Abstract
Background
The 21st century has seen a decline in employment rates in the US at the same time that it has experienced a historically unprecedented rise in the numbers of adults under criminal justice system control. Both low employment and high incarceration have posed serious challenges for public health.
Methods
Using data from the National Survey on Drug Use and Health from 2002–2021, we estimated employment rates by community supervision status. Variations by sociodemographic subgroups were explored as well as correlations between employment and a range of mental and behavioural health variables.
Results
Those on probation were twice as likely as those not to live in poverty. They experienced higher rates of poor mental and behavioural health, including three times the rate of substance use. Employment rates varied little by community supervision status. Health risk factors were associated with more risk and protective factors did less to mitigate risk for those under community supervision.
Conclusions
Despite the range of adversities faced by individuals under criminal justice system control, their employment rates are remarkably close to those not. Despite near equivalent involvement in the labour force, this population has substantially poorer health and substantially reduced likelihood of escaping poverty.
Keywords: mental health, probation, employment, criminal justice
Introduction
A substantial body of research demonstrates a strong relationship between employment status and health, including mental and behavioural health, with causal relationships in both directions. 1 With the persistent decline in US employment rates since 2000, the public health consequences of a less-employed adult population has become an area of concern, particularly for vulnerable populations, such as adults with criminal justice system contact and have indicated a need for a fuller understanding of the health of this population.2,3
An estimated 2 million people are incarcerated in the US with an estimated 10 million cycling through jails each year4,5 For those who experience incarceration, the physical removal from the workforce, with increasingly lengthy placements in correctional facilities and lifelong criminal records, has been found to have direct impacts on both employment and health across the life course.6 The incarcerated population is majority low-income men of colour, a population already vulnerable to poor health and low employment rates.7 A less studied but comparably vulnerable—and much larger—criminal justice-involved population consists of those under community supervision.8 The emergence of mass incarceration in the US at the end of the 20th century coincided with what has come to be called ‘mass supervision’—4.6 million people under correctional control in the community who will have lifetime public criminal records.9
Limited research exists on the employment rates of individuals under community supervision10 and even less exists on their health.11 According to public health research and theory, employment is a social determinant of health.12 Lack of employment has been found to have a strong relationship with psychological distress and depressive symptoms.13 Mortality rates have been found to be 60% higher for the unemployed compared to the employed.14 Risks for suicide and substance use disorder are higher amongst the unemployed.15 Likewise, substance use disorder increases risk for unemployment.16
We hypothesized that, given the combined pressures of disease burden, community supervision stipulations and criminal labelling, those on probation would have significantly lower employment rates than those not. Additionally, we hypothesized that this population would have poorer mental and behavioural health outcomes with associated poorer employment outcomes. We present findings from a series of analyses on employment status and a range of mental and behavioural health correlates using a national dataset, comparing those on probation in the past year to the general population.
Methods
Data and sample
We used the National Survey on Drug Use and Health (NSDUH), an annual cross-sectional household survey conducted across all 50 US states and the District of Columbia. The NSDUH is a nationally representative survey that uses a multi-stage probability sampling design to obtain a sample of civilian, non-institutionalized US residents, ages 12 years and older. It offers a comprehensive collection of data related to substance use, mental health, and other health-related topics. The detailed methodology and design of the NSDUH are available elsewhere.17 For the present study, we used data collected between 2002 and 2021 and included a total of 778 162 adults, ages 18 years and above, of which 23 537 were on probation in the past year.
Measures
To measure ‘community supervision status’, respondents were asked if they had been on probation at any time in the past 12 months. ‘Employment status’ was coded as either not employed/not in labour force or employed full/part time. Employment status was measured in NSDUH each year from 2002 to 2021, but the methodology in the data collection of this variable changed in 2015 and again in 2020. Specifically, in 2015, the employment questions changed from being interviewer-administered to being self-administered. In 2020, the NSDUH methodology changed again with the addition of web-based interviewing.
‘Behavioural health’ variables included measures of substance use and other health risk behaviours. To measure substance use we used four self-reported measures of cigarette use, binge alcohol use, cannabis use and illicit drug use. All items were measures of past-year use except for binge alcohol use, which was measured for the past month. To measure other health risk behaviours, we used past-year driving under the influence, drug selling, theft and arrest as well as a risk propensity score. A risk propensity score was created with methods consistent with previous studies.18,19 Respondents were asked the following two questions: ‘How often do you like to test yourself by doing something a little risky?’ and ‘How often do you get a real kick out of doing things that are a little dangerous?’ The scores were then summed to create an ordinal measure indicating low (0), medium (1) or high (2) risk propensity.
‘Mental health’ was measured based on past-year depressive episodes and serious psychological distress. For depressive episodes, one or more past-year major depressive episodes was coded as ‘yes’ following criteria in the DSM-IV.20 Serious psychological distress was measured using the K6 scale and indicated by a score over 12.21
For detailed descriptions of variable construction, please see Appendix.
Data analysis
Statistical analyses were weighted to account for the NSDUH’s stratified cluster sampling design and carried out using Stata 18SE. First, we estimated the employment rates between 2002 and 2021 by community supervision status. Second, we examined the linear trend of employment rates within sociodemographic subgroups by community supervision status. For this step, we used survey-adjusted logistic regression analysis, as suggested by the Centers for Disease Control and Prevention’s guidelines for secular trend analysis. This approach assesses change at all time points, not just at first and last, using record-level data. It incorporates survey design and sample weights, adjusting for correlations between years. Given the breakages in comparability of the employment variable, we also tested the linear trends between 2002–2006, 2010–2014, 2015–2019 and 2020–2021 separately, in addition to 2002–2021. Third, we examined the correlations between employment and a range of variables, comparing those on probation and those not. Specifically, we conducted a series of logistic regressions to examine the association between each of the behavioural and mental health variables, whilst controlling for sociodemographic factors and survey year. Correlation analyses were limited to the most recent survey years with continued comparability (2015–2019). Supplemental analyses of interactions between community supervision status and correlates were conducted and provided results consistent with our stratified approach.
