Trajectories of Resilience-Related Traits and Their Impact on Health Outcomes in Schizophrenia: Results From a 4-Year Longitudinal Study

Abstract

Background and Hypothesis

For the rapidly growing population of older people living with schizophrenia (PLWS), psychological resilience, or the capacity to adapt to adversity, is an understudied target for improving health. Little is known about resilience and its longitudinal impact on outcomes among PLWS. This study assesses trajectories of resilience-related traits in PLWS and a nonpsychiatric comparison group (NCs) and longitudinal interactions between resilience and health.

Study Design

This sample included 166 PLWS and 155 NCs (mean age 48 years, 52% women), with a 4.1-year mean follow-up time. The groups were comparable in age, sex, and follow-up time. We assessed resilience-related traits, physical well-being, obesity, hyperglycemia, positive symptoms, and negative symptoms. We conducted linear mixed-effects models to examine linear trends over time and continuous-time structural equation models (CTSEM) to assess the longitudinal relationships (cross-lagged effects between resilience and health).

Study Results

People living with schizophrenia had lower resilience levels, compared with NCs. While resilience was stable over time for White individuals, younger non-White individuals with less education had increases in resilience over time. We found bidirectional 1-year cross-lagged effects of resilience with physical well-being and obesity, but not with hyperglycemia. Among PLWS, there were 1-year cross-lagged effects of resilience with both positive symptoms and negative symptoms.

Conclusions

These findings highlight the importance of resilience and its link to physical and mental health over time. Resilience may be a key protective factor in aging among PLWS, and the potential to improve resilience is an important and understudied approach for improving outcomes for older PLWS.

Introduction

Despite the increased life expectancy for the general US population over the past 70 years,¹ people living with schizophrenia (PLWS) continue to have premature mortality, dying 15 years earlier.² The excess early mortality among PLWS is primarily attributable to cardiovascular and metabolic disease.^3,4 Risks for cardiovascular and metabolic disease are multifactorial among PLWS, including psychotropic medications, smoking, sedentary behaviors, unhealthy lifestyles, and obesity.⁵ Lifestyle and pharmacologic interventions may improve physical health, though such interventions can be challenging to implement due to psychopathology, cost, and access.⁶ Novel alternative approaches are warranted to improve overall health and longevity among PLWS. One understudied approach to improve physical health is enhancing psychological resilience.

Defined as “the process and outcome of successfully adapting to difficult or challenging life experiences,” psychological resilience⁷ has been shown to improve health behaviors and subsequent health outcomes within the general population. Individuals with higher levels of resilience were less likely to smoke, had lower stress levels⁸ and more balanced diets, were more physically active, and slept more,⁹ compared with those with lower levels of resilience. Resilience has been linked to better treatment adherence^10,11 and increased safety behaviors.¹² Higher resilience levels are linked to a lower prevalence of obesity, diabetes, and hypertension.⁹ Older adults with high resilience had fewer hospitalizations and emergency room visits as well as lower prescription drug spending, compared with older adults with low resilience.^13,14 Resilience is also associated with better mental health—including satisfaction with life and happiness, lower levels of psychological distress,¹⁵ as well as less anxiety and depression.¹⁶ In the general population, resilience has been shown to have a positive longitudinal impact on positive coping strategies,¹⁷ stress,¹⁸ and cognition.^19,20 One study reported that high levels of resilience at baseline were associated with a decreased likelihood of cognitive impairment and cognitive decline over a 2.8-year follow-up.²¹

However, little is known about the longitudinal impact of resilience on physical and mental health in PLWS. On average, PLWS have lower levels of resilience compared with nonpsychiatric comparison participants (NCs).²² Among PLWS, resilience is associated with better self-esteem, better quality of life,^23,24 less severe positive and negative symptoms,^23,25 less cognitive disorganization,²⁵ longer duration of illness,²⁴ and greater likelihood of maintaining symptom remission and remaining medication-free after 15 years.²⁶ One study found that PLWS whose resilience increased over 4 years were more likely to achieve recovery after their first episode.²⁷ However, the longitudinal impact of resilience on physical health among PLWS has not been studied.

