Introduction

COVID-19 is an infectious disease caused by the coronavirus, which first affected Wuhan, China in December 2019. By March 2020, it had spread across continents and was recognized as a global pandemic. In addition to the negative effects on global health and the global economy, scientists have predicted that the pandemic will have extensive effects on long-term mental health worldwide [1,2,3,4,5]. Research on frontline healthcare workers and individuals recovered from COVID-19 supports this prediction, with evidence showing increased rates of PTSD, depression, and anxiety [6,7,8,9,10,11,12].

Researchers across the world have begun to examine the effects of COVID-19 on those with pre-existing mental illnesses, but research in this area is still nascent. Multiple experts have posited that the COVID-19 pandemic will lead to widespread reductions in mental health treatment utilization and increases in acute symptoms will follow in those who had been diagnosed with a psychotic disorder prior to the COVID-19 pandemic [13,14,15,16]. Furthermore, there is also a concern in the mental health community that pandemic-related stressors, such as social isolation, restricted healthcare access, and lower physical activity, may contribute to psychosis conversion in those at clinical high-risk (CHR) for develo** psychosis (i.e., those with prodromal syndromes) [17,18,19,20].

Initial reports of the effects of the COVID-19 pandemic on symptoms in individuals with schizophrenia-spectrum disorders have been inconsistent. A recent study by Esposito and colleagues [21] assessed first-episode psychosis patients hospitalized during the beginning of the shelter-in-place in Italy and compared these findings to 2019 data in the same timeframe. They found that the average age of individuals hospitalized for first-episode psychosis in Italy was significantly older during the pandemic (mean age: 43.5 years), which may reflect an increase in stress-induced psychosis due to the negative impact of the COVID-19 pandemic. Wynn and colleagues [22] analyzed clinical outcomes in veterans with psychosis and veterans who had recently experienced homelessness. They found that veterans with psychosis and veterans who had recently experienced homelessness had increased levels of depression, anxiety, and loneliness compared to retrospective self-report of pre-pandemic levels. However, this study did not examine the frequency of hallucinations or delusions. Research conducted during the pandemic does indicate higher rates of attenuated psychosis than expected in the general population [1, 23, 24]. Attenuated symptoms have also been reported to be maximal early in the pandemic and to have decreased over time [25,26,27,28]. However, Pinkham and colleagues [29] found no significant changes in positive symptoms in patients diagnosed with schizophrenia-spectrum disorders or bipolar disorder. They found that patients reported higher levels of well-being during the pandemic. Additionally, Strauss and colleagues [30] found that individuals with schizophrenia reported significantly higher negative symptoms during the pandemic compared to pre-pandemic levels across all five domains (alogia, blunted affect, anhedonia, avolition, and asociality), and youth at clinical-high risk had reported increases in anhedonia and avolition during the pandemic compared to pre-pandemic. Determining the pandemic’s effects—or lack thereof—on symptoms is crucial to informing ongoing treatment practices used to manage and prevent serious mental illness during the COVID-19 pandemic.

To mitigate infection rates and deaths, governments throughout the globe began to regulate social interactions through shelter-in-place orders, instituting quarantines and self-isolations, along with social distancing policies [31,32,33,34,35,36]. Social isolation is associated with worse mental health outcomes for individuals with psychosis, including increased risk of suicide, and social isolation has been found to be a mediator in the relationships between suicidal ideation and hallucinations, delusions, and depression [37]. Research during the pandemic has shown that individuals on the schizophrenia spectrum have been adherent to COVID-19 health and safety recommendations, which indicates that they are following social distancing protocols and thus may be at risk for such mental health sequelae [38]. Due to these regulations, measures have been implemented across the mental healthcare system to prevent the spread of COVID-19, and healthcare providers have had to transition to telehealth platforms to deliver care [2, 39, 40]. However, it is unclear whether the shift to telehealth services is as effective for individuals with schizophrenia-spectrum disorders, nor is it clear whether patients will adhere to treatment via online mental health services. Treatment adherence with antipsychotic medications is already a well-known issue in schizophrenia patients, along with patients who have chronic illnesses in general [41,42,43,44]. Considering medications in the treatment of schizophrenia are associated with improved clinical outcomes and quality of life [45], medication adherence is a critical issue during the pandemic. Furthermore, self-efficacy has been found to be a protective factor that is associated with the likeliness to cope with day-to-day stressors in individuals with early psychosis [46]. This suggests that variables such as self-efficacy and treatment adherence are important to examine as protective factors during the pandemic.