Results
Employment trends, 2002–2021
In 2002, employment rates for both groups of study participants were ~70%. Over the next two decades, employment rates fell for both groups, with those on probation in the past year experiencing a steeper decline (see Fig. 1). In 2021, ~51% of the probation group were employed compared to 57% of those not on probation. Table 1 provides detailed results from our trend analyses of the four time periods studied. These analyses revealed statistically significant (P < 0.05) downward trends in employment for both groups, with those on probation declining, on average, three percentage points per time period [trend adjusted odds ratio (AOR) = 0.97, 95% confidence interval (CI) = 0.96–0.98] compared to those not on probation declining, on average, two percentage points per time period (trend AOR = 0.98, 95% CI = 0.98–0.98). The interaction between survey year and probation status was significant (AOR = 0.99, 95% CI = 0.98–1), indicating that the two groups’ trends significantly differed.
Fig. 1.
Year-by-year percentage employed among U.S. adults on probation and not on probation, National Survey on Drug Use and Health (NSDUH) 2002-2021 (N=778,162).
Table 1.
Employment rates of US adults on probation and not on probation, National Survey on Drug Use and Health 2002–2021 (N = 778 162)
| Adults not on probation (N = 754 625) % Employed full/part time (95% CI) |
Adults on probation (N = 23 537) % Employed full/part time (95% CI) |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 2002–2006 | 2010–2014 | 2015–2019 | 2020–2021 | Trend AOR (95% CI) | 2002–2006 | 2010–2014 | 2015–2019 | 2020–2021 | Trend AOR (95% CI) | |
| Full sample | 68.22% (67.79–68.65) | 64.35% (63.94–64.75) | 62.61% (62.21–63.00) | 57.90% (57.25–58.56) |
0.98 (0.98–0.98) |
70.27% (68.47–72.00) | 60.75% (58.60–62.857) | 59.31% (57.44–61.15) | 50.27% (44.24–56.29) |
0.97 (0.96–0.98) |
| Age | ||||||||||
| 18–29 | 75.49 (75.02–75.96) |
70.27 (69.81–70.73) |
71.26 (70.75–71.76) |
66.05 (64.81–67.27) |
0.97 (0.97–0.98) |
71.49 (69.72–73.20) |
64.03 (61.86–66.15) |
63.26 (60.97–65.49) |
63.11 (55.83–69.85) |
0.97 (0.96–0.98) |
| 30–49 | 83.08 (82.62–83.53) |
80 (79.56–80.42) |
78.25 (77.92–78.57) |
74.58 (73.67–75.46) |
0.97 (0.97–0.98) |
70.58 (66.94–73.98) |
62.07 (58.36–65.64) |
63.28 (60.44–66.03) |
52.18 (44.86–59.41) |
0.97 (0.95–0.98) |
| 50 or older | 49.41 (48.57–50.25) |
49.28 (48.57–49.99) |
47.35 (46.72–47.99) |
42.49 (41.4–43.58) |
0 .98 (0.98–0.99) |
63.36 (53.76–72.00) |
45.45 (38.11–52.99) |
42.40 (36.01–49.06) |
31.05 (18.06–47.92) |
0.94 (0.91–0.98) |
| Gender | ||||||||||
| Female | 61.6 (61.06–62.14) |
58.98 (58.39–59.57) |
57.63 (57.13–58.13) |
53.16 (52.21–54.11) |
0.98 (0.98–0.98) |
56.08 (52.17–59.91) |
51.31 (47.62–54.99) |
51.03 (47.53–54.53) |
42.75 (31.42–54.88) |
0.98 (0.96–0.99) |
| Male | 75.53 (74.97–76.08) |
70.24 (69.72–70.75) |
68.02 (67.5–68.54) |
63.01 (61.68–64.32) |
0 .97 (0.97–0.98) |
75.30 (73.35–77.16) |
64.72 (62.18–67.18) |
62.96 (60.65–65.21) |
54.12 (47.24–60.84) |
0.96 (0.95–0.97) |
| Race/ethnicity | ||||||||||
| Non-Hispanic White | 67.94 (67.43–68.45) |
64.30 (63.82–64.79) |
62.99 (62.48–63.49) |
58.45 (57.63–59.26) |
0.98 (0.98–0.99) |
71.86 (69.42–74.17) |
60.50 (57.35–63.57) |
60.79 (58.19–63.33) |
56.03 (48.32–63.46) |
0.98 (0.96–0.99) |
| Non-Hispanic Black/African American | 66.84 (65.78–67.89) |
60.42 (59.25–61.58) |
59.26 (58.26–60.25) |
54.97 (52.79–57.14) |
0.97 (0.97–0.98) |
63.48 (58.01–68.62) |
52.55 (48.28–56.79) |
53.74 (49.17–58.25) |
41.40 (28.58–55.50) |
0.96 (0.94–0.98) |