Resilience-enhancing interventions (eg, social support, mindfulness, cognitive-behavioral therapy) can increase resilience levels in the general population,²⁸ though few studies have explored resilience interventions among PLWS. Multicomponent early intervention strategies (including social support and individual therapy) can improve resilience and cognitive outcomes for individuals at high risk for schizophrenia and with first-episode psychosis.²⁹ While studies in adults with chronic schizophrenia are lacking, resilience may be a key target for intervention that could contribute to improved health outcomes.

Despite the health benefits associated with resilience, little is known about how resilience may be linked to physical health outcomes among individuals with chronic schizophrenia, a highly vulnerable group for physical comorbidities. In this study, we assessed characteristics of resilience as the primary predictor using the Connor-Davidson Resilience Scale (CD-RISC), a commonly used and high-validity measure.³⁰ For the results of this study, resilience is assessed as a disruptive process that results in the acquisition of qualities such as optimism, adaptability to change, etc.³¹ The CD-RISC does not assess the resiliency process or the multilevel systems of resilience.^31,32 We examined the temporal relationships between characteristics of resilience and physical health outcomes among PLWS and NCs. We hypothesized that PLWS would have lower resilience compared with NCs. We also hypothesized that resilience trajectories would be stable in both PLWS and NCs over time. Lastly, we hypothesized that resilience levels and physical health outcomes would be associated with each other, such that lower resilience would lead to worse physical health outcomes in the future and vice versa. We also explored the longitudinal relationships of resilience with positive and negative symptoms among PLWS.

Methods

Participants

Participants in this study included individuals with schizophrenia or schizoaffective disorder (PLWS) and NCs who were recruited from the San Diego Area. Participants with schizophrenia or schizoaffective disorder were recruited through presentations at board and care facilities (assisted living facilities), word-of-mouth, and previous enrollment from other studies. The NCs were recruited using numerous recruitment avenues such as distributing recruitment flyers, ResearchMatch.org, and word-of-mouth. All participants were English speaking. Those with a diagnosis of schizophrenia or schizoaffective disorder and NCs were screened by trained staff with structured clinical interviews. Exclusion criteria included having significant neurological disorders, dementia, intellectual disabilities, recent substance abuse or dependence, or medical conditions impeding consent or study completion. This study received approval from the UC San Diego Human Research Protections Program (IRB# 101631), and all participants provided written informed consent to participate in the study. Participants were evaluated by study staff every 12-18 months for up to 90 months.

Sociodemographic and Clinical Characteristics

Demographic information such as sex, age, race/ethnicity, and education along with schizophrenia-related factors such as age of onset and medication dosages were gathered during individuals’ interviews. Antipsychotic medication dosages were calculated using the World Health Organization guidelines.³³

Resilience Measure

Resilience-related characteristics were assessed with the Connor-Davidson Resilience Scale (CD-RISC), a validated and commonly used 10-item questionnaire.³⁴ Participants were asked to rate themselves on their ability to handle perceived stress, negative emotions, relationships, spirituality, and other stressors over the past month on a 5-point Likert scale (0 being “not true at all” and 4 being “true nearly all the time”). Higher CD-RISC scores indicated greater levels of resilience-related characteristics.

Psychiatric Symptom Severity and General Mental Health

Depression levels were assessed using the Patient Health Questionnaire (PHQ-9),³⁵ a 9-item self-report scale where higher scores indicated greater severity of depression. For PLWS only, the Scale for the Assessment of Positive Symptoms and Negative Symptoms (SAPS and SANS, respectively)³⁶ were used to measure positive and negative symptoms, respectively. Higher SAPS and SANS scores were indicative of worse positive and negative symptoms, respectively.

Physical Health Measures

We administered the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36)³⁷ to assess self-assessed physical well-being, with higher scores correlating with better health. The number and severity of medical comorbidities were evaluated using the Cumulative Illness Rating Scale (scores ranging from 0 to 56, with higher scores indicating a greater number of conditions and increased severity of illness).^38,39 Participants’ body mass index (BMI) was calculated based on their height and weight measurements.

Fasting hemoglobin A1c (HbA1c) levels, a standard indicator of hyperglycemia, were analyzed at the UC San Diego Hospital laboratory using standard assays. Elevated HbA1c levels suggested worse blood sugar control and an increased risk of diabetes and associated complications.

