One hundred eighty-three participants from three community mental health centers (CMHCs) participated in this study. The CMHCs were located in Nashua, New Hampshire (one site) and Boston, Massachusetts (two sites). In order to be eligible to participate in the study, participants needed to be 50 years of age or older. We used the cutoff of 50 years as a designation for “older adults” in this study. Previous work related to aging adults with serious mental illness has defined the threshold for “older” to begin at age 50 or 55 to account for the early development of aging-related functional impairment in this population (Bartels et al., 2014). The following requirements also had to be met: (a) having a diagnosis of a serious mental illness (i.e., bipolar disorder, major depressive disorder, schizoaffective disorder, or schizophrenia) based on the Structured Clinical Interview for DSM-IV (SCID-CV; First, Spitzer, Gibbon, & Williams, 1996); (b) exhibiting functional impairments; (c) being enrolled in the CMHC research site for a minimum of three months; and (d) being fluent in English. Individuals were deemed ineligible to participate in the study if they met the following criteria: (a) were currently residing in an inpatient institution or nursing home; (b) met criteria for significant cognitive impairment either by having a diagnosis of dementia or based on a Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975) score of less than 20; (c) were diagnosed with a terminal disease with a life expectancy of less than one year; or (d) were currently diagnosed with substance dependence over the past 12 months based on the substance use disorder module of the SCID-CV (First et al., 1996).

Sociodemographic characteristics, including age, gender, race, marital status, education, living situation were ascertained by self-report through the initial screening and baseline questionnaires. Diagnostic information on health conditions were collected through medical documentation.

The Medical Outcomes Study Short-Form Health Survey (SF-36; Ware, Kosinski, & Keller, 1996) was utilized to measure pain-related activity interference (defined as the degree to which pain interferes with daily activities and functioning). The SF-36 is a self-report measure with 36 items, which assesses health on eight subscales covering functional status, well-being, and overall health status. One of the physical subscales from the SF-36 includes two items related to pain. In line with prior survey research among older adults using pain-related activity interference as either a predictor or outcome (Scudds and Østbye, 2001; Thomas, Peat, Harris, Wilkie, & Croft, 2004), we used the item “During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)?”, which is rated on a 5-point scale ranging from 1 (not at all) to 5 (extremely). The physical subscales that include pain items from the SF-36 have demonstrated factorial validity (Ware et al., 1998), relative validity from psychometric tests and clinical tests (McHorney, Ware, & Raczek, 1993), and have been used among adults ≥50 years (Hawker et al., 2011). Previous research has also provided empirical support for the use of the physical subscales from the SF-36 for individuals with serious mental illness, suggesting good internal consistency, test-retest reliability, and concurrent and discriminative validity for this group (Tunis, Croghan, Hellman, Johstone, & Obenchain, 1999).

The Dartmouth-Northern New England Primary Care Cooperative Information Project Functional Assessment Charts (COOP; Nelson et al., 1987) were used to measure several variables related to pain and functional impairment. The self-report charts include illustrations that depict a level of functioning or well-being during the past two weeks. For the current study, we used the feelings chart for bothersome emotional problems, daily activities chart for physical and emotional health-related activity restriction, social activities chart for physical and emotional health-related social limitations, and the pain chart for pain severity. The feelings chart asks “How much have you been bothered by emotional problems such as feeling anxious, depressed, irritable or downhearted and blue?” and is rated on a 5-point rating scale ranging from 1 (not at all) to 5 (extremely). The daily activities chart asks “How much difficulty have you had doing your usual activities or tasks, both inside and outside the house because of your physical and emotional health?” and is rated on a 5-point rating scale ranging from 1 (no difficulty at all) to 5 (could not do). The social activities chart asks “Has your physical and emotional health limited your social activities with family, friends, neighbors or groups?” and is rated on a 5-point rating scale ranging from 1 (no pain) to 5 (severe pain). Finally, the pain chart prompts participants with “How much bodily pain have you generally had?” and is rated on a 5-point rating scale ranging from 1 (no pain) to 5 (severe pain). Single-item rating scales are considered psychometrically sound and standard assessments for pain severity (Hawker, Mian, Kendzerska, & French, 2011; Hoffman, Sadosky, Dukes, & Alvir, 2010). The COOP has demonstrated reliability and convergent and discriminant validity for adult and older adult populations, even when compared to multidimensional measures (McHorney et al., 1992; Nelson et al., 1987). However, this is the first known study in which the COOP was utilized among individuals with serious mental illness.