| Hispanic | 70.16 (68.95–71.33) |
66.69 (65.69–67.67) |
62.71 (62.00–63.42) |
57.36 (55.21–59.48) |
0.97 (0.96–0.97) |
72.27 (68.08–76.10) |
69.43 (64.49–73.97) |
60.05 (55.31–64.6) |
43.60 (29.91–58.34) |
0.95 (0.93–0.97) |
| Other | 70.09 (68.12–71.99) |
66.26 (64.69–67.79) |
64.24 (63.10–65.37) |
59.09 (56.06–62.05) |
0.98 (0.97–0.98) |
67.49 (59.49–74.60) |
60.69 (52.44–68.38) |
60.01 (52.51–67.08) |
37.13 (17.70–61.86) |
0.99 (0.96–1.01) |
| Family income | ||||||||||
| Living in poverty | 44.51 (43.45–45.58) |
42.29 (41.45–43.14) |
39.61 (38.77–40.46) |
33.03 (31.34–34.76) |
0.98 (0.97–0.98) |
49.89 (45.67–54.10) |
43.75 (40.31–47.26) |
41.53 (37.95–45.19) |
30.44 (22.54–39.68) |
0.98 (0.96–0.99) |
| Family income up to 2× poverty line | 56.75 (55.78–57.72) |
55.69 (54.83–56.55) |
52.00 (51.29–52.71) |
47.29 (45.74–48.85) |
0.98 (0.98–0.98) |
71.66 (68.21–74.86) |
61.98 (58.09–65.72) |
61.27 (57.15–65.25) |
57.33 (45.18–68.67) |
0.97 (0.96–0.99) |
| Family income more than 2× poverty line | 75.00 (74.49–75.49) |
71.78 (71.34–72.22) |
70.59 (70.14–71.04) |
66.30 (65.40–67.19) |
0.98 (0.98–0.98) |
81.29 (78.78–83.57) |
74.92 (71.79–77.82) |
70.86 (68.18–73.40) |
68.21 (59.81–75.57) |
0.96 (0.94–0.97) |
| Education | ||||||||||
| Less than high school | 48.96 (47.97–49.96) |
44.80 (43.73–45.87) |
41.08 (40.21–41.96) |
36.72 (34.79–38.69) |
0.98 (0.97–0.98) |
64.18 (60.55–67.65) |
51.28 (48.21–54.34) |
46.75 (43.25–50.29) |
39.32 (26.37–53.96) |
0.96 (0.95–0.98) |
| High school graduate | 65.35 (64.52–66.18) |
58.85 (58.15–59.56) |
54.55 (53.93–55.16) |
48.92 (47.61–50.22) |
0.98 (0.97–0.98) |
68.16 (65.01–71.14) |
61.85 (58.71–64.89) |
60.19 (57.18–63.13) |
49.56 (40.77–58.37) |
0.98 (0.97–0.99) |
| Some college/associate degree | 73.52 (72.88–74.15) |
67.99 (67.27–68.71) |
66.02 (65.48–66.55) |
60.34 (59.23–61.43) |
0.98 (0.97–0.98) |
76.52 (72.49–80.13) |
63.24 (59.30–67.02) |
63.91 (60.36–67.32) |
52.87 (41.96–63.51) |
0.96 (0.94–0.98) |
| College graduate | 78.65 (77.87–79.41) |
75.41 (74.80–76.01) |
73.91 (73.35–74.46) |
70.96 (69.91–71.99) |
0.98 (0.98–0.99) |
88.35 (81.51–92.88) |
81.95 (74.46–87.61) |
72.33 (64.08–79.29) |
75.55 (61.85–85.48) |
0.94 (0.89–0.98) |
| Marital status | ||||||||||
| Married | 69.85 (69.27–70.44) |
67.12 (66.58–67.65) |
64.62 (64.11–65.12) |
59.93 (58.74–61.1) |
0.98 (0.98–0.98) |
72.77 (68.34–76.78) |
65.89 (61.01–70.46) |
62.34 (59.24–65.34) |
54.07 (41.05–66.57) |
0.96 (0.94–0.98) |
| Divorced/separated/ widowed | 56.14 (55.08–57.2) |
51.27 (50.3–52.25) |
49.11 (48.26–49.95) |
44.96 (43.39–46.55) |
0.98 (0.98–0.99) |
66.77 (61.42–71.72) |
52.77 (47.04–58.42) |
51.32 (46.70–55.92) |
41.62 (29.44–54.91) |
0.96 (0.94–0.98) |
| Never married | 74.21 (73.66–74.76) |
68.80 (68.24–69.35) |
68.32 (67.81–68.83) |
63.17 (62.03–64.30) |
0.97 (0.97–0.98) |
70.45 (68.41–72.42) |
61.93 (59.45–64.36) |
61.60 (59.08–64.05) |
52.79 (45.04–60.42) |
0.97 (0.96–0.98) |
Note: all estimates adjusted for the NSDUH’s complex sampling design. Bolded trend AOR indicates significant linear trend at P < 0.05.
Demographic and socioeconomic correlates (2002–2021 and 2015–2019)
Both our linear trend analyses (2002–2021) and our logistic regression analyses of correlates (2015–2019) revealed relatively small but consistent patterns in which demographic and socioeconomic factors had weaker relationships to employment for those on probation than for those not. Whilst the results are fully presented in Tables 1 and 2, we present a selection of these correlates here in order to summarize our findings.
Table 2.
