Psychological Distress in Solitary Confinement: Symptoms, Severity, and Prevalence in the United States, 2017–2018

Abstract

Objectives. To specify symptoms and measure prevalence of psychological distress among incarcerated people in long-term solitary confinement.

Methods. We gathered data via semistructured, in-depth interviews; Brief Psychiatric Rating Scale (BPRS) assessments; and systematic reviews of medical and disciplinary files for 106 randomly selected people in solitary confinement in the Washington State Department of Corrections in 2017. We performed 1-year follow-up interviews and BPRS assessments with 80 of these incarcerated people, and we present the results of our qualitative content analysis and descriptive statistics.

Results. BPRS results showed clinically significant symptoms of depression, anxiety, or guilt among half of our research sample. Administrative data showed disproportionately high rates of serious mental illness and self-harming behavior compared with general prison populations. Interview content analysis revealed additional symptoms, including social isolation, loss of identity, and sensory hypersensitivity.

Conclusions. Our coordinated study of rating scale, interview, and administrative data illustrates the public health crisis of solitary confinement. Because 95% or more of all incarcerated people, including those who experienced solitary confinement, are eventually released, understanding disproportionate psychopathology matters for developing prevention policies and addressing the unique needs of people who have experienced solitary confinement, an extreme element of mass incarceration.

Long-term solitary confinement expanded across the United States in the 1980s; by 1997, nearly every state had built a “supermax,” creating an estimated total of 20 000 new solitary cells.^1,2 Human rights agencies characterize the practice as torture^3,4; policy analysts criticize it as expensive and ineffective.^2,4 Yet the epidemiological basis for understanding solitary confinement is weak. Current estimates of the annual US solitary confinement population vary from 80 000 to 250 000.^5,6 Likewise, the conditions (how much isolation with how few privileges), purposes (discipline, protection, or institutional security), and labels (administrative segregation, supermax, restrictive housing, intensive management) defining solitary confinement are contested.^2,5,6 Many studies document psychological harms of segregation, including associations between solitary confinement and self-harm, anxiety, depression, paranoia, and aggression, among other symptoms,^7–9 but other recent findings suggest that psychological impacts are limited.^10–12 Correctional officials use solitary confinement at their discretion, often with few procedural protections, limited available alternative responses, and no external oversight.² Researchers and policymakers are therefore limited not only in access to data and populations, but also by these populations’ fluidity.

A standard instrument for assessing psychological impacts of incarceration is the Brief Psychiatric Rating Scale (BPRS). Originally developed to rate the severity of symptoms in hospitalized psychiatric patients and track changes in status over time,^13,14 the BPRS is increasingly used for research within carceral settings.^12,15,16,17 The current scale assesses 24 observable or self-reported symptoms. Extensive research on the BPRS’s reliability and validity confirms its efficacy in identifying indicators of serious mental illness.¹⁴

In Washington State, interviewers administered the BPRS to a random sample of 87 incarcerated people during qualitative interviews (and also conducted 122 medical chart reviews),^1,9,15 concluding that solitary confinement reveals “a concentration of some of the most important negative effects of the entire prison complex.”^1(p1692) In a widely cited subsequent study, in Colorado, the BPRS was included in a battery of tests designed to measure psychological “constructs” associated with solitary confinement (for 270 matched participants), but generated few reliable results. The study relied on a pencil-and-paper test, the Brief Symptom Inventory, “a 53-item self-report measure . . . to assess a broad range of psychological symptoms,” and concluded that people in solitary confinement sometimes experienced improvements in their psychological well-being, and those with mental illnesses did not deteriorate over time.^11(p52)

Our study builds on these investigations, relying not only on psychometric instruments but also on mental and physical health and disciplinary records and in-depth interview data to assess the psychological well-being of 106 randomly sampled incarcerated people in long-term solitary confinement in the Washington State Department of Corrections (WADOC) from 2017 to 2018. Triangulation of sources gives this study a robust basis for understanding the psychological effects of solitary confinement.