The current longitudinal study provides a report of findings from the University of Georgia PACE Study (Psychosis Assessment of COVID-19 Effects), which was designed to evaluate COVID-19-related changes in symptom severity and their moderators in those with chronic schizophrenia (SZ) and individuals at clinical high-risk for psychosis (CHR). This manuscript focuses on the effects of the COVID-19 pandemic on positive symptoms specifically (i.e., presence of hallucinations and delusions). The following hypotheses were evaluated: (1) Hallucinations and delusions would occur more frequently and evoke greater distress in clinical groups (SZ, CHR) compared to CN at all time points; (2) The frequency of and distress resulting from positive symptoms would increase during the pandemic for both clinical groups; and (3) medication adherence, access to telehealth, and protective factors (i.e., self-efficacy) would be associated with lower frequency of positive symptoms and distress resulting from positive symptoms in the clinical groups.

Material and methods

Participants

Data were collected from two samples and results related to negative symptoms and COVID-19 pandemic health/safety precautions have been reported elsewhere [30, 38]. Sample 1 consisted of outpatients with SZ and matched community CN, whereas sample 2 consisted of CHR and matched community CN.

Sample 1

Participants included 32 outpatients meeting DSM-5 criteria for schizophrenia, schizoaffective disorder, or bipolar disorder with psychotic features (SZ) and 31 healthy controls (CN). Participants with SZ were originally recruited for studies investigating mechanisms of negative symptoms that occurred prior to the COVID-19 pandemic and had heterogeneous symptom presentations [47,48,49,50,51]. Original recruitment occurred at outpatient mental health clinics in northeast Georgia, USA and through online or printed advertisements. Patients were evaluated during periods of clinical stability as indicated by no self-reported change in medication type of dose within the past 4 weeks. Diagnosis was established via the Structured Clinical Interview for the DSM-5 (SCID-5) [52]. SZ were generally in the chronic phase of illness, had experienced multiple episodes, and were experiencing mild to moderate symptoms.

Healthy control participants (CN) were recruited through printed and online advertisements. CN completed a diagnostic interview, including the SCID-5 [52] and SCID-5 for personality disorders [53], and did not meet criteria for any current psychiatric disorder or schizophrenia-spectrum personality disorder. CN also had no family history of psychosis and did not meet the lifetime criteria for psychotic disorders.

No participants met criteria for substance use disorders (other than tobacco) and all denied a lifetime history of neurological disorders associated with cognitive impairment (e.g., Traumatic Brain Injury, Epilepsy).

Individuals with SZ and CN did not significantly differ in age, parental education, sex, or ethnicity; however, SZ had lower personal education than CN (see Table 1).

Table 1 Demographic characteristics for study 1 and study 2
Full size table

Sample 2

Participants included 25 CHR participants and 30 healthy controls (CN) who were originally recruited for studies examining reward processing mechanisms underlying negative symptoms and psychosis risk [47, 48, 54, 55]. CHR participants were recruited from the Georgia Psychiatric Risk Evaluation Program (G-PREP), which receives referrals from local clinicians to perform diagnostic assessment and monitoring evaluations for youth displaying psychotic experiences. CHR participants were also recruited via online and printed advertisements. CHR participants were included if they met criteria for a prodromal syndrome on the Structured Interview for Prodromal Syndromes (SIPS) [56]. All CHR participants met SIPS criteria for Attenuated Positive Symptoms (i.e., SIPS score of at least 3–5 on at least one positive symptom item, with a frequency of occurring at least once per week; 13 progression, 11 persistence, 1 partial remission). CHR participants did not meet lifetime criteria for a DSM-5 psychotic disorder as determined via the SCID-5 and two participants in the CHR sample had been prescribed an antipsychotic. No CHR participants met the current criteria for a substance use disorder.