Sociodemographic correlates of being employed full/part time amongst US adults on probation and not on probation, National Survey on Drug Use and Health 2015–2019 (N = 213 869)
| Adults not on probation (N = 209 054) | Adults on probation (N = 4815) | |||||||
|---|---|---|---|---|---|---|---|---|
| n (%) | Employment rate (95% CI) | AOR | 95% CI | n (%) | Employment rate (95% CI) | AOR | 95% CI | |
| Age | ||||||||
| 18–29 | 86,703 (41%) | 71.26% (70.74–71.77) | 1.00 | – | 2717 (56%) | 63.26% (60.93–65.53) | 1.00 | – |
| 30–49 | 78,783 (38%) | 78.25% (77.92–78.58) | 1.17 | 1.12–1.23 | 1769 (37%) | 63.28% (60.39–66.07) | 1.03 | 0.87–1.22 |
| 50 or older | 43,568 (21%) | 47.35% (46.71–48.00) | 0.25 | 0.24–0.27 | 329 (7%) | 42.40% (35.91–49.17) | 0.42 | 0.30–0.59 |
| Gender | ||||||||
| Female | 112,817 (54%) | 57.63% (57.12–58.14) | 1.00 | – | 1639 (34%) | 51.03% (47.47–54.59) | 1.00 | – |
| Male | 96,237 (46%) | 68.02% (67.49–68.55) | 1.53 | 1.48–1.59 | 3176 (66%) | 62.96% (60.61–65.25) | 1.59 | 1.31–1.93 |
| Race/ethnicity | ||||||||
| Non-Hispanic White | 126,202 (60%) | 62.99% (62.47–63.5) | 1.00 | – | 2504 (52%) | 60.79% (58.14–63.37) | 1.00 | – |
| Non-Hispanic Black/African American | 26,110 (12%) | 59.26% (58.24–60.27) | 1.12 | 1.06–1.20 | 864 (18%) | 53.74% (49.09–58.33) | 0.89 | 0.69–1.15 |
| Hispanic | 35,929 (17%) | 62.71% (61.98–63.43) | 1.23 | 1.19–1.28 | 909 (19%) | 60.05% (55.23–64.68) | 1.11 | 0.83–1.48 |
| Other | 20,813 (10%) | 64.24% (63.08–65.39) | 0.87 | 0.81–0.92 | 538 (11%) | 60.01% (52.37–67.20) | 1.19 | 0.87–1.63 |
| Family income | ||||||||
| Living in poverty | 37,182 (18%) | 39.61% (38.75–40.47) | 1.00 | - | 1592 (33%) | 41.53% (37.89–45.26) | 1.00 | – |
| Family income up to 2× poverty line | 44,358 (21%) | 52.00% (51.28–52.73) | 1.94 | 1.85–2.02 | 1325 (28%) | 61.27% (57.07–65.32) | 2.08 | 1.59–2.73 |
| Family income more than 2× poverty line | 125,354 (60%) | 70.59% (70.13–71.04) | 4.15 | 3.95–4.37 | 1875 (39%) | 70.86% (68.13–73.44) | 2.91 | 2.33–3.61 |
| Education | ||||||||
| Less than high school | 26,462 (13%) | 41.08% (40.20–41.97) | 1.00 | – | 1179 (24%) | 46.75% (43.19–50.35) | 1.00 | – |
| High school graduate | 55,055 (26%) | 54.55% (53.92–55.17) | 1.55 | 1.47–1.63 | 1797 (37%) | 60.19% (57.13–63.18) | 1.46 | 1.18–1.81 |
| Some college/associates | 70,316 (34%) | 66.02% (65.47–66.56) | 2.17 | 2.07–2.27 | 1535 (32%) | 63.91% (60.30–67.37) | 1.58 | 1.21–2.07 |
| College graduate | 57,221 (27%) | 73.91% (73.35–74.47) | 2.78 | 2.64–2.93 | 304 (6%) | 72.33% (63.92–79.40) | 2.11 | 1.40–3.18 |
| Marital status | ||||||||
| Married | 86,979 (42%) | 64.62% (64.10–65.13) | 1.00 | – | 890 (18%) | 62.34% (59.19–65.39) | 1.00 | – |
| Divorced/separated/widow | 28,926 (14%) | 49.11% (48.25–49.97) | 0.91 | 0.87–0.96 | 804 (17%) | 51.32% (46.62–56.00) | 0.74 | 0.57–0.96 |
| Never been married | 93,149 (45%) | 68.32% (67.80–68.84) | 0.97 | 0.93–1.02 | 3121 (65%) | 61.60% (59.04–64.10) | 0.89 | 0.70–1.14 |
| Urbanicity | ||||||||
| Large metro | 94,515 (45%) | 64.96% (64.40–65.51) | 1.00 | – | 1855 (39%) | 61.28% (58.17–64.3) | 1.00 | – |
| Small metro | 73,837 (35%) | 60.56% (59.92–61.19) | 0.94 | 0.91–0.97 | 1754 (36%) | 59.28% (56.39–62.11) | 0.98 | 0.82–1.17 |
| Non-metro | 40,702 (19%) | 57.62% (56.79–58.44) | 1.00 | 0.96–1.05 | 1206 (25%) | 54.26% (49.49–58.94) | 0.93 | 0.72–1.20 |
Notes: data from years 2015–2019 are pooled. AOR were estimated with year and sociodemographic factors adjusted for (age, gender, race/ethnicity, household income, education, marital status and urbanicity). Bolded AOR are statistically significant at P < 0.05. All estimates adjusted for the NSDUH’s complex sampling design.