METHODS

WADOC is a midsized (39th highest rate of incarceration in the United States), fully state-funded correctional system with a long history of inviting academic researchers to independently evaluate carceral practice.^1,9,18,19 Fieldwork was conducted over 2 separate 3-week periods in the summers of 2017 and 2018, by a total of 13 research team members (9 women and 4 men) all affiliated with the University of California, Irvine. In total, 106 incarcerated people were interviewed in 2017, and 80 incarcerated people were reinterviewed in 2018. We also collected medical and disciplinary data, including serious mental illness (SMI) and self-harm data.

Sample and Data Collections

WADOC has 5 geographically dispersed intensive management units (IMUs); people in these all-male units have usually violated an in-prison rule and are in solitary confinement for durations ranging from months to years, with highly restricted access to phones, radios, televisions, time out of cell, and visitors. As a result of WADOC efforts to reform and reduce IMU use, the population in these units fluctuated, with a high of more than 600 (in 2011) to a low of 286 incarcerated people (in 2015) on “maximum custody” status: for indeterminate terms, contingent on meeting specific benchmarks.²⁰ In 2017, when the initial sample for this research was drawn, there were 363 maximum custody status people assigned to the IMU.

We selected participants from a randomly ordered list in proportion to the population of each IMU, accounting for 29% of the total population in each of the 5 units. For recruitment and consent processes, see Appendix A (available as a supplement to the online version of this article at http://www.ajph.org). The interview refusal rate was 39% (67 out of 173 approached), comparable to similar studies of incarcerated people.^9,21

The 96-question semistructured interview instrument included a range of questions used in previous studies on incarcerated people’s experiences,^22,23 covering conditions of daily life, physical and mental health treatment, and IMU programming. BPRS self-report items were embedded throughout the interview; we evaluated observational items immediately following each interview.²⁴ Interviews lasted between 45 minutes and 3 hours.

Following interviews, participants were given an option to consent to medical file reviews and to participate in 1-year follow-up interviews. All participants consented to reinterviews, and all but 2 participants (n = 104) consented to medical file reviews. Following year-1 interviews, WADOC provided electronic administrative health and disciplinary files for all 104 consenting participants (along with comparable, population-level data for the prison system in 2017).

In summer 2018, the research team returned to Washington and reconsented and reinterviewed every available participant—notably including those no longer housed in the IMU—for a total of 80 reinterviews. Because of refusals (n = 4), institutional transfers and parole (n = 21), and 1 death, we were unable to follow-up with 26 respondents (25%). This drop-out rate is low compared with similar studies.^25,26 Follow-up interviews lasted between 45 minutes and 2 hours. The condensed year-2 instrument contained approximately 70 questions, with variation by current housing status.

For the steps taken to protect vulnerable imprisoned research participants and details of the training research team members completed, establishing high interrater reliability in administering the BPRS,²⁴ see Appendix A (available as a supplement to the online version of this article at http://www.ajph.org).

Data Analysis

All interviews were assigned a randomly generated identifier, digitally recorded, transcribed in Microsoft Word (Microsoft Corporation, Redmond, WA), translated (1 interview was conducted in Spanish), systematically stripped of identifying details (names, dates of birth), and entered into Atlas-ti (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany) for analysis. See Appendix A for an explanation of the thematically grounded, open-coding process.²⁷ We entered all BPRS paper rating sheets, completed following year-1 and year-2 interviews, into Microsoft Excel (Microsoft Corporation, Redmond, WA). We linked each participant’s BPRS rating, by random identifier, to extracted data from qualitative interviews, medical file reviews, and administrative data from WADOC.

Relevant variables extracted from administrative health data included SMI, a critical classification because it implies that treatment is medically necessary and, therefore, is an obligation of the prison system while the person is under its care. WADOC operationally defines SMI by standardized criteria combining diagnosis, medication, and frequency of psychiatric encounters, and history of suicide attempts or other self-harm.

We then imported BPRS and other administrative data into SPSS version 26 (IBM, Armonk, NY) to generate descriptive statistics, including prevalence of clinically significant ratings on BPRS items and factors (subscales of co-occurring symptom groups), including positive symptoms (unusual thought content, hallucinations, conceptual disorganization), negative symptoms (blunted affect, emotional withdrawal, motor retardation), depression-anxiety-guilt symptoms (including somatic concerns; DAGS), and mania (excitability, elevated mood, hyperactivity, distractibility).¹⁴ We ran correlational analyses (cross-tabs and t test) to evaluate the relationships between BPRS ratings and other independent assessments of well-being, such as existing diagnosis of SMI.