CN recruitment and inclusion/exclusion were identical to sample 1. CHR did not significantly differ from their matched CN group on age, sex, race, or parental education; however, CHR had lower personal education than CN.

Procedures

During studies where initial recruitment occurred, SZ, CHR, and CN participants had all consented to be recontacted for future studies. Recontact was done via email, text message, or phone call to determine interest in participating in an online study.

Participation in data collection occurred between July 9, 2020 and October 5, 2020 for Times 1 and 2 (T1 and T2) and August 2, 2021-September 19, 2021 for Time 3 (T3). There were 8 participants (6 SZ, 2 CN) in Study 1 and 11 participants (5 CHR, 6 CN) in Study 2 who did not complete the second wave (T3). For context, the state of Georgia ordered shelter-in-place on April 3, 2020, in response to the COVID-19 pandemic. At the end of Time 2 data collection, the COVID-19 pandemic state of emergency was still in effect. COVID precautions (e.g., restrictions on certain businesses being open, mask wearing, etc.) were widely in place throughout the Time 2 data collection period. During T3, there was still a local emergency declaration in place that included a mask requirement indoors in Athens-Clarke county.

All participants completed an online consent for a protocol approved by the University of Georgia Institutional Review Board and in accordance with the ethical standards outlined in the 1964 Declaration of Helsinki and its later amendments. After consenting, participants were automatically directed to complete a series of questionnaires online by themselves Qualtrics that took approximately 1–2 h. Participants received a $40 check payment for participating.

Online questionnaires covered a range of content: demographics, COVID-19 health and safety behaviors, environmental factors, positive symptoms, general symptoms (e.g., anxiety, depression, mania, sleep), internet/social media use, and protective factors. Only the positive symptom data is the focus of this report. Participants were asked to answer the questions in relation to their general experience during the following time frames: prior to shelter-in-place (T1), during shelter-in-place (T2), and after shelter-in-place with ongoing mask requirement indoors (T3). A total of 3 (SZ = 0, CN = 0; CHR = 2, CN = 1) participants reported having contracted COVID at times 1 and 2 and 19 participants (SZ = 4, CN = 3; CHR = 6, CN = 6) at time 3. The proportion of participants contracting COVID did not differ between clinical and control groups in either study.

Measures

Brief Assessment of Positive Symptoms The Brief Assessment of Positive Symptoms (BAPS) is a 17-item self-report report questionnaire assessing the frequency of and distress resulting from hallucinations and delusions over the past week. The frequency of hallucinations was assessed in relation to the five sensory domains (visual, auditory, gustatory, tactile, and olfactory), with response options ranging from “never” to “everyday for most of the day.” There was also a single item assessing the distress resulting from any of these hallucinations anchored from 0 to 100. The frequency of nine types of delusions was evaluated (persecution, guilt, grandeur, reference, mind reading, thought insertion/withdrawal, nihilism, control, and somatic delusions). There was also a single item assessing the magnitude of distress resulting from any of those delusions (0–100 scale). The BAPS showed good internal consistency across time for frequency (ɑ = 0.88–0.89) and distress (ɑ = 0.78–0.88). Scree plot and exploratory factor analysis supported a one-factor solution for frequency items (see Supplemental Table S8 and Supplemental Table S9). The items are presented in supplemental materials. An overall positive symptom frequency score was calculated as the average of all frequency items; separate hallucination and delusion frequency scores were calculated as the average of only those items.