We conducted a secondary analysis of baseline data collected before the start of the Helping Older People Experience Success (HOPES) intervention study. Prior to the intervention study, the research protocol was approved by the Committee for the Protection of Human Subjects at Dartmouth College.

All statistical analyses were performed using the Statistical Package for the Social Science V.24 (SPSS; IBM Corp., 2016). Data for all predictor and outcome variables were screened using SPSS V.24 for accuracy, multivariate outliers, and normality. The presence of multicollinearity was assessed by examining the variance inflation factors (VIF) and tolerance. None of the VIF values exceeded 10 for any of the predictor variables in the analyses (range = 1.22 to 1.81), and none of the tolerance values were less than .10 (range = .55 to .86), suggesting that there was not multicollinearity in the data and that no large changes in the coefficients would result from adding or deleting variables from the dataset. The Pearson product-moment correlation coefficients among the outcome variable and the predictor variables ranged from small to moderate in size.

Using the COOP pain chart, pain was bifurcated into moderate-to-severe pain (1 = score of 4 or 5, which relates to Moderate pain to Severe pain) and no-to-mild pain (0 = score of 1, 2, or 3, which relates to No pain to Mild pain) based on prior research showing that moderate to higher levels of pain severity are considered to be clinically significant pain due to the consistent links with functional impairment or activity interference caused by pain (Woo et al., 2015). Potential covariates were also selected based on previous research (e.g., Sánchez, Rosenthal, Chan, Brooks, & Bezyak, 2016) and analyzed using exploratory logistic regressions on the pain severity outcome. Sociodemographic characteristics, health conditions, and functional impairment covariates producing ≥10% change in the point estimates in exploratory logistic regressions were used in the logistic regression models in Tables 2 and 3.

To examine the associations with moderate-to-severe pain (referent=no-to-mild pain) as the binary outcome variable, we conducted the logistic regression analyses for the selected sociodemographic characteristics, health conditions, and functional impairment measures among older adults with serious mental illness. We combined serious mental illness diagnosis groups for data analyses based on past studies on pain, health promotion, and mental health among older adults with serious mental illness and functional impairment from community-based mental health programs (e.g., Bartels et al., 2014; Brooks et al., 2018). Sociodemographic characteristics and health conditions included age (divided by 10 to improve interpretation of the findings), gender (1=Female, 0=Male), marital status (1=Currently or previously married, 0=Never married), housing situation (1=Supervised/supported housing, 0=Independent living), serious mental illness diagnosis (1=Bipolar disorder or Major depressive disorder, 0=Schizophrenia or Schizoaffective disorder), and arthritis diagnosis (1=Yes, 0=No). To produce comparable coefficient estimates for the functional impairment variables, raw scores were transformed into z-scores (M=0, SD=1) in order for each logistic regression coefficient to represent change in odds of the outcome for each standard deviation increase in functional impairment variable scores.

To examine whether the logistic regression analyses varied on the basis of separating predictor variables into ordered sets of predictors, we sequentially entered four blocks of variables: (1) age and gender, (2) sociodemographic variables (marital status, housing situation), (3) health conditions (serious mental illness diagnosis, arthritis diagnosis), and (4) functional impairment variables (pain-related activity interference, physical and emotional health-related activity restriction, bothersome emotional problems, and physical and emotional health-related social limitations).