RESULTS

See Table 1 for summary characteristics of the all-male participant population (there are no women in IMUs in WADOC) and the general WADOC population. As in other studies of solitarily confined incarcerated people,⁶ our sample was generally younger, more violent (in terms of criminal history), and serving longer sentences than those in the general population. Latinos and gang affiliates are both overrepresented in our IMU sample, likely because of the salience of conflicts among rival Latino factions as an institutional security concern.² Although our IMU participants differed from the general prison population, there were no significant differences in either demographic variables or criminal history characteristics between our random sample and the overall IMU population, except that our participant pool was slightly older than the overall IMU population.

TABLE 1—

Characteristics of Sample of People in Solitary Confinement Compared With General Prison Population: Washington State Department of Corrections, 2017

	IMU Population (n = 106)	General Population (n = 16 465)a
Age, y
Mean	35	40
Median	34	38
Range	20–65	18–94
Race/ethnicity, % (no.)
White	42 (44)	59 (9746)
African American	12 (12)	18 (2935)
Latino	23 (24)	14 (2276)
Other	23 (24)	9 (1508)
IMU length of stay
Mean	14.5 mo	. . .
Median	6 mo	. . .
Range	< 1 wk–151 mo	. . .
Current offense category, % (no.)
Murder and manslaughter	17 (18)	16 (2623)
Sex offenses	12 (13)	19 (3195)
Robbery and assault	57 (60)	34 (5608)
Property offenses	8 (9)	18 (2933)
Drugs or other	6 (6)	13 (2106)
Prison convictionsb
Mean	5	4
Median	4	3
Range	1–18	1–27
Prison length of stay, mo
Mean	103	97
Median	72	45
Range	3–456	2–600
Ever in prison gang,c % (no.)
Yes	60 (64)	32 (5410)
No	36 (38)	68 (11 659)
Missing	4 (4)	. . .
Serious mental illness,d % (no.)	19 (16)	9 (1589)
Self-harm attempt,e % (no.)	18 (17)	Not available
Suicide attempt,e % (no.)	22 (22)	Not available

Note. IMU = intensive management unit.

^aGeneral population data excludes 761 nonsentenced and 718 resentenced incarcerated people. Both categories returned to prison for technical violations of conditions on underlying drug or sex offenses, a politically selective and narrow set of offenses that would distort the general population primary offense profile.

^bNumber of convictions to prison, excluding out-of-state convictions, often significant for IMU residents.

^cGang status was self-reported. Figure is calculated from 102 respondents who disclosed this information.

^dSerious mental illness data were provided for 85 respondents; figure is calculated from this sample.

^eSelf-harm and suicide data were provided for 94 respondents; figure is calculated from this sample.

Range and Prevalence of Psychological Symptoms Identified

Our initial sample of 106 participants had a mean BPRS rating of 37 and a median rating of 33 (possible range from 24 to 168), suggesting mild psychiatric symptoms among the study population at the time of our interviews.¹⁴ However, analysis of individual scale items showed clinically significant ratings (of 4 or higher of a possible 7) for as much as one quarter of the population sampled, especially for the depression and anxiety symptoms (Table 2). Further analysis of BPRS factors, as opposed to individual items, provided additional evidence of clinically significant psychiatric distress in as much as half of the population sampled (i.e., DAGS factor; Table 2).

TABLE 2—

Brief Psychiatric Rating Scale Symptom and Factor Prevalence: Washington State Department of Corrections, 2017–2018

	IMU 2017 (n = 106), % (No.)	IMU 2018 (n = 28), % (No.)	Non-IMU 2018 (n = 52), % (No.)
Symptomsa
Depression	24.50 (26)	25.00 (7)	15.38 (8)
Anxiety	24.50 (26)	32.14 (9)	28.85 (15)
Somatic concern	15.10 (16)	21.43 (6)	7.69 (4)
Guilt	17.90 (19)	17.86 (5)	7.69 (4)
Hostility	11.30 (12)	17.86 (5)	17.31 (9)
Hallucinations	9.40 (10)	14.29 (4)	11.54 (6)
Excitement	10.40 (11)	14.29 (4)	7.69 (4)
Factorsb
Positive	16.00 (17)	17.86 (5)	11.54 (6)
Negative	4.70 (5)	0 (0)	1.92 (1)
DAGS	49.10 (52)	53.57 (15)	36.54 (19)
Mania	17.00 (18)	14.81 (4)	17.31 (9)

Note. DAGS = depression, anxiety, guilt, and somatization; IMU = intensive management unit; mania = elevated mood, distractibility, motor hyperactivity, and excitement; negative = blunted affect, emotional withdrawal, and motor retardation; positive = hallucinations, unusual thought content, and conceptual disorganization.