Data analysis

Hypotheses 1 (Group Effects) and 2 (Time Effects)

Mixed model ANOVAs were conducted for Study 1 (SZ) and Study 2 (CHR) to examine the effects of Group (CN and SZ or CHR), Time (T1, T2, T3), and the Group × Time interaction on: (1) the frequency of hallucinations and delusions, and (2) the intensity of distress from hallucinations and delusions. A main effect of Time would indicate a significant change in the frequency or intensity of distress from positive symptoms during the pandemic. The Group × Time interactions in these two ANOVAs inform hypotheses regarding a change in positive symptoms between groups through the pandemic. Post hoc analyses consisted of Tukey tests that were corrected for multiple comparisons.

Hypothesis 3

All correlations conducted involved difference scores unless otherwise specified. Change was calculated as T2–T1, T3–T2, and T3–T1 (i.e., a positive value indicates an elevation from the previous time). To calculate medication change, two new variables were created: the first variable was created to represent a change in medication status from T1 to T2 from the answer to the question, “Did you take medications?” which was asked at each time point. If participants had a different answer between the two-time points, they were given a score of “1”; if participants had the same answer between the two-time points, they were given a score of “0”. The same process was repeated to find the change from T2 to T3. This question was designed to target whether changes in medication status during the pandemic had an impact on positive symptoms. This process was repeated to create two new variables examining telehealth utilization change from T1 to T2 and T2 to T3 based on dichotomous responses to the question, “Have you received remote healthcare services (i.e., telemedicine, teletherapy)?”.

One-way ANOVAs were conducted to examine whether medication adherence predicted positive symptoms frequency and/or distress and whether access to telehealth was predictive of the frequency of positive symptoms and distress resulting from them. This was completed by analyzing participants’ dichotomous responses to the question, “Did you miss any medications?” at each time point, and comparing those responses to the positive symptoms frequency and distress responses at the corresponding time point (i.e., did they miss medications at T1, and how frequent/distressing were their symptoms at T1).

Three separate difference scores (T1–T2, T2–T3, and T1–T3) were calculated from survey items measuring efficacy as a protective factor. Pearson correlations were conducted between the protective factors summary scores for T1–T2, T2–T3, and T1–T3 with difference scores of positive symptom frequency and distress from T1–T2, T2–T3, and T1–T3.

Results

Preliminary analyses

The BAPS showed good internal consistency across time for frequency (ɑ = 0.88–0.89) and distress (ɑ = 0.78–0.88). Scree plot and exploratory factor analysis supported a one-factor solution for frequency items (see Supplemental Table S8 and Supplemental Table S9).

Hypothesis 1. Elevated positive symptoms in clinical groups relative to CN

Repeated measures ANOVAs are presented in Table 2. Across all dependent variables, there was a significant main effect of Group such that SZ and CHR endorsed greater positive symptoms than their respective CN groups across all time points. Additionally, an exploratory analysis was conducted on whether SZ and CHR differed in positive symptom frequency and distress (see Supplemental Table S10 for ANOVA results), and no group differences were found.

Table 2 Study 1 (Schizophrenia) and Study 2 (Clinical High-Risk) pre-pandemic to during pandemic changes in frequency and distress
Full size table

Hypothesis 2. Changes in positive symptoms across time

Sample 1 (SZ) Repeated measures ANOVAs revealed a significant effect of Time for hallucination frequency (see Table 2). Post-hoc contrasts reveal that hallucinations decreased significantly in frequency from T1 to T3 (d = 0.47) and from T2 to T3 (d = 0.56). Hallucinations increased in frequency from T1–T2, but not to a significant degree (d = − 0.09).

There was a significant Group × Time interaction for hallucination distress such that distress decreased significantly from T2 to T3 in SZ (d = 0.57) but not CN (d = − 0.11). There was no significant increase from T1–T2 or T1–T3 in both groups.

Repeated measures ANOVAs revealed no significant effects of Time for delusion frequency and delusion distress. There were nonsignificant Group × Time interactions for delusion frequency and delusion distress.