Age
For those not on probation, those age 50 and older experienced the ‘least’ decline compared to other age groups, falling seven percentage points (trend AOR = 0.98, 95% CI = 0.98–0.99). Meanwhile, for their counterparts who were on probation, their employment rates fell the ‘most’ compared to other age groups, dropping by 30 percentage points (Trend AOR = 0.94, 95% CI = 0.91–0.98). Within-group comparisons using logistic regression analyses revealed additional patterns in the relationship between age and employment for both groups. Using data from the study period from 2015–2019, we found that for those 30–49 years old (roughly, the prime working age group) who were not on probation, 78% were employed. Compared to those under 30, this group had a 17% greater likelihood of employment (AOR = 1.17, 95% CI = 1.12–1.23). For those 30–49 on probation, however, there was no statistically significant difference in rate of employment when compared to those under 30. The employment rate was 63% for both age groups.
Race
For, participants who were not on probation, those who identified as Non-Hispanic White experienced the least decline over the four time periods examined in our trend analyses (trend AOR = 0.98, 95% CI = 0.98–0.99) compared to those who identified as Non-Hispanic Black (Trend AOR = 0.97, 95% CI = 0.97–0.98) or Hispanic (Trend AOR = 0.97, 95% CI = 0.96–0.97). Whilst Non-Hispanic White participants saw a decline in employment at about two percentage points per time period, Non-Hispanic Black and Hispanic participants saw a decline of three per time period. This pattern was similar for those on probation, however, the difference by race/ethnicity was larger.
Family income
In our trend analysis, we found that each income bracket of those not on probation experienced about the same rate of decline in employment, about two percentage points per time period. For those on probation, a higher income was associated with greater decline in employment rates over time. Living in poverty was almost twice as common for those on probation as those not (33% compared to 18%).
Education
In our trend analyses, for those with a college degree, employment rates fell by about eight percentage points for those not on probation between time period one and time period four (trend AOR = 0.98, 95% CI = 0.98–0.99). At the same time, employment rates fell by 13 percentage points for those with a college degree who were on probation (trend AOR = 0.94, 95% CI = 0.89–0.98). Similarly, in our logistic regression analyses, which, again, focused exclusively on the study period from 2015–2019, we found that those not on probation with some college were 2.2 times more likely to be employed than those who did not finish high school (AOR = 2.17, 95% CI = 2.07–2.27). Those on probation with some college, by comparison, were 1.6 times more likely to be employed than their counterparts who did not finish high school (AOR = 1.58, 95% CI = 1.21–2.07).
Behavioural and mental health correlates
We explored a range of behavioural and mental health correlates in our logistic regression analyses of the study period 2015–2019 which are presented in full in Table 3.
Table 3.
Behavioural and mental health correlates of being employed full/part time amongst community supervised and non-community supervised US adults, National Survey on Drug Use and Health 2015–2019 (N = 213 869)
| Non-community supervised (N = 209 054) | Community supervised (N = 4815) | |||||||
|---|---|---|---|---|---|---|---|---|
| n (%) | Employment rate (95% CI) | AOR | 95% CI | n (%) | Employment rate (95% CI) | AOR | 95% CI | |
| Past-year cigarette use | ||||||||
| No | 152,763 (73%) | 62.38% (61.92–62.83) | 1.00 | – | 1623 (34%) | 59.73% (56.66–62.73) | 1.00 | – |
| Yes | 56,291 (27%) | 63.39% (62.79–63.99) | 1.12 | 1.08–1.16 | 3192 (66%) | 59.07% (56.67–61.42) | 1.01 | 0.85–1.19 |
| Past-month binge drinking | ||||||||
| No | 144,342 (69%) | 58.3% (57.86–58.74) | 1.00 | – | 2898 (60%) | 55.58% (53.00–58.13) | 1.00 | – |
| Yes | 64,712 (31%) | 74.72% (74.10–75.32) | 1.65 | 1.59–1.71 | 1917 (40%) | 65.43% (62.49–68.25) | 1.37 | 1.12–1.67 |
| Past-year marijuana use | ||||||||
| No | 165,530 (79%) | 61.16% (60.70–61.62) | 1.00 | – | 2792 (58%) | 59.27% (56.83–61.66) | 1.00 | – |
| Yes | 43,524 (21%) | 70.7% (70.01–71.38) | 1.17 | 1.12–1.22 | 2023(42%) | 59.38% (56.35–62.33) | 0.86 | 0.72–1.02 |
| Past-year illicit drug use | ||||||||
| No | 185,312 (89%) | 61.97% (61.55–62.39) | 1.00 | – | 3207 (67%) | 60.22% (57.66–62.72) | 1.00 | – |
| Yes | 23,742 (11%) | 69.38% (68.58–70.16) | 1.08 | 1.04–1.13 | 1608 (33%) | 57.27% (54.91–59.60) | 0.81 | 0.69–0.95 |
| Major depressive episode | ||||||||
| No episodes | 188,032 (91%) | 63% (62.58–63.42) | 1.00 | – | 4044 (85%) | 60.67% (58.39–62.91) | 1.00 | – |
| One or more episodes | 18,818 (9%) | 59.09% (57.85–60.32) | 0.77 | 0.73–0.82 | 697 (15%) | 51.26% (45.67–56.82) | 0.68 | 0.51–0.90 |
| Serious psychological distress | ||||||||
| No | 176,974 (85%) | 62.8% (62.39–63.21) | 1.00 | – | 3450 (72%) | 60.83% (58.41–63.20) | 1.00 | – |
| Yes | 32,080 (15%) | 61.06% (60.20–61.91) | 0.82 | 0.79–0.85 | 1365 (28%) | 54.88% (51.09–58.61) | 0.78 | 0.64–0.95 |
| Driving under influence* | ||||||||
| No | 142,195 (86%) | 60.67% (60.23–61.11) | 1.00 | – | 2603 (71%) | 58.08% (55.60–60.52) | 1.00 | – |
| Yes | 22,685 (14%) | 79.74% (78.80–80.66) | 1.75 | 1.63–1.88 | 1051 (29%) | 66.77% (63.74–69.67) | 1.24 | 1.02–1.52 |
| Drug selling | ||||||||
| No | 204,403 (98%) | 62.75% (62.34–63.16) | 1.00 | – | 4339 (91%) | 59.97% (57.91–61.99) | 1.00 | – |
| Yes | 3953 (2%) | 57.07% (54.64–59.46) | 0.77 | 0.69–0.87 | 437 (9%) | 54.09% (47.79–60.26) | 0.74 | 0.54–1.02 |
| Theft | ||||||||
