Association of low Hourly Wages in Middle Age With Faster Memory Decline in Older Age

Evidence From the Health and Retirement Study

Katrina L. Kezios; Adina Zhang; Soohyun Kim; Peiyi Lu; M. Maria Glymour; Tali Elfassy; Adina Zeki Al Hazzouri

Am J Epidemiol. 2022;191(12):2051-2062.

Abstract and Introduction

Abstract

Little research has investigated the long-term relationship between low wages and memory decline, despite the growing share of low-wage workers in the US labor market. Here, we examined whether cumulative exposure to low wages over 12 years in midlife is associated with memory decline in later life. Using 1992–2016 data from the Health and Retirement Study, we analyzed data from 2,879 individuals born in 1936–1941 using confounder-adjusted linear mixed-effects models. Low-wage work was defined as an hourly wage lower than two-thirds of the federal median wage for the corresponding year and was categorized into "never," "intermittent," and "sustained" based on wages earned from 1992 to 2004. Memory function was measured at each study visit from 2004 to 2016 via a memory composite score. The confounder-adjusted annual rate of memory decline among "never" low-wage earners was −0.12 standard units (95% confidence interval: −0.13, −0.10). Compared with this, memory decline among workers with sustained earning of low midlife wages was significantly faster (β_{time×sustained} = −0.014, 95% confidence interval: −0.02, −0.01), corresponding to an annual rate of −0.13 standard units for this group. Sustained low-wage earning in midlife was significantly associated with a downward trajectory of memory performance in older age. Enhancing social policies to protect low-wage workers may be especially beneficial for their cognitive health.

Introduction

In the United States, the federal minimum wage has remained $7.25 per hour since 2009.^[1] While economic growth has increased since then, wage and salary growth for employees—particularly those in low-wage jobs—has slowed over time,^[2–4] and the minimum wage has not kept up with inflation.^[5] Regularly earning low wages is potentially a public health problem,^[6] as lower socioeconomic status (SES) is a determinant of numerous adverse health outcomes.^[7–10] However, limited literature has examined the relationship between low wages and health.^[11,12]

Of particular concern is how low wages may affect cognitive health. Structural forces such as race- and sex-based discrimination and educational inequality drive certain groups of individuals into sustained low-wage employment, producing a low-wage workforce disproportionately composed of, for example, women, Black workers, and workers with low education;^[13] prior research has also documented inequalities in aging-related disorders among these groups.^[14–18] However, whether sustained earning of low wages affects cognitive function is an open question. Low-wage jobs—particularly those also considered "low-skill"^[19,20]—are often characterized by factors that may harm cognition, including toxic exposures,^[21–23] injuries,^[21,22,24] and work-related stress.^[25,26] Low-wage jobs have been previously associated with health outcomes such as depressive symptoms,^[27] obesity,^[28] and hypertension,^[29,30] which are risk factors for cognitive aging.^[31–34] However, to our knowledge, no prior studies have examined the specific relationship between low wages during working years and later-life cognitive functioning.

To fill this gap, we estimated how low-wage employment history during the peak earning years in midlife affected subsequent longitudinal changes in memory function in older age among US adults in the Health and Retirement Study (HRS). We hypothesized that a prolonged history of low wages would be associated with faster memory decline in old age. Further, because women make up a disproportionate share of the low-wage workforce^[13] and have been shown to have a higher lifetime risk for cognitive disorders such as Alzheimer disease,^[14–16] we explored whether the relationship varied by sex.