Exploratory item-level analyses for Study 1 are presented in Supplemental Table S1. Generally, effects did not differ from those found with overall scores; however, there was a significant effect of Time for olfactory hallucinations (F[1, 126] = 3.53, p = 0.032) in the SZ group. Post-hoc contrasts indicated that the difference in frequency of olfactory hallucinations in both groups was greater at T1 compared to T2 and T3; however, the frequency of olfactory hallucinations was relatively low in this sample and findings may therefore be influenced by the relative infrequency of this phenomenon.

Sample 2 (CHR) The repeated measures ANOVAs revealed nonsignificant effects of Time on hallucination frequency and hallucination distress. There were nonsignificant Group × Time interactions.

Repeated measures ANOVAs revealed a significant effect of Time for delusion distress. Post hoc pairwise comparisons showed a significant decrease in delusion distress from T2–T3 (d = 0.57), although the change between T1–T2 was nonsignificant. There was no significant effect of Time on delusion frequency; nonsignificant Group × Time interactions were observed within Sample 2.

Exploratory item-level analyses (see Supplemental Table S2) revealed that there was a significant Group × Time interaction for visual hallucinations (F[2, 114] = 3.55, p = 0.032) in the CHR group. Post-hoc contrasts indicated that there were significant decreases in visual hallucination frequency in the CHR group between T1–T2 (p = 0.006) and T1–T3 (p = 0.005), while changes in the CN group were nonsignificant. In the CHR group, visual hallucinations were greatest at T1 and declined at T2 and T3. There were no other significant item-level analyses.

Hypothesis 3. Medication adherence, telehealth access, and protective factors

Sample 1 (SZ) One-way ANOVAs (see Supplemental Table S3) revealed there was a significant difference in delusion frequency at T3 between those who reported missing medications at T3 and those who did not miss medications, F(1, 24) = 4.39, p = 0.047; participants who reported missing medications reported increased frequency of delusions than those who did not miss medications. Only one SZ participant reported a change in medication status from T1–T2 and T2–T3, therefore, models could not be run to examine whether changes in medication status had an effect on symptoms. There were nonsignificant differences in positive symptoms between those who endorsed utilizing telehealth and those who did use telehealth (see Supplemental Table S5). In addition, there were no significant differences based on a change in utilization of telehealth services from T1–T2 and T2–T3 (see Supplemental Table S6).

Pearson correlations revealed that there were significant negative associations between efficacy scores from T1 to T2 and delusion frequency, hallucination frequency, delusion distress, and hallucination distress in SZ from T1 to T2. However, none of the correlations survived correction for multiple comparisons (see Supplemental Table S7).

Sample 2 (CHR) One-way ANOVAs (Supplemental Table S3) indicated nonsignificant group differences in positive symptoms between those who missed medications (see Supplemental Table S3) and nonsignificant differences between those who had a change in medication status and those who did not (see Supplemental Table S4). There were nonsignificant differences in positive symptoms between those who endorsed utilizing telehealth and those who did use telehealth at T2 and T3 (see Supplemental Table S5). Only one participant endorsed utilizing telehealth at T1, therefore, models could not be run at that time point. Additionally, there were nonsignificant differences between those who had a change in utilization of telehealth services during the pandemic and those who did not (see Supplemental Table S6).

Pearson correlations revealed that there were significant negative associations between efficacy scores from T1 to T2 and delusion frequency, hallucination frequency, and delusion distress in CHR from T1 to T2, but hallucination distress was nonsignificant. However, none of the correlations survived correction for multiple comparisons (see Supplemental Table S7).

Discussion

The aim of the current study was to examine changes in positive symptoms over the course of the pandemic, the effects of medication adherence and telehealth utilization on positive symptoms, and how protective factors may have mitigated increases in the frequency of or distress resulting from positive symptoms. Overall, in Study 1 (SZ), distress resulting from hallucinations increased at T2 compared to T1 (though not to a significant degree), but then reduced from T2 to T3, while hallucination frequency decreased from T1 to T3 and T2 to T3. In contrast, delusion frequency and delusion distress in SZ did not significantly change. In Study 2, CHR and CN displayed a decrease in delusion distress from T2-T3, but not hallucination distress nor frequency of symptoms across all time points. Additionally, the frequency of visual hallucinations in the CHR group declined significantly during the pandemic.

