None reported	1.57 (1.23–2.00)
Cisgender woman	1.55 (1.33–1.80)
Nonbinary	1.58 (0.82–3.05)
Other	1.08 (0.50–2.35)
Transgender	0.97 (0.47–1.97)
Employer industry
Finance and insurance	1 (reference)
Accommodation and food services, construction, mining, quarrying, and oil and gas extraction, utilities	1.68 (0.93–3.04)
Information	0.87 (0.70–1.08)
Manufacturing, retail trade, transportation and warehousing, and wholesale trade	1.04 (0.79–1.37)
Professional, scientific, and technical services	1.05 (0.91–1.21)
Region
West	1 (reference)
Midwest	0.95 (0.80–1.14)
Northeast	0.89 (0.76–1.05)
South	0.94 (0.81–1.08)
Urban/Rural
Rural	1 (reference)
Urban	1.27 (0.95–1.71)
HHS-HCC chronic condition indicator^b
No chronic conditions identified	1 (reference)
1 or more chronic conditions	1.8 (1.57–2.07)

aOR = adjusted odds ratio of having frequent total unhealthy days from a logistic regression that included all variables included in the table.

Department of Health and Human Services Hierarchical Condition Categories (HHS-HCC) assigns a score to chronic conditions identified in medical claims. Members with any chronic conditions during 6 months prior to receiving the survey were categorized as “1 or more chronic conditions”; members with no chronic conditions or insufficient data during that time were categorized as “no chronic conditions identified.”

CI, confidence interval.

In addition, the odds of frequent total unhealthy days were statistically significantly higher for members who were younger (aOR = 0.98 per additional year of age, P < 0.001), cisgender women (aOR = 1.55, P < 0.001), queer, bisexual, or multiple sexualities (aOR = 2.16, P = 0.019; aOR = 1.49, P = 0.046; aOR = 1.91, P = 0.015), and had a positive HHS-HCC case mix index (aOR = 1.80, P < 0.001). Relative to people who were White, people who were Asian or Black had lower odds of reporting frequent unhealthy days (aOR = 0.67, P < 0.001 and aOR = 0.70, P = 0.011, respectively). US region, urban residence, and employer industry were not statistically significant.

Discussion

To the author’s knowledge, this is the first study to look at the association of zip code SDI and frequent unhealthy days among a commercially insured population accessing virtual care and navigation services through a digital application. This study demonstrates that even among adults with employer-provided insurance and health benefits, variation in frequent unhealthy days by zip code remains.

The sample of adults in this study was unique in that all individuals had registered for virtual care and navigation services offered through an employer. It is therefore not surprising that there was a much higher proportion who reported frequent unhealthy days than in other published literature. For example, only 12.3% of commercially insured BRFSS 2021 respondents reported frequent mentally unhealthy days and 6.1% reported frequent physically unhealthy days compared with 20.8% and 13.7% of members in this study, respectively.²⁷

This study shows that adults with the greatest zip code socioeconomic deprivation were 1.21 times more likely to report frequent unhealthy days than those living in zip codes with the lowest deprivation, even after accounting for individual characteristics that have previously been associated with self-reported health. The direction and magnitude of the relationship are similar to prior literature that examined the relationship between healthy days and county and state social deprivation.^12,15,16 One study found Medicare Advantage enrollees living in counties in the top quartile for 10 SDOH factors had a 6%–12% higher risk of frequent unhealthy days compared to those in counties in the lowest quartile.¹² Another study based on the 2016 BRFSS found that states with the highest poverty rates, unemployment rates, and lowest median income had higher predicted physically and mentally unhealthy days after controlling for an individual’s SDOH factors.¹⁶

Similar to prior literature, this study also found that sexually diverse members and cisgender women had higher odds of worse HRQoL.^18,19 Specifically queer, bisexual, or individuals with multiple sexual identities had higher odds of frequent total unhealthy days than heterosexuals. While many differences by gender identity were not statistically significant due to low sample size, cisgender women had greater likelihood of frequent unhealthy days than cisgender men. As the population of sexually and gender diverse members who complete the SMS survey increases, additional insights can be gleaned on the differences in healthy days for these populations which can contribute to existing literature.