1 2 3 4

Next Section

Abstract and Introduction
Methods
Results
Discussion
References

Table 1. Baseline Sample Characteristics ^a According to Category of Low-Wage Employment History, 1936–1941 Birth Cohort of the Health and Retirement Study, 1992–2016
Sample Characteristic	Total Sample (n = 2,879)		Low-Wage Exposure History (1992–2004)
	Total Sample (n = 2,879)		Never Exposed (n = 1,749)		Intermittent Exposure (n = 851)		Sustained Exposure (n = 279)
	No.	%	No.	%	No.	%	No.	%
Age at exposure baseline (1992), years^b	53.0 (1.8)		53.0 (1.8)		53.1 (1.8)		53.1 (1.8)
Age at beginning of cognitive follow-up (2004), years^b,c	64.8 (1.8)		64.8 (1.8)		64.9 (1.8)		64.9 (1.8)
Duration of education, years^b	13.0 (2.6)		13.5 (2.6)		12.5 (2.4)		11.1 (2.3)
No. of exposure time points with wage information^b	4.7 (1.7)		4.7 (1.7)		5.0 (1.6)		3.8 (1.7)
No. of memory assessments^b	5.7 (1.8)		5.7 (1.9)		5.8 (1.8)		5.4 (2.0)
Household wealth, dollars^d,e	111,701 (42,423–225,283)		130,715 (56,251–263,484)		94,096 (31,622–179,242)		49,971 (7,494–137,181)
Female sex	1,422	49.4	729	41.7	473	55.6	220	78.9
Born in the South	1,069	37.1	592	33.8	338	39.7	139	49.8
Race/ethnicity^f
Non-Hispanic White	2,318	80.5	1,461	83.5	661	77.7	196	70.3
Non-Hispanic Black	490	17.0	258	14.8	161	18.9	71	25.4
Other	71	2.5	30	1.7	29	3.4	12	4.3
Parental duration of education
<12 years	1,264	43.9	705	40.3	392	46.1	167	59.9
≥12 years	1,492	51.8	987	56.4	420	49.4	85	30.5
Missing data	123	4.3	57	3.3	39	4.6	27	9.7
Occupational level
High-skilled	940	32.7	709	40.5	203	23.9	28	10.0
Low-skilled	1,665	57.8	931	53.2	566	66.5	168	60.2
Missing data	274	9.5	109	6.2	82	9.6	83	29.7
Self-employed
Yes	413	14.3	213	12.2	154	18.1	46	16.5
No	2,191	76.1	1,427	81.6	614	72.2	150	53.8
Missing data	275	9.6	109	6.2	83	9.8	83	29.7
Marital status
Married/cohabiting	2,204	76.6	1,357	77.6	658	77.3	189	67.7
Other	591	20.5	334	19.1	178	20.9	79	28.3
Missing data	84	2.9	58	3.3	15	1.8	11	3.9
Retirement status at cognitive follow-up^g
Completely retired	1,125	39.1	779	44.5	255	30.0	91	32.6
Other	1,499	52.1	813	46.5	529	62.2	157	56.3
Missing data	255	8.9	157	9.0	67	7.9	31	11.1

^aUnless otherwise specified, all baseline covariates were measured in 1992.
^bValues are expressed as mean (standard deviation).
^cEither age at outcome period baseline (2004) or age at which the first memory assessment occurred during the outcome period (2004–2016).
^dValues are expressed as median (interquartile range).
^eHousehold wealth, including secondary residence and equivalizing for household size by dividing wealth by the square root of the number of household members, was calculated in 2016 dollars and adjusted for inflation.
^fFor analytical purposes (to preserve statistical power), the "non-Hispanic Black" and "other" categories were combined into 1 group.
^gEither retirement status at outcome period baseline (2004) or retirement status when the first memory assessment occurred during the outcome period (2004–2016).

Table 2. Distribution of Hourly Wages ^a, Overall and by Sex, in the 1936–1941 Birth Cohort of the Health and Retirement Study, 1992–2016
Year	Federal Median Hourly Wage, dollars	Definition of Low Wage, dollars^c	Overall (n = 3,803)	Sex
Year	Federal Median Hourly Wage, dollars	Definition of Low Wage, dollars^c	Overall (n = 3,803)	Male (n = 1,913)	Female (n = 1,890)
1992	9.40	6.27	11.54 (7.50–17.33)	14.49 (9.61–20.38)	9.08 (6.35–13.60)
1994	9.74	6.49	12.00 (7.91–18.27)	15.00 (10.00–21.92)	9.00 (6.00–14.00)
1996	10.62	7.08	12.19 (8.00–19.24)	15.20 (10.00–2.93)	10.00 (7.00–15.00)
1998	11.83	7.89	12.75 (8.33–20.19)	16.03 (10.26–24.50)	10.12 (7.33–15.72)
2000	13.18	8.79	13.34 (8.75–22.12)	16.67 (10.08–26.49)	10.90 (7.85–16.67)
2002	13.31	8.87	13.30 (9.00–22.47)	17.00 (10.07–29.34)	11.00 (8.12–17.31)
2004	13.98	9.32	12.97 (8.60–22.50)	15.00 (9.62–26.92)	11.42 (8.50–18.01)