^aOnly clinically significant symptoms (rating of 4 or higher) that were reported by 10% or more of the sample are presented.

^bFactors combine 3 or 4 different symptoms that are commonly associated with one another.¹⁴

Administrative data support the finding of long-term psychological distress. Among our respondents, 19% had SMI designations, 22% had a documented suicide attempt, and 18% had documentation of other self-harm, all at some point during their incarceration, either before or during their time in the IMU (Table 1). Moreover, respondents with SMI designations were much more likely to report positive symptoms and slightly more likely to report all other factored symptoms than non-SMI respondents (Table 3). These findings support the validity of the BPRS assessments.

TABLE 3—

Serious Mental Illness Status and 2017 Brief Psychiatric Rating Scale Factor Prevalence: Washington State Department of Corrections, 2017–2018

	SMI (n = 16), % (No.)	Non-SMI (n = 69), % (No.)
Positive	50 (8)	10.14 (7)
Negative	6.30 (1)	4.40 (3)
DAGS	56.30 (9)	47.80 (33)
Mania	18.75 (3)	13 (9)
Populationa	18.80 (16)	81.20 (69)

Note. DAGS = depression, anxiety, guilt, and somatization; mania = elevated mood, distractibility, motor hyperactivity, and excitement; negative = blunted affect, emotional withdrawal, and motor retardation; positive = hallucinations, unusual thought content, and conceptual disorganization; SMI = serious mental illness.

^aMental health data were available only for 85 of 106 sampled incarcerated people.

Qualitative interview data revealed symptoms not otherwise captured by the BPRS and medical files. (Such data will be used illustratively here, for reasons of space, and will be considered exhaustively in subsequent analyses). Two classes of symptoms were reported by a majority of respondents: descriptions of the severity of the emotional toll of being in the IMU (80% of respondents; cumulatively, the topic was mentioned 359 times) and feelings of social isolation (73% of respondents; cumulatively, the topic was mentioned 192 times). This interview excerpt exemplifies the “emotional toll” descriptions:

I bet you couldn’t walk in my shoes because all the stuff you got to endure behind these walls of pain. There’s a lot you got to go through . . . [and] I’ve been doing this for 11 years . . . people adapt to their surroundings, but to get used to this life, I don’t [think] you can. (Michael, a pseudonym, as with all subsequent quotations)

And this quotation exemplifies social isolation:

You’re not around people. I’m around somebody right now with handcuffs and shackles on like I’m an animal. It’s dehumanizing. No human contact. As [a] human being, I feel like we’re meant to socialize, and it does have an effect on your mentality while you’re sitting in the cell. (Chase)

Two additional symptoms were as prevalent as other clinically significant BPRS items like anxiety: references to sensory hypersensitivity (16% of respondents mentioned this at least once) and loss of identity (25% of respondents mentioned this at least once). Respondents discussed hypersensitivity to sounds, smells, “[and . . .] tiny things” (Giovanni). In particular, the sounds of doors opening and closing aggravated many respondents:

All you got to do is hold it. I mean, you don’t got to slam it. It’s like [correctional officers] showing their power. . . . That ain’t cool. You wouldn’t do that in your house, would you? (Tyler).

Respondents also talked about the institution taking over their identity:

I’ve been in the hole so long that it defines the person. If you’ve been in the box for so long, you can’t play well with others. . . . We’re so confined in that box. It’s like a safety blanket. (Eli).