This study suggests that using SDI to identify members who may benefit from additional support could be valuable for employers and their benefit partners, in the absence of other data. Another study found that using SDI alone identifies only half of individuals with SDOH needs.²⁸ However, employers and their partners typically lack data on the SDOH needs of individuals, especially before the member engages with the health care system. Therefore, they could proactively identify individuals living in areas with a high SDI who may benefit from additional support. Specifically, the rise in digital health, which by definition has no geographic boundaries, provides a unique opportunity to offer services to members living in areas with greatest SDI.²⁹

This study also demonstrates the value in capturing self-reported HRQoL, a predictor of future spend and health status, from members and patients through technology-driven solutions.^30
–32 First, frequent unhealthy days vary by member demographic characteristics, underscoring the importance of collecting HRQoL as a part of a broader health equity strategy. Second, real-time member-reported HRQoL can help health care organizations identify who may benefit from clinical outreach. For organizations, this real-time data capture is more timely and actionable than the commonly used claims-based algorithms, which may not receive the relevant data until months after the health care encounter. Finally, HRQoL collected prior to engaging with a member can serve as a baseline against which an organization may measure the overall effectiveness of health care interventions in improving health outcomes.

This study has several limitations. First, the results may lack generalizability outside the study sample for 2 main reasons. The study relied on a nonprobability sample of commercially insured adults who chose to engage with a digital application. However, >6000 respondents were geographically dispersed across 49 states (all but Hawaii) and the District of Columbia and were not limited to a single health system or health insurer. In addition, this study had a high nonresponse rate similar to other survey-based studies of Healthy Days.^14,33
–35 While research indicates higher response rates do not always reduce nonresponse bias, the findings should be interpreted cautiously outside of this study sample.^36,37 Moving forward, Included Health will shift from using SMS texts for surveys to collecting Healthy Days within the digital health application. This change is expected to greatly improve response rates among members. Second, the data available for this study were not inclusive of all factors that may influence healthy days, such as an individual’s job type or income. Third, less than one-quarter of respondents had a positive HHS-HCC case mix index, which limited the ability to account for heterogeneity among respondents with chronic conditions. Fourth, more than one-third of the sample had missing race, ethnicity, sexual orientation, or gender identity. Despite this, the inclusion of these variables is a strength of this study. It contributes to the existing literature that has found lower self-reported health among sexual and gender diverse communities and is based on a larger sample size than many prior studies. In addition, given the strong relationship between sexual orientation and gender identity with healthy days, it was important to control for these characteristics in the regression model.

Conclusion

This study found that higher zip code socioeconomic deprivation is positively correlated with frequent unhealthy days among commercially insured adults using health care services through a digital application. In the absence of other data, this highlights an opportunity for employers and their partners to use zip code SDI when considering which member segments to prioritize for engagement, particularly when intervention resources are limited. This study also demonstrates that member-reported HRQoL varies widely by demographic characteristics, even among those engaging with an employer-offered health benefit. Overall, these findings underscore the potential for zip code SDI and self-reported HRQoL data to guide targeted health interventions and optimize resource allocation for improving health outcomes.

Footnotes

Acknowledgment

The authors want to thank the entire Included Health team for the inclusion of healthy days, race and ethnicity, sexual orientation, and gender identity as part of the member data collection strategy. This made the study possible.

Authors’ Contributions

J.M.: Writing—original draft, project administration, methodology, supervision, conceptualization, and resources. O.-J.M.B.: Writing—review and editing, methodology, and data curation. N.L.: Methodology, formal analysis, visualization, and writing—review and editing. T.T.: Writing—review and editing and conceptualization. A.Y.: Writing—review and editing, resources, and conceptualization.

Data Availability Statement

The data collected and analyzed for this study are not publicly available.

Author Disclosure Statement

All authors are employed by Included Health.

Funding Information

The authors received no funding for this study.