Abbreviations: OEWS, Occupational Employment and Wage Statistics; SSA, Social Security Administration.
^aData are presented as median (interquartile range) in 2016 dollars, adjusted for inflation.
^bThe federal median hourly wage was based on data obtained from either the Bureau of Labor Statistics' OEWS (for the years 2002–2004) or the SSA's Average Wage Index (for the years 1992–2000). For 2002 and 2004, we used the reported federal hourly median wages across all occupations from the respective OEWS databases. However, from 1992 to 2000, this value was not reported in the OEWS databases. Thus, for 1992–2000, we instead estimated the federal median hourly wage from the average net compensation for these years, converting it to dollars per hour, from the SSA database. We used the average net compensation from the SSA rather than the median net compensation because when converted to hourly wages, using average versus median net compensation resulted in values that were more closely aligned with those reported in the OEWS databases.
^c"Low wage" was calculated as two-thirds of the federal median wage.

Table 3. Association Between Low-Wage Employment History (1992–2004) and Cognitive Trajectories (2004–2016) in the 1936–1941 Birth Cohort of the Health and Retirement Study ^a
Model Parameter	Model 1^b		Model 2^c		Model 3^d		Excess Years of Cognitive Aging (Over 10 Years)^e
Model Parameter	β	95% CI	β	95% CI	β	95% CI	Excess Years of Cognitive Aging (Over 10 Years)^e
Low-wage exposure history
Never	0	Referent	0	Referent	0	Referent
Intermittent	−0.003	−0.034, 0.027	0.014	−0.016, 0.045	0.014	−0.016, 0.045
Sustained	−0.009	−0.058, 0.040	0.013	−0.037, 0.063	0.019	−0.033, 0.071
Time (annual change)^f	−0.065^g	−0.067, −0.062	−0.121^g	−0.134, −0.108	−0.118^g	−0.134, −0.102
Low-wage exposure × time^f
Intermittent × time	−0.005	−0.009, 0.000	−0.001	−0.006, 0.003	−0.002	−0.007, 0.002	0.2
Sustained × time	−0.026^g	−0.034, −0.019	−0.015^g	−0.023, −0.008	−0.014^g	−0.022, −0.006	1.2

Abbreviation: CI, confidence interval.
^aThe sample size for all models was n = 2,879. The sample size was n = 1,749 for never low-wage exposure, n = 851 for intermittent low-wage exposure, and n = 279 for sustained low-wage exposure. The memory score outcome is presented in standard units, where lower scores indicate worse cognitive function.
^bModel 1 results were unadjusted, and the model only includes terms for low-wage exposure, number of time points with wage information reported by the participant over the exposure time period, age at beginning of cognitive follow-up (2004), time since the start of cognitive follow-up, and the interaction between exposure and time.
^cModel 2 included all terms from model 1 and adjusted for sex, race/ethnicity, education, place of birth, marital status, parental education, and each covariate's interaction with time.
^dModel 3 additionally adjusted for employment in lower-skilled occupations, household wealth in 1992 (equivalized for household size), retirement status at first memory assessment during the outcome period, and each of these covariates' interactions with time.
^eFollowing the procedures outlined by Weuve et al. (47), using the coefficients from model 3, we assumed that the annual rate of change in memory scores among never low-wage earners represented "cognitive aging if one never experienced low wages" (this is captured by the main effect term for time). Then, we divided the estimated differences in the annual rates of change in memory function for "intermittent" versus "never" and "sustained" versus "never" low-wage exposure by the annual rate of change among never low-wage earners and multiplied by 10, to obtain excess years of cognitive aging per 10-year period. Positive values indicate that individuals with a history of "intermittent" or "sustained" low-wage earning experienced more (or "excess") cognitive aging over a 10-year period than "never" low-wage earners, while negative values indicate that they experienced less cognitive aging.
^fThe estimate for "time" represents the estimated slope of the memory trajectory for "never" low-wage earners; the estimated slopes of the memory trajectories for "intermittent" and "sustained" low-wage earners (for a given time period) can be obtained by summing the main effects for time and the interaction between low-wage exposure history and time for these groups.
^g P < 0.001.