The number of older adults with serious mental illness will disproportionately increase among the baby boomer population in the United States (Bartels et al., 2014). These results suggest that mental health clinicians may, in turn, observe an increasing trend in clients with schizophrenia and other serious mental illness diagnoses reporting pain and should be prepared to assist with identifying and managing pain-related issues. It may also be equally important to closely monitor older adults with serious mental illness experiencing pain for the exacerbation of perceived stress and psychiatric symptoms. It is known that persons with chronic pain are more likely to experience psychological distress, anxiety, depression, and other mild-to-moderate mental health disorders (Keefe et al., 2002). For a population with an inherent susceptibility to psychological problems, it is possible that older individuals with serious mental illness may be at greater risk for the psychiatric repercussions of physical pain, which may further limit functioning.

This study’s findings also indicate there might be causal processes for the associations between moderate-to-severe pain and functional impairment, sociodemographic characteristics, and health conditions. Given the results from the post hoc analysis, potential moderating and mediating factors between pain and sociodemographic characteristics (e.g., age) or health conditions (e.g., arthritis) might include pain-related activity interference and physical and emotional health-related social limitations. These results suggest the need to target health services for older adults with serious mental illness toward providing assistance with daily and social activities affected by pain and associated health problems within community-based mental health programs. It is well-documented that individuals with serious mental illness are more likely to experience social problems due to illness-related, personal, and socioenvironmental factors (Tschopp & Frain, 2009), therefore, social isolation and loneliness might be key factors related to pain and disability (Jaremka et al., 2014). The findings of this study might, then, also suggest that clinicians should evaluate pain as an independent factor affecting diminished involvement in activities and social functioning among middle-aged and older adults with serious mental illness who already experience substantial physical and emotional health-related social limitations. Overall, psychiatric rehabilitation practitioners might consider using evaluations and methods addressing pain and related functional impairment for older adults with serious mental illness through brief screening tools, comprehensive assessments, non-pharmacological pain management, and health promotion interventions.

This is one of the few studies to describe self-reported, non-experimentally-induced, moderate-to-severe pain and related health and contextual factors among older adults with serious mental illness. However, several limitations should be taken into account when interpreting our findings. First, this study was conducted using secondary data, which restricted the sample to older adults meeting the eligibility criteria set in the original study. For instance, persons with significant cognitive limitations and substance dependence were excluded from participation. However, older adults with pain may be more likely to have cognitive impairment or substance use disorders such as prescription opioid misuse (Gfroerer et al., 2003; Weiner et al., 2006), which may have impacted the current study’s ability to assess pain and related factors in other subpopulations. Using secondary data also limited our dataset to the variables selected for the original study. For example, we assessed pain-related activity interference and pain using the SF-36 and COOP charts, and although 1-item rating scales are routinely used in pain survey research (Herr & Garand, 2001), the SF-36 item and COOP chart have limited response ranges. Moreover, pain-related activity interference was specific to work performed inside and outside of the home; therefore, generalizability to non-work-related activities such as social and community engagements or recreation and leisure activities may be limited. In addition, neither pain chronicity nor pain duration were assessed. Next, causality could not be determined due to the cross-sectional study design. It is possible that associations among the variables are bidirectional or that diminished functioning might contribute to greater perceived pain. We also understand that our subsample of individuals with moderate-to-severe pain is small and that we may have restricted statistical power and may have limited our ability to identify variables associated with pain compared to those with no-to-mild pain. Third, we recruited participants from CMHCs; thus limiting the generalizability of our results only to older adults with serious mental illness who participate in CMHCs. Finally, although we attempted to exclude participants from our study who had cognitive disorders or substance dependence, we utilized self-report data, which are susceptible to affective bias, poor insight, and recent life events (Atkinson, Zibin, & Chuang, 1997).

Despite these limitations, the study findings are largely consistent with biopsychosocial models of pain and related disability (Gatchel et al., 2007; World Health Organization, 2001). Subsequent work would benefit from exploring other biomedical/physical factors (e.g., pain duration, prescription opioid use), psychological factors (e.g., anxiety, anger, trauma, personality traits, or psychosis), or social factors (e.g., specific forms of social support, including family, friend, and significant other support) relevant to pain. Future research should investigate how these factors may interact and influence pain in older adults with serious mental illness.