References

Dwyer-Lindgren

, Bertozzi-Villa

, Stubbs

, et al. Inequalities in life expectancy among us counties, 1980 to 2014: Temporal trends and key drivers. JAMA Intern Med., 2017; 177(7):1003–1011; doi: 10.1001/jamainternmed.2017.0918

Shanahan

, Subramanian

, Burdick

, Monuteaux

, Lee

, Fleegler

. Association of neighborhood conditions and resources for children with life expectancy at birth in the US. JAMA Netw Open., 2022; 5(10):e2235912; doi: 10.1001/jamanetworkopen.2022.35912

Tan

, Hinman

, Abdel Magid

, Nelson

, Odden

. Association between income inequality and county-level COVID-19 cases and deaths in the US. JAMA Netw Open., 2021; 4(5):e218799; doi: 10.1001/jamanetworkopen.2021.8799

Khan

, Vaughan

, Harrington

, et al. US county–level variation in preterm birth rates, 2007-2019. JAMA Netw Open., 2023; 6(12):e2346864; doi: 10.1001/jamanetworkopen.2023.46864

Hood

, Gennuso

, Swain

, Catlin

. County health rankings: Relationships between determinant factors and health outcomes. Am J Prev Med, 2016; 50(2):129–135; doi: 10.1016/j.amepre.2015.08.024

Hostetter

, Klein

, Commonwealth Fund . Using patient-reported outcomes to improve health care quality. Available from: https://www.commonwealthfund.org/publications/newsletter-article/using-patient-reported-outcomes-improve-health-care-quality [Accessed August 7, 2024].

Phillips

Jr , Etz

. CMS’s universal foundation measures are not universally good for primary care. Health Affairs Forefront; 2023.

HealthyPeople.gov. Foundation health measures. General Health Measures. 2017. Available from: https://wayback.archive-it.org/5774/20170103200800/https://www.healthypeople.gov/2020/about/Foundation-Health-Measures [Accessed August 26, 2024].

Ochiai

, Kigenyi

, Sondik

, et al. Healthy People 2030 leading health indicators and overall health and well-being measures: Opportunities to assess and improve the health and well-being of the nation. J Public Health Manag Pract, 2021; 27(Suppl 6):S235–S241; doi: 10.1097/PHH.0000000000001424

10.

Centers for Disease Control and Prevention. Measuring Healthy Days. Atlanta, Georgia: CDC, November 2000. Available from: https://archive.cdc.gov/#/details?url=https://www.cdc.gov/hrqol/pdfs/mhd.pdf [Accessed July 15, 2024].

11.

Moriarty

, Zack

, Kobau

. The Centers for Disease Control and Prevention’s Healthy Days measures – population tracking of perceived physical and mental health over time. Health Qual Life Outcomes, 2003; 1:37; doi: 10.1186/1477-7525-1-37

12.

Cordier

, Song

, Cambon

, et al. A bold goal: More healthy days through improved community health. Popul Health Manag, 2018; 21(3):202–208; doi: 10.1089/pop.2017.0142

13.

Rudy

, McNamara

, Goldberg

, Parker

, Nash

. A call for consistent measurement across the social determinants of health industry landscape. Popul Health Manag, 2022; 25(5):698–701.

14.

Shih

, Aylward

, Kunkle

, Graziani

. Health-related quality of life among members using an on demand behavioral health platform: Pilot observational study. JMIR Form Res, 2022; 6(7):e35352.

15.

Jia

, Moriarty

, Kanarek

. County-level social environment determinants of health-related quality of life among US adults: A multilevel analysis. J Community Health, 2009; 34(5):430–439; doi: 10.1007/s10900-009-9173-5

16.

Liu

, Jiang

, Fan

, Thompson

, Ding

, Ni

. Investigating the associations between socioeconomic factors and unhealthy days among adults using zero-inflated negative binomial regression. SAGE Open, 2023; 13(3):21582440231194163.

17.

Downing

, Przedworski

. Health of transgender adults in the U.S., 2014-2016. Am J Prev Med, 2018; 55(3):336–344; doi: 10.1016/j.amepre.2018.04.045

18.

Lagos

. Looking at population health beyond “male” and “female”: Implications of transgender identity and gender nonconformity for population health. Demography., 2018; 55(6):2097–2117.

19.

Gonzales

, Henning-Smith

. Health disparities by sexual orientation: Results and implications from the behavioral risk factor surveillance system. J Community Health, 2017; 42(6):1163–1172; doi: 10.1007/s10900-017-0366-z

20.