Table 4. Fully Adjusted Associations Between Low-Wage Employment History (1992–2004) and Cognitive Trajectories (2004–2016), by Sex, in the 1936–1941 Birth Cohort of the Health and Retirement Study ^a
Model Parameter	Sex
	Male			Female
	β	95% CI^b	Excess Years of Cognitive Aging (Over 10 Years)^c	β	95% CI^b	Excess Years of Cognitive Aging (Over 10 Years)^c
Low-wage exposure history
Never	0	Referent		0	Referent
Intermittent	0.023	−0.008, 0.055		0.006	−0.040, 0.051
Sustained	−0.027	−0.103, 0.048		0.026	−0.038, 0.091
Time (annual change)^d	−0.127^e	−0.150, −0.104		−0.094^e	−0.119, −0.070
Low-wage exposure × time^d
Intermittent × time	−0.005	−0.012, 0.001	0.4	0.000	−0.007, 0.006	0.0
Sustained × time	−0.028^f	−0.046, −0.010	2.2	−0.010	−0.020, 0.000	1.1

Abbreviation: CI, confidence interval.
^aThe sample size was n = 1,457 for males and n = 1,422 for females. For males, the sample size was n = 1,020 for never low-wage exposure, n = 378 for intermittent low-wage exposure, and n = 59 for sustained low-wage exposure. For females, the sample size was n = 729 for never low-wage exposure, n = 473 for intermittent low-wage exposure, and n = 220 for sustained low-wage exposure. The memory score outcome is presented in standard units, where lower scores indicate worse cognitive function.
^bAssociations were adjusted for the number of time points with wage information reported by the participant over the exposure time period, age at the beginning of cognitive follow-up (2004), race/ethnicity, education, place of birth, marital status, parental education, employment in lower-skilled occupations, household wealth in 1992 (equivalized for household size), retirement status at first memory assessment during the outcome period, and each covariate's interaction with time.
^cFollowing the procedures outlined by Weuve et al.,⁴⁷ using the coefficients from model 3, we assumed that the annual rate of change in memory scores among never low-wage earners represented "cognitive aging if one never experienced low wages" (this is captured by the main effect term for time). Then, we divided the estimated differences in the annual rates of change in memory function for "intermittent" versus "never" and "sustained" versus "never" low-wage exposure by the annual rate of change among never low-wage earners and multiplied by 10, to obtain excess years of cognitive aging per 10-year period. Positive values indicate that individuals with a history of "intermittent" or "sustained" low-wage earning experienced more (or "excess") cognitive aging over a 10-year period than "never" low-wage earners, while negative values indicate that they experienced less cognitive aging.
^dThe estimate for "time" represents the estimated slope of the memory trajectory for "never" low-wage earners; the estimated slopes of the memory trajectories for "intermittent" and "sustained" low-wage earners (for a given time period) can be obtained by summing the main effects for time and the interaction between low-wage exposure history and time for these groups.
^e P < 0.001.
^f P < 0.01.

processing....

Association of low Hourly Wages in Middle Age With Faster Memory Decline in Older Age

Evidence From the Health and Retirement Study

Abstract and Introduction

Abstract

Introduction

Tables

Tables

Tables

Tables

References

Authors and Disclosures

Authors and Disclosures

Email This

Feedback