| No | 206,399 (99%) | 62.76% (62.35–63.17) | 1.00 | – | 4457 (93%) | 60.09% (58.15–61.99) | 1.00 | – |
| Yes | 2055 (1%) | 49.07% (46.13–52.02) | 0.61 | 0.53–0.71 | 329 (7%) | 50.35% (44.19–56.51) | 0.67 | 0.50–0.92 |
| Arrest | ||||||||
| No | 204,273 (98%) | 62.78% (62.37–63.19) | 1.00 | – | 2025 (52%) | 63.11% (60.03–66.09) | 1.00 | – |
| Yes | 3958 (2%) | 54.32% (52.14–56.48) | 0.70 | 0.63–0.78 | 1895 (48%) | 55.74% (52.43–58.99) | 0.78 | 0.62–0.98 |
| Risk propensity score | ||||||||
| Low | 158,027 (76%) | 60.54% (60.08–61.00) | 1.00 | – | 2872 (60%) | 56.26% (53.95–58.55) | 1.00 | – |
| Medium | 23,874 (11%) | 69.65% (68.69–70.59) | 1.15 | 1.09–1.22 | 753 (16%) | 66.37% (59.96–72.22) | 1.34 | 1.02–1.76 |
| High | 25,711 (12%) | 74.4% (73.53–75.26) | 1.29 | 1.22–1.36 | 1175 (24%) | 63.57% (59.76–67.21) | 1.02 | 0.81–1.28 |
Notes. Data from years 2015–2019 are pooled. AOR were estimated with year and sociodemographic factors adjusted for (age, gender, race/ethnicity, household income, education, marital status and urbanicity). Bolded AOR are statistically significant at P < .05. All estimates adjusted for the NSDUH’s complex sampling design.
Data from years 2016–2019 were pooled because this variable was not comparable before and after 2016.
Behavioural health
Smoking was far more common amongst those on probation than those not (66% compared to 27%). Marijuana was also used by about twice as many of those under supervision (42% compared to 21%) as were illicit drugs (33% compared to 11%). Amongst study participants who were not on probation, those who indicated that they used substances had higher odds of employment than those who indicated they did not. This relationship was statistically significant across all types of substance use measured and ranged from an 8% greater likelihood for those using illicit substances (AOR = 1.08, 95% CI = 1.04–1.13) to a 65% greater likelihood for those who engaged in binge drinking (AOR 1.65, 95% CI = 1.59–1.71). For those on probation, however, there were no statistically significant relationships between cigarette or marijuana use and employment. Moreover, use of illicit drugs for those on probation had an inverse relationship with employment, with those indicating illicit drug use having 20% less likelihood of employment compared to those who did not indicate drug use (AOR = 0.81, 95% CI = 0.69–0.95).
Mental health
For those on probation, mental health symptoms were associated with worse employment outcomes than for those not on probation. The differences were small but consistent across measures. For those experiencing depressive episodes, employment odds were 23% lower for those not on probation (AOR = 0.77, 95% CI = 0.73–0.82) compared to 32% lower for those on probation (AOR = 0.68, 95% CI = 0.51–0.90). For those experiencing serious psychological distress, employment odds were 18% lower for those not on probation (AOR = 0.82, 95% CI = 0.79–0.85) compared to 22% lower for those on probation (AOR = 0.78, 95% CI = 0.64–0.95).
Discussion
Main findings of this study
Importantly, we found that those on probation in the past 12 months experienced depressive episodes and psychological distress twice as often as those not. We also found that whilst poor mental health was associated with lower employment rates for both study groups, the employment rates were lower for those on probation. This group also used tobacco twice as often (with 2/3 indicating they smoked) and used illicit substances three times as often (with 1/3 indicating illicit drug use). Drug use had minimal relationship to employment status for those not on probation but reduced odds of employment by 20% for those who were. In other words, those under criminal justice system control were more likely to have poorer health and their poorer health was more likely to be associated with lack of employment.
What is already known on this topic
Our study is consistent with the literature demonstrating that poverty is the defining characteristic of the American criminal justice system.22 Over 60% of our sample under community supervision had family incomes that were below twice the federal poverty line. For comparison, in 2019, ~26% of the US population had family incomes in this range.23 Of note, amongst our study participants on probation, those in the highest income group (above twice the poverty line) were at greatest risk for unemployment, suggesting that even when escape from poverty is possible, it remains precarious. It has been well established that poverty harms health.24
What this study adds
Our study contributes to the literature establishing that the intersection of poverty and the criminal justice system harms health in unique ways,3 not only for those in jail and prison but also for those under criminal justice control in the community. Further, our study finds that the relationship between employment and health is more fraught for those under community supervision, likely compounding the challenges for this population to escape poverty. We found that, in general, just as risk factors were associated with more risk for those on probation, protective factors offered less mitigation against risk. Specifically, those with a college degree on probation experienced the most dramatic decline in employment rates over the 20-year study period. Education, it appears, did not hold the same promise for those under criminal justice system control. Our findings indicated depressed employment rates across the life course for those on probation, with prime working-age status offering little advantage and the final working years having high risk of disengagement with the labour market, especially after the Great Recession (2007–2009) when the employment rate for this group fell to 30%.