Slabaugh

, Shah

, Zack

, et al. Leveraging health-related quality of life in population health management: The case for Healthy Days. Popul Health Manag, 2017; 20(1):13–22; doi: 10.1089/pop.2015.0162

21.

Social deprivation index (SDI). Robert Graham Center - policy studies in family medicine & primary care. November 29, 2021. Available from: https://www.graham-center.org/maps-data-tools/social-deprivation-index.html [Accessed November 5, 2018].

22.

Bevan

, Nasir

, Rajagopalan

, Al-Kindi

. Socioeconomic deprivation and premature cardiovascular mortality in the United States. Mayo Clin Proc, 2022; 97(6):1108–1113; doi: 10.1016/j.mayocp.2022.01.018

23.

Patel

, Krasnow

, Long

, Shirey

, Dickert

, Morris

. Excess 30-day heart failure readmissions and mortality in black patients increases with neighborhood deprivation. Circ Heart Fail, 2020; 13(12):e007947; doi: 10.1161/CIRCHEARTFAILURE.120.007947

24.

United States Census Bureau. North American Industry Classification System. September 3, 2024. Available from: https://www.census.gov/naics/?58967?yearbck=2022

25.

Kautter

, Pope

, Ingber

, et al. The HHS-HCC risk adjustment model for individual and small group markets under the Affordable Care Act. Medicare Medicaid Res Rev, 2014; 4(3):mmrr2014–m004-03-a03; doi: 10.5600/mmrr2014-004-03-a03

26.

Marshall

, Zhang

, Green

. Adults’ willingness to report sexual orientation and gender identity when registering for a digital health application: A cross-sectional quantitative study. PLoS One, 2023; 18(11):e0292739; doi: 10.1371/journal.pone.0292739

27.

Centers for Disease Control & Prevention. Behavioral risk factor surveillance system web enabled analysis tool. August 15, 2024. Available from: https://nccd.cdc.gov/weat/#/crossTabulation/selectYear

28.

Cottrell

, Hendricks

, Dambrun

, et al. Comparison of community-level and patient-level social risk data in a network of community health centers. JAMA Netw Open., 2020; 3(10):e2016852; doi: 10.1001/jamanetworkopen.2020.16852

29.

Zimlichman

, Nicklin

, Aggarwal

, Bates

. Health care 2030: The coming transformation. NEJM Catalyst Innovations in Care Delivery . 2021. Available from: https://catalyst.nejm.org/doi/full/10.1056/CAT.20.0569

30.

Antol

, Hagan

, Nguyen

, et al. Change in self-reported health: A signal for early intervention in a Medicare population. Healthc (Amst), 2022; 10(1):100610; doi: 10.1016/j.hjdsi.2021.100610

31.

Cordier

, Slabaugh

, Havens

, et al. A health plan’s investigation of healthy days and chronic conditions. Am J Manag Care, 2017; 23(10):e323–e330.

32.

McNamara

, Rudy

, Rogers

, et al. The cost of unhealthy days: A new value assessment. Popul Health Manag, 2024; 27(5):307–311; doi: 10.1089/pop.2024.0102

33.

Casebeer

, Antol

, Hopson

, et al. Retracted: Using the healthy days measure to assess factors associated with poor health-related quality of life for patients with metastatic breast, lung, or colorectal cancer enrolled in a Medicare Advantage health plan. Popul Health Manag, 2018; 22(2):186; doi: 10.1089/pop.2018.0024

34.

Gliske

, Welsh

, Braughton

, Waller

, Ngo

. Telehealth services for substance use dissimples during the COVID-19 pandemic: Longitudinal assessment of intensive outpatient programming and data collection practices. JMIR Ment Health, 2022; 9(3):e36263; doi: 10.2196/36263

35.

Cole

, Zbikowski

, Renda

, Wallace

, Dobbins

, Bogard

. Examining changes in healthy days after health coaching. Am J Health Promot, 2019; 33(5):774–777; doi: 10.1177/0890117118816286

36.

Lee

. Propensity score adjustment as a weighting scheme for volunteer panel web surveys. J Offici Stati, 2006; 22(2):329.

37.

National Academies of Sciences, Engineering, and Medicine. Nonresponse in social science surveys: A research agenda. The National Academies Press: Washington, DC; 2013.