Statistical Analysis

All variables were assessed for normality prior to analyses. HbA1c levels were log-transformed for all analyses to reduce variance and improve efficiency. Sociodemographic characteristics and clinical variables were summarized, and differences were compared between the two diagnostic groups (PLWS and NCs) using independent sample t-tests and chi-square tests.

We employed separate linear mixed-effects models (LMMs) with random intercept and slope to examine linear trends over time (years since baseline visit) in resilience and health outcomes, as well as explore possible differences between groups. These LMMs were constructed while controlling for baseline age, sex, race/ethnicity, and education level. Race was dichotomized into White versus other races, due to the low number of participants from other racial backgrounds. We tested a model that included the main effects mentioned above, interactions between time and the diagnostic group, and interactions between time and all covariates. The LMM results include t-statistics. Statistical significance was set at P <.05 (2-tailed).

Classic longitudinal models, such as latent change score models,⁴⁰ cross-lagged panel models,^41–43 or vector autoregressive models,^44,45 are common dynamic modeling approaches used to explore the dynamics and process of an individual, as well as the variations in these dynamics among individuals. However, these models typically treat time as discrete.^45,46 Thus, their dynamic parameters rely on the specific time intervals, falling short of representing the continuous nature of developmental processes or providing information regarding other temporal durations.⁴⁷

On the other hand, continuous-time structural equation modeling (CTSEM) overcomes this limitation.⁴⁷ In CTSEM, time-related information is explicitly integrated into the model, ensuring that predictions between successive measurements are tied to the precise duration that has passed, rather than simply depending on the number of observations, as is typical in discrete-time models. While there is an exact relationship between the discrete and continuous-time forms in many cases when the time intervals between measurements are equal, the continuous-time form is generally more appropriate when these intervals differ.

In our data, as the time intervals between all measurements vary widely, using CTSEM is preferred for investigating predictive inter-relationships among longitudinal measures with variable assessment intervals (Figure 1). We employed CTSEM to examine cross-lagged effects (eg, resilience → BMI and BMI → resilience) and the potential moderation of these effects by group, while controlling for baseline age, sex, race/ethnicity, and education level. These moderators influence drift effects, the latent process means at the first time point, and continuous-time intercepts to capture the main effects of the moderators. Continuous moderators (baseline age and education level) were standardized. Since CTSEM was modeled using a Bayesian approach, credibility intervals were used instead of confidence intervals.⁴⁸ Multiple comparison is not directly applicable as the Bayesian framework is employed for model estimates. Statistical significance was set at P = .05 (2-tailed).

We also compared the sizes of two cross-lagged effects by testing liberal (assume effects can be different) and restricted (assume effects are the same) models using a chi-squared test. Statistical significance was also set at P < .05 (1-tailed). All analyses were performed in R⁴⁹ and SPSS.⁵⁰

Results

Sample Characteristics

The present study sample included 160 NCs and 173 PLWS, age 40-70 years in the San Diego area (Table 1). The groups were comparable by age, but the PLWS had fewer years of education, more non-White participants, and fewer female participants; compared with the NCs. PLWS had fewer years of education compared with NCs.

Resilience Trajectories

PLWS had lower baseline resilience compared with NCs at all time points (M = 23.0, SD=8.5 and M = 32.5, SD=6.0, respectively; t(330)=11.7, P < .001, d = 1.28) (Figure 2). Resilience levels were relatively stable over time for White individuals, regardless of their diagnostic group, baseline age, gender, or years of education (Supplementary Table S1). Younger non-White individuals with less education had increases in resilience over time. On average, the PLWS group had 4.4 total visits (SD 2.49), while the NC group had 4.65 total visits (SD 2.51).

Relationships Between Resilience and Physical Well-Being

PLWS had significantly poorer physical well-being compared with NCs at all time points (Supplementary Table S2). Older adults had significantly poorer physical well-being compared with younger people at all time points. Men had significantly better physical well-being than women at the beginning of the study. More education was associated with better physical well-being. There is a positive association between education length and physical well-being at the early stages of the study.