Given the range of barriers those on probation face when attempting to enter and remain in the job market, we found it remarkable that our study participants under supervision were employed as much as they were. In fact, we found that both study groups were employed at equal rates prior to the Great Recession. When disaggregated by race, we found that Whites in both study groups were employed at equal rates over the ‘entire’ study. For participants of colour under supervision, their employment rates fell more sharply during the recession and did not recover as much after.
Limitations of this study
Limitations of this study include the use of self-report data which rely on the recall of study participants. Of note, multiple studies have demonstrated convergence between official records and self-report of justice-involved individuals.25–27 Second, NSDUH data are the result of a series of cross-sectional surveys. This does not allow for any causal claims or temporal ordering. Third, for our analyses, we collapsed data across multiple time periods (2002–2006, 2010–2014, 2015–2019, and 2020–2021). This allowed us to reach adequate sample sizes for testing a wide range of variables but limited our ability to capture nuances of annual or seasonal changes. Fourth, it should be noted that arrest, whilst categorized in this study as a behavioural health risk and commonly used in research on this population, has limited validity as a measure of an individual’s behaviour as law enforcement policies and practices are involved and vary widely by jurisdiction.28 Lastly, this study was only able to look at individuals on probation at some point in the prior 12 months. The relationship between important related factors, such as length of probation, time spent incarcerated, or type of conviction, and employment could not be considered.
Conclusion
The dramatic growth in the numbers of adults under criminal justice system control in the US has been found to have a wide range of negative public health consequences. Researchers have found that communities with high rates of incarceration have poorer health,29 less access to education,30 and employment31 and shorter life expectancies.32 Recent research has begun to explore whether these incarceration effects are also seen with community supervised populations.11 Our findings indicate that those on probation carry a high health burden, that they experience distinct disadvantage in the labour market associated with this health burden, and that protective factors such as education and income offer less protection. We believe that this points to the importance of holistic approaches to services for this population. For example, mental and behavioural health assessment and support services may need to be integrated into employment-specific services in order for employment services to be effective and employment training and support services may need to be integrated into mental and behavioural health services. Given the complex health needs of this population, it will be important for policy makers, researchers, and practitioners to look outside the criminal justice field for guidance on responding to the service needs of this population (e.g. nursing,33 psychiatry,34 addiction35) and to avoid repeating the harms of mass incarceration.
Supplementary Material
Maria Morrison, PhD, MSW
Audrey Hang Hai, PhD, MSW
Yohita Shraddha Bandaru, MSW
Christopher P. Salas-Wright, PhD, MSW
Michael G. Vaughn, PhD
Contributor Information
Maria Morrison, Saint Louis University, School of Social Work, 1 North Grand Boulevard, St. Louis, MO 63103, USA.
Audrey Hang Hai, Tulane University, School of Social Work, 127 Elk Place, New Orleans, LA 70112, USA.
Yohita Shraddha Bandaru, Saint Louis University, School of Social Work, 1 North Grand Boulevard, St. Louis, MO 63103, USA.
Christopher P Salas-Wright, Boston College, School of Social Work, 140 Commonwealth Ave., Chestnut Hill, MA 02467, USA.
Michael G Vaughn, Saint Louis University, School of Social Work, 1 North Grand Boulevard, St. Louis, MO 63103, USA.
Funding
The project described was supported by grant number T32MH019960 from the National Institute of Mental Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health or the National Institutes of Health.
Data availability statement
Data from the National Study on Drug Use and Health (NSDUH) is publicly available and can be accessed at https://datatools.samhsa.gov/#/.
References
- 1. Pratap P, Dickson A, Love M. et al. Public health impacts of underemployment and unemployment in the United States: exploring perceptions, gaps and opportunities. Int J Environ Res Public Health 2021;18(19):10021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Harding DJ, Western B, Sandelson JA. From supervision to opportunity: reimagining probation and parole. Ann Am Acad Pol Soc Sci 2022;701(1):8–25. [Google Scholar]
- 3. Cloud DH, Garcia-Grossman IR, Armstrong A. et al. Public health and prisons: priorities in the age of mass incarceration. Annu Rev Public Health 2023;44:407–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kang-Brown J, Montagnet C, Heiss J. People in Jail and Prison in 2020. Vera Institute of Justice, New York, 2021. [Google Scholar]
- 5. Zeng Z. Jail inmates in 2016. Bureau of Justice Statistics, Washington, DC, 2018. [Google Scholar]
- 6. Western B, Sirois C. Racialized re-entry: labour market inequality after incarceration. Soc Forces 2019;97(4):1517–42. [Google Scholar]
- 7. Chetty R, Stepner M, Abraham S. et al. The association between income and life expectancy in the United States, 2001-2014. JAMA 2016;315(16):1750–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Morrison M, Hai AH, Bandaru YS. et al. Opioid misuse and associated health risks among adults on probation and parole: prevalence and correlates 2015–2020. Subst Use Misuse Published online 2023;59:20–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Schiraldi V. Mass Supervision: Probation, Parole, and the Illusion of Safety and Freedom. The New Press, New York, 2023. [Google Scholar]
- 10. Capece J. Community supervision and employment. Ann Am Acad Pol Soc Sci 2022;701(1):61–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Wildeman C, Goldman AW, Wang EA. Age-standardized mortality of persons on probation, in jail, or in state prison and the general population, 2001–2012. Public Health Rep 2019;134(6):660–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Thornton RLJ, Glover CM, Cené CW. et al. Evaluating strategies for reducing health disparities by addressing the social determinants of health. Health Aff 2016;35(8):1416–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Perreault M, Power N, Touré EH. et al. Transitional employment and psychological distress: a longitudinal study. Psychiatric Quarterly 2020;91(3):735–47. [DOI] [PubMed] [Google Scholar]
- 14. Roelfs DJ, Shor E, Davidson KW. et al. Losing life and livelihood: a systematic review and meta-analysis of unemployment and all-cause mortality. Soc Sci Med 2011;72(6):840–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Milner A, Page A, Lamontagne AD. Cause and effect in studies on unemployment, mental health and suicide: a meta-analytic and conceptual review. Psychol Med 2014;44(5):909–17. [DOI] [PubMed] [Google Scholar]
- 16. Henkel D. Unemployment and substance use: a review of the literature (1990–2010). Curr Drug Abuse Rev 2011;4(1):4–27. [DOI] [PubMed] [Google Scholar]
- 17. Substance Abuse and Mental Health Services Administration . What Is NSDUH? RTI International, Research Triangle Park, NC, 2021. https://nsduhweb.rti.org/respweb/homepage.cfm (22 February 2023, date last accessed).