Another respondent echoed a frequent complaint about the lack of mirrors contributing to the loss of identity:

This IMU has mirrors in the cell. The majority of them do not. And it gets really stressful when you can’t even see your own reflection. . . . I mean when you can’t even look at yourself, you lose some of your self-identity. (Eric)

Comparing Symptoms in and out of Solitary Confinement (2018)

Of the 80 respondents reinterviewed in the second year of this study, 28 were in IMU custody and 52 were in the general prison population. These 2 subpopulations provide important comparison groups between IMU residents and people in the general population, because all initially entered the study through a random sample of IMU residents. These subpopulations also provide a longitudinal view of how incarcerated people experience IMU conditions over 1 year and how they recover from these conditions as they re-enter the general population. In Table 2, we compare, cumulatively by subpopulation, symptom and factor scores in 2017 for IMU residents to 2018 scores for IMU respondents and respondents not in the IMU. For respondents still in the IMU in 2018, all clinically significant symptoms that were prevalent among at least 10% of the population were at least as prevalent in 2018, and 2 clinically significant factor scores were more prevalent (positive, DAGS). For respondents not in the IMU in 2018, the prevalence of clinically significant symptoms varied from more prevalent than in the 2017 sample (e.g., anxiety) to less prevalent (e.g., somatic concerns and guilt), and factor scores were either lower (i.e., positive, negative, DAGS) or similar (for mania) for respondents not in the IMU in 2018. Despite having an exceptionally large sample size for a study of a solitary confinement population, our study was not powered to establish statistically significant differences between the 2017 and 2018 data sets.

DISCUSSION

In this study, we combined qualitative interview data with structured, quantitative measures of psychological and psychiatric outcomes in solitary confinement among 106 randomly sampled incarcerated people in Washington State, documenting both a wide range and high prevalence of symptoms of psychological distress. We highlight 4 major implications of this.

First, while the overall BPRS ratings we analyzed indicated limited psychological distress, as documented in earlier studies,^11,12 a closer examination of specific items and factors revealed that as many as half of respondents had at least 1 clinically significant symptom within the BPRS anxiety–depression factor. Because other studies using the BPRS in solitary confinement settings employed earlier 18-item versions of the scale,¹⁵ used the scale in combination with other scales,¹¹ or analyzed only total ratings,¹² our findings are not directly comparable with those in other BPRS studies. However, our findings are consistent with other studies, including findings that 20% or more of Washington incarcerated people in solitary exhibited a “marked or severe degree of distress,”^15(p774) and that more than half of California incarcerated people in solitary reported “symptoms of psychological distress.”^28(p133) Our findings therefore highlight the importance of analyzing specific components of BPRS scores, and not only aggregates, which mask variation in both prevalence and severity of specific symptoms.

Second, administrative data confirmed that our participants had relatively high rates of documented mental health problems, including rates of SMI and self-harming behavior (Table 1). SMI rates, typically estimated at 10% to 15% of prison populations,^8,29 are measured at 9% in Washington’s general prison population but 20% in our IMU sample. Likewise, our qualitative data confirmed that people in solitary confinement experience symptoms specific to those conditions not captured in standard psychiatric assessment instruments.³⁰ Both findings suggest an affirmative answer to the question of whether solitary confinement is associated with more and worse psychopathology than general population confinement. As longitudinal case studies have illustrated,^9,30 disproportionate representation of incarcerated people with psychopathology in solitary confinement reflects the interaction of clinical and security factors in prison custody decisions: solitary confinement responds to behavior expressing psychopathology, often undiagnosed, and also aggravates the propensity of some incarcerated people to break down or act out.³¹ For these reasons, the causal role of solitary confinement is not established by aggregate comparisons of IMU and non-IMU populations.

Third, the comparisons we were able to make across multiple sources of data allowed us to identify a broader range of symptoms of distress than studies that have focused on only 1 or 2 sources of data, such as administrative data,⁸ psychiatric assessments,¹¹ or qualitative interviews.^28,30 Symptoms such as anxiety and depression were especially prevalent in this population, along with symptoms ostensibly specific to solitary confinement, such as sensory hypersensitivity and a perceived loss of identity (as found in other studies exploring solitary-specific symptoms^{7,9,15,28,30,32}).