Importantly, distress resulting from delusions and hallucinations was impacted during the pandemic, but not frequency (with the exception of visual hallucinations in the CHR group). This suggests that while symptoms may be relatively comparable to pre-pandemic times in terms of their rate of occurrence, the impact on patients’ emotional lives was heightened in the early stages of the pandemic and then reduced as the pandemic persisted. This pattern of findings may be consistent with what is observed for general psychiatric symptoms (i.e., depression and anxiety) [21, 22], and suggest that the pandemic-related fluctuations in distress across time points in the pandemic may be due to changes in global negative affect. Alternatively, this may be novel evidence for resiliency in individuals on the schizophrenia spectrum during the pandemic, which supports findings from Pinkham and colleagues [29] who found an increase in well-being during the pandemic in individuals with severe mental illness (i.e., bipolar disorder and psychosis).

The study also examined whether medication adherence, telehealth utilization, and protective factors had an impact on positive symptoms during the pandemic. Muruganadam and colleagues [57] found that one in five patients with severe mental illness reported that they stopped taking their medications as prescribed during the pandemic. The current results indicated a significant difference in the frequency of delusions at T3 in those who endorsed missing medications compared to those who did not miss medications in the SZ group. However, this effect was nonsignificant at T1 and T2, which may be a consequence of growing distrust of the medical system over the course of the pandemic [58]. Telehealth utilization did not have any significant association with symptom frequency or distress. There is limited research on the use of telehealth with individuals on the schizophrenia spectrum (see Santesteban-Echarri and colleagues [59], for a review), and future research should try to identify whether long-term telehealth-based care is as efficacious as in-person treatment in this clinical population. These are critical treatment targets for healthcare providers, and it may be beneficial to take additional measures to ensure that patients are adhering to treatment regimens during the pandemic.

In recent years, the treatment of severe mental illness has shifted to a paradigm of recovery and treatment focusing on well-being and empowerment, in which self-efficacy is a key theme [60, 61]. Self-efficacy was not found to have any significant effects on positive symptoms throughout the pandemic once analyses were corrected for multiple comparisons. This result was contrary to expectations, but may be an indication that the measure did not effectively capture self-efficacy or that other protective factors not assessed in this study (e.g., social support) have a greater effect. Further research on protective factors during the pandemic for individuals with schizophrenia-spectrum disorders is necessary.

The present study had some limitations. First, administering an online study during the pandemic was challenging and limited the sample due to the fact that it required participants to have internet access and a device to complete the survey. Some participants who were invited were unable to complete the study due to internet access issues. This may limit the generalizability of this data due to these barriers to participate. Second, our sample has more females and higher education levels than past study samples conducted in person in our lab. Third, participants were asked to retrospectively rate their positive symptoms at T1 during T2 data collection, and both data points were collected at the same time. This may lend to some caution in interpreting T1 data due to the cognitive burden on the participant to accurately recall their symptoms prior to the pandemic. However, increases in positive symptoms were not associated with precautionary measures such as social distancing [38]. Finally, when asking participants about missing medications, the survey did not specify “antipsychotic”, “psychotropic”, or “psychiatric” medications, therefore, we cannot determine whether participants who responded were responding about missing antipsychotic medications, or if they were missing other non-psychiatric medications. Future research should examine medication effects with greater granularity.

Conclusions

The current findings indicate that although the frequency of positive symptoms was stable over time, distress in response to these symptoms was impacted during the pandemic. Distress from positive symptoms may mirror general distress during the pandemic, which was initially elevated but reduced as people became accustomed to the limitations and stressors imposed by the pandemic [62]. Additionally, it is essential that healthcare providers ensure continuity of care throughout the pandemic to prevent exacerbations in positive symptoms. As the pandemic persists, additional longitudinal research is necessary to further understand how the pandemic has impacted positive symptoms in individuals on the schizophrenia spectrum.