- 18. Hai AH, Carey KB, Vaughn MG. et al. Simultaneous alcohol and marijuana use among college students in the United States, 2006–2019. Addict Behav Rep 2022;16:100452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Vaughn MG, Nelson EJ, Salas-Wright CP. et al. Racial and ethnic trends and correlates of non-medical use of prescription opioids among adolescents in the United States 2004–2013. J Psychiatr Res 2016;73(1):17–24. 10.1016/j.jpsychires.2015.11.003. [DOI] [PubMed] [Google Scholar]
- 20. American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders, 4th edn. American Psychiatric Association, Washington, DC, 2000. [Google Scholar]
- 21. Kessler RC, Barker PR, Colpe LJ. et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry 2003;60(2):184–9. 10.1001/archpsyc.60.2.184. [DOI] [PubMed] [Google Scholar]
- 22. Kurlychek MC, Johnson BD. Cumulative disadvantage in the American criminal justice system. Annual Review of Criminology 2019;2(1):291–319. [Google Scholar]
- 23. Semega J, Kollar M, Shrider EA. et al. Income and Poverty in the United States: 2019, U.S. Census Bureau, Washington, DC, 2020.
- 24. Bor J, Cohen GH, Galea S. Population health in an era of rising income inequality: USA, 1980–2015. The Lancet 2017;389(10077):1475–90. 10.1016/S0140-6736(17)30571-8. [DOI] [PubMed] [Google Scholar]
- 25. Dubow EF, Huesmann LR, Boxer P. et al. Childhood predictors and age 48 outcomes of self-reports and official records of offending. Crim Behav Ment Health 2014;24(4):291–304. 10.1002/cbm.1929. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Farrington DP, Ttofi MM, Crago RV. et al. Prevalence, frequency, onset, desistance and criminal career duration in self-reports compared with official records. Crim Behav Ment Health 2014;24(4):241–53. 10.1002/cbm.1930. [DOI] [PubMed] [Google Scholar]
- 27. Pollock W, Menard S, Elliott DS. et al. It’s official: predictors of self-reported vs. officially recorded arrests. J Crim Justice 2015;43(1):69–79. 10.1016/j.jcrimjus.2014.12.004. [DOI] [Google Scholar]
- 28. Fogliato R, Xiang A, Lipton Z. et al. On the validity of arrest as a proxy for offense: race and the likelihood of arrest for violent crimes. In: AIES 2021 - Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society. Association for Computing Machinery, New York, 2021. 10.1145/3461702.3462538 [DOI] [Google Scholar]
- 29. Hatzenbuehler ML, Keyes K, Hamilton A. et al. The collateral damage of mass incarceration: risk of psychiatric morbidity among nonincarcerated residents of high-incarceration neighborhoods. Am J Public Health 2015;105(1):138–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Hagan J, Foster H. Intergenerational educational effects of mass imprisonment in America. Sociol Educ 2012;85(3):259–86. 10.1177/0038040711431587. [DOI] [Google Scholar]
- 31. Simes JT. Place after prison: Neighborhood attainment and attachment during reentry. J Urban Aff 2019;41(4):443–63. 10.1080/07352166.2018.1495041. [DOI] [Google Scholar]
- 32. Wildeman C, Wang EA. Mass incarceration, public health, and widening inequality in the USA. The Lancet 2017;389(10077):1464–74. 10.1016/S0140-6736(17)30259-3. [DOI] [PubMed] [Google Scholar]
- 33. Utomo SF, Dewi IP, Seman N. et al. Holistic rehabilitation in action: a data-driven analysis of the impact on quality of life among individuals with psychoactive substance abuse. J Holist Nurs 2024;42(25):135–43. [DOI] [PubMed] [Google Scholar]
- 34. Hayes A, Herlinger K, Paterson L. et al. The neurobiology of substance use and addiction: evidence from neuroimaging and relevance to treatment. BJPsych Advances 2020;26(6):367–78. [Google Scholar]
- 35. Babor TF, McRee BG, Kassebaum PA. et al. Screening, brief intervention, and referral to treatment (SBIRT): toward a public health approach to the management of substance abuse. Alcohol/Drug Screening and Brief Intervention 2023;9(1):7–30. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data from the National Study on Drug Use and Health (NSDUH) is publicly available and can be accessed at https://datatools.samhsa.gov/#/.