One-year cross-lagged effects were significant for resilience and physical well-being in both directions, such that higher resilience levels were associated with better subsequent physical well-being and better physical well-being was associated with higher subsequent resilience levels (Table 2A). While the diagnostic groups had similar cross-lagged effects of higher resilience being associated with better future physical well-being, the PLWS group had a significantly higher 1-year cross-lagged effect of baseline physical well-being being associated with higher resilience, compared with the NC group.

Relationships Between Resilience and Obesity (BMI)

PLWS have significantly higher BMI compared with the NCs at baseline, but the difference in BMI between the diagnostic groups significantly decreases over time (Supplementary Table S3). Males demonstrate a significantly lower BMI than females at the beginning of the study. Education has a negative association with BMI, indicating that a longer duration of education is linked to lower BMI. White individuals significantly exhibit lower BMI compared with non-White individuals.

Both 1-year cross-lagged effects were significant for resilience and obesity (BMI)—with no differences by diagnostic group (Table 2B).

Relationships Between Resilience and Hyperglycemia (HbA1c Levels)

PLWS have significantly higher HbA1c levels compared with the control group at the early time of the study, but the difference becomes insignificant after 2 years (Supplementary Table S4). Older adults have a significantly higher mean HbA1c level compared with the younger people at the beginning of the study. White individuals have a significantly lower mean HbA1c level compared with the non-White individuals.

There were no significant cross-lagged effects between resilience and HbA1c levels (Table 2C).

Relationships Between Resilience and Positive Symptoms in Schizophrenia

Among PLWS, positive symptoms were stable over time (Supplementary Table S5). There were significant 1-year cross-lagged effects between resilience and positive symptoms (Table 3A). Higher resilience levels were associated with better subsequent positive symptoms, and better positive symptoms were associated with higher subsequent resilience levels. The cross-lagged relationship of resilience to subsequent positive symptoms was stronger than the cross-lagged relationship of positive symptoms to subsequent resilience levels (P = .017).

Relationships Between Resilience and Negative Symptoms in Schizophrenia

PLWS with more years of education exhibit significantly less severe negative symptoms compared with those with less education (Supplementary Table S6). There were significant 1-year cross-lagged effects between resilience and negative symptoms (Table 3B). Higher resilience levels were associated with better subsequent negative symptoms, and better negative symptoms were associated with higher subsequent resilience levels. The cross-lagged relationship of resilience to subsequent negative symptoms was stronger than the cross-lagged relationship of negative symptoms to subsequent resilience levels (P = .032).

Discussion

Our study findings partially supported our hypotheses. While PLWS had lower resilience levels at baseline compared with NCs, there were longitudinal increases in resilience levels for non-White participants in both diagnostic groups. There were significant two-way cross-lagged relationships of resilience with better physical well-being and lower BMI, but not with hyperglycemia. Among PLWS, resilience was also associated with less severe positive and negative symptoms. Notably, resilience appeared to have a larger effect on subsequent psychopathology than the reverse direction.

Our findings of differences in resilience levels by the diagnostic group are consistent with prior studies that also found lower levels of resilience among individuals with schizophrenia spectrum disorders⁵¹ and serious mental illnesses,²³ compared with nonpsychiatric comparison participants. Different risk factors related to schizophrenia (eg, cognitive deficits, disability, lack of routine/work, capacity or self-esteem/motivation/self-efficacy, psychotic symptoms) may contribute to lower internal resources and thus, lower resilience.

The current study found increasing levels of resilience among non-White participants, though resilience trajectories were relatively stable among White participants. This finding was consistent with one study that found higher levels of resilience among African-American and Hispanic women, compared with non-Hispanic White women.⁵²

However, other studies have not shown any differences in resilience scores by race or ethnicity.^53,54 The mixed findings may reflect potential influences of age and sample size. Studies have described the skin-deep resilience phenomenon,^55–57 whereby individuals from minoritized and disadvantaged groups who achieve academic and psychological well-being effects are deemed to be resilient in their ability to overcome difficult circumstances. However, the resilience may only be “skin-deep” as these individuals have been shown to have evidence of physical consequences from these challenges including increased inflammation⁵⁷ and increased allostatic load (higher stress hormone levels, elevated blood pressure, higher BMI).⁵⁶ Thus, our current findings may reflect the increased adversity faced by minoritized groups, though the increase in psychological resilience should be interpreted in the context of other consequences of adversity experienced by these populations.