Finally, consistent with previous studies,^11,12 we found that the prevalence of psychiatric distress did not significantly increase over time for incarcerated people that either stay or are released from the IMU 1 year later. Yet our qualitative data suggest that the BPRS may not be capturing actual psychopathology, as respondents pointed to psychiatric distress—in profoundly existential terms, as in the previously mentioned quotations regarding selfhood and identity—beyond the 2-week time period evaluated by the BPRS and outside the scope of the instrument. Moreover, although symptoms were not cumulatively found to worsen, they did persist at high rates, for incarcerated people in and out of the IMU, in 1-year follow-up assessments. These latter findings are also consistent with other studies, underscoring the need for additional research comparing incarcerated people’s experiences across different contexts and over time.^1,7,15,28,32

Limitations

Five specific limitations are especially notable. First, although our initial sample was relatively large for a solitary confinement population, our 1-year follow-up group, especially the number of respondents remaining in solitary confinement in the second year, was relatively small, limiting our ability to establish statistically significant findings about change over time and across contexts from BPRS data. Second, as our interview results revealed, the BPRS does not capture the full spectrum of psychiatric distress incarcerated people experience in solitary confinement. Third, assessments of psychological well-being would ideally occur at multiple times, beyond the 2 we were able to conduct within the constraints of this multimethod study. Fourth, Washington State is not representative of most state prison systems in terms of the prevalence of people with mental illnesses in solitary confinement, as WADOC has undertaken reforms in both treatment of mental illness and imposition of solitary confinement over the past 20 years, including reforms designed to divert people with serious mental illness to specialized treatment units.³³ Moreover, these reforms have radically improved systematic mental health record-keeping; we would expect not only a lower prevalence of psychiatric symptoms and less deterioration in WADOC in IMUs but also a higher rate of documentation of those symptoms that are present. Finally, although people in solitary confinement may exhibit distinctive or disproportionately severe psychopathology, causal inference regarding the relationship between solitary confinement and psychopathology is beyond the analysis we are able to perform here.

Conclusions and Implications

We found a wide range and high prevalence of symptoms of psychiatric distress in this population, including BPRS symptoms associated with anxiety and depression among as many as half of our participants, administrative indicators of SMI among at least one fifth of our participants, and condition-specific symptoms, such as feelings of extreme social isolation, in well more than half of our participants. Moreover, these symptoms persisted in the second year for participants in and out of solitary confinement.

If we study people in solitary confinement solely with instruments validated with nonincarcerated populations, such as the BPRS, we may fail to capture the extent of incarecerated people’s psychological distress. A respondent’s rating on a given symptom may not be “high enough”; symptoms may not be experienced within the instrument’s designated time frame; or the discursive strategies incarcerated people use to articulate their suffering might not correspond with clinical language. Moreover, past research reveals that incarcerated people develop coping mechanisms for solitary,^1,2,32 and these, along with the fact that speaking openly about psychological distress conflicts with institutional norms of self-protection in prison,^1,2,30 likely contribute to a systematic underreporting of distress. These are critical limitations of standardized assessments of incarcerated people whose symptoms may fluctuate substantially in presence and severity during time in solitary.^1,7,32 Apart from symptoms or their severity, this fluctuation, itself, is an integral aspect of incarcerated people’s psychological distress,³⁴ but a need for repeated measurement makes it especially difficult to capture.

Our findings still point to the importance of using standardized instruments, which provide a baseline for assessing and interpreting the psychological effects of solitary confinement. Nonetheless, additional sources of evidence—interviews, clinician observations, staff observations, medical files—are crucial for capturing the range of symptoms that people in solitary exhibit, and those symptoms’ prevalence, duration, and severity over time. Without the benefit of mixed methods and improved instruments, researchers and policymakers alike will continue not only to lack desired data but also to not know what data we lack. Increasing the transparency of both conditions of confinement and the associated health effects is critical to both question formulation and data gathering.

As 5% to 15% of the United States’ 1.6 million incarcerated people are held in solitary confinement for at least part of their incarceration,^5,6 and virtually all of those people will be released, all members of society have a vested interest in limiting the induction of psychopathology suggested by findings such as those presented here. At least some of the symptoms we described here, including identity loss and hypersensitivity, resulted directly from specific conditions of confinement, such as the absence of mirrors and the repetitive slamming of doors. To the extent that solitary is meant to make people more manageable, its association with psychopathology calls into question its usefulness, let alone its justice. And to the extent that solitary confinement has any causative role in psychopathology, our collective goal should be prevention.