The current study observed bidirectional links of resilience with physical well-being and obesity, that are consistent with cross-sectional studies in older women⁹ and older adults.⁵⁸ These findings may reflect how resilience is associated with healthy lifestyle habits including greater physical activity and lower likelihood of cigarette smoking^8,9 that contribute to better overall health. More directly, individuals with higher resilience are more likely to adhere to medical treatments,^10,11 likely contributing to better management of physical and psychiatric conditions as well as better preventative care. Of note, our findings suggested that better physical health and obesity could contribute to higher resilience in the future. There may be a complex interplay where higher resilience levels improve health and contribute to a positive feedback loop that further increases resilience, analogous to the upward spiral of positive affect.⁵⁹ Similarly, the resilience process may foster the development of key resources (eg, self-esteem,²³ self-efficacy) that further expand an individual’s capacity to adapt to adversity and overcome future obstacles.

The current study found that, among PLWS, lower resilience levels were associated with more severe positive and negative symptoms in the future. These findings are supported by studies that have observed cross-sectional relationships of resilience with positive and negative symptoms.^25,60,61 These findings may reflect the impact of resilience on key mediators of psychopathology including medication adherence,^10,11 better sleep^62,63 possibly due to more consistent routines or lifestyle habits, and the ability to manage stressors.^25,64

The study strengths include the longitudinal design, large sample, and novel approach to examine bi-temporal relationships of resilience with health and psychopathology. The study’s limitations include the potential lack of generalizability of these findings to populations of PLWS who live outside of the San Diego area, who are inpatients or acutely symptomatic, or who do not speak English. Furthermore, the assessment of resilience (using the CD-RISC) is primarily a measure of resilience-related traits, thus we are not able to directly assess the resilience process or response to adversity. The study does not include assessments of lifetime adversity.

The temporal associations of resilience with physical and mental health outcomes hold promise for resilience as a treatment target. Resilience-enhancing interventions have been shown to improve mental health outcomes and coping strategies.^65–67 While published interventions are highly heterogeneous, approaches include cognitive-behavioral therapy, mindfulness training, teaching coping strategies, psychosocial skills or communication training, and problem-solving skills training. Among PLWS, Individual Resiliency Training (IRT) has been shown to improve outcomes for individuals with first-episode psychosis.⁶⁸ IRT includes psychoeducation regarding illness self-management, cognitive-behavioral therapy for psychosis, and teaching rehabilitation skills. Similarly, Individual Coping Awareness Therapy (I-CAT) focuses on mindfulness and developing personalized goal-driven daily routines to foster adaptive responses to stress for adults with early schizophrenia (<5 years of antipsychotic medication treatment).⁶⁹ These results highlight the potential of resilience-enhancing interventions to improve outcomes for PLWS, though adapting such interventions for older adults with chronic schizophrenia is warranted. In light of the physical and cognitive burden among older PLWS, novel interventions are needed to improve aging in schizophrenia.

Author Contributions

Ellen Lee designed the study, oversaw the data analyses, assisted with data interpretation, and wrote the first draft of the manuscript. Tsung-Chin Wu and Xin Tu conducted the data analysis and assisted with data interpretation and manuscript writing. Stephanie Ibrahim assisted with manuscript writing. Angelina VanDyne assisted with data interpretation. Lisa Eyler assisted with data interpretation and manuscript writing.

Funding

This work was supported by the National Institute of Mental Health [an R01 grant (R01MH094151-01) to L.T.E. and a K23 grant (K23 MH119375-01) to E.E.L.]; the National Institutes of Health [NIH UL1TR001442 of CTSA (PI: Gary Firestein, MD)]; a Havens Established Investigator Grant from The Brain & Behavior Research Foundation to E.E.L.; the American Psychiatric Association Kempf Award to E.E.L.; the Desert-Pacific Mental Illness Research Education (for E.E.L. and L.T.E.) and Clinical Center at the VA San Diego Healthcare System; and the Stein Institute for Research on Aging at the University of California San Diego.

Conflicts of Interest

None declared.

Role of Funding Source

The funding sources had no other role in this publication. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

References