INTRODUCTION

The COVID-19 pandemic has been associated with adverse consequences for non-cancer pain patients. Pandemic-related stress and mood disruption have been linked to worse pain intensity and pain-related impaired functioning 1,2,3,4 which in turn can contribute to increased opioid use and high-risk concomitant benzodiazepine use.5

Cross-sectional survey studies found respondents with chronic pain and post-surgical pain perceived worsening pain severity and worsening pain interference early in the COVID-19 pandemic.1,3 In some cases, this perception was correlated with worsening mood, sleep problems, worries about the future, and feelings of insecurity.2,3 In contrast, Mun and colleagues’6 survey of chronic pain patients in the first year of the pandemic revealed little to no worsening in pain severity, pain interference, emotional distress, and opioid misuse.

The pandemic is associated with reduced access to, and utilization of, non-pharmacological pain therapies7,8 and postponed surgical interventions,9 all of which may contribute to worsening pain and increased opioid use.10 However, there is mixed evidence regarding changes in use of prescription opioids during the first year of COVID-19.8,11,12,13 Recent analyses of nationally distributed medical claims data observed that compared to the same time period prior to the pandemic, patients with pain were more likely to receive prescription opioids for longer duration and greater potency in the first year of the pandemic.14 In addition, patients were more likely to transition from non-pharmacological pain management to opioids during the first 9 months of the COVID-19 pandemic compared to the first 9 months of 2019.14

The long-term effects of the pandemic on prescription opioid use are not clear. We carried out the present research using data from the Prescription Opioids and Depression Pathways Cohort Study.15 This is an on-going, prospective cohort study of new 30–90-day prescription opioid users who complete baseline, 6-month, and 12-month follow-up assessments. Although enrollment is on-going and only 30% of the cohort has had the opportunity to complete the 12-month follow-up, we believe using data from participants who completed baseline and the 6-month follow-up is appropriate. There is an urgent need to communicate the potential for COVID-19-related stress to contribute to long-term opioid therapy (LTOT) as the pandemic waxes and wanes but remains a public health crisis.

Our first objective was to determine if any COVID-19-related stressor (e.g., concerns about health, concerns about social isolation, reporting pandemic worsened pain, etc.) was associated with persistent prescription opioid use over a 6-month period. Second, we determined if this association was independent of demographic factors, change in pain measures, mental well-being, and social support.

METHODS

Patient Recruitment

Participants were enrolled in the Prescription Opioids and Depression Pathways Cohort Study, henceforth termed the Pathways Study, which is designed to determine the mental health consequences of LTOT and risk factors for prescription opioid use disorder. The study protocol has been reported in detail.15 In brief, eligible patients were identified from the electronic health records of Saint Louis University’s academic medical practice in St. Louis, MO, and Henry Ford Health in Detroit, MI. Patients were eligible to enroll in the study if they were starting a new period of prescription opioid use, defined as no opioid use in the prior 3 months, and were free of cancer. Opioid use and cancer status were identified in electronic health records and confirmed in screening questions. Patients who completed the baseline assessment were invited to complete 6- and 12-month follow-up surveys. The current 6-month retention rate is approximately 82%. The present study uses data from the first 477 patients who completed baseline, 6-month follow-up assessments, and answered questions about COVID-19 stressors. Baseline enrollment began in November 2019. Measures of COVID-19-related stress were added to the 6-month follow-up survey in September 2020. Prior to incorporating the COVID-19 questions, 62 participants completed baseline. Therefore, the current study uses data from 477 participants who completed baseline and 6-month follow-up between September 2020 and January 2022.

All assessments were administered in REDCap and either completed by the subject on the internet or by a telephone interviewer entering answers into REDCap. Patients provided informed consent prior to participation. Patients were given a $50 gift card for each survey completed. All procedures were approved by Saint Louis University and Henry Ford Health’s Institutional Review Boards.

Variables

Outcome

Prescription opioid use at baseline and 6-month follow-up was based on self-report of the opioid type, frequency (daily vs. non-daily), and dose. Persistent opioid use was defined as using a prescription opioid at 6-month follow-up.

Primary Exposure

COVID-19-related stressors were derived from the Complementary and Integrative Research (CAIR) Pandemic Impact Questionnaire (C-PIQ).16 The C-PIQ measures experiences in the past 2 weeks and includes the following questions: (1) How much are you reading, watching, listening, talking or thinking about coronavirus/COVID-19? (2) How much do you worry about your health or the health of your friends or family? (3) How stressful have changes in social contacts, like family or friends, been for you? (4) How stressful have changes in your way of life, like changes in finances, education, living situation, childcare, etc. been for you? (5) How much has your mental/emotional health been worsened by the COVID-19 pandemic? We created a sixth question based on C-PIQ format which was as follows: (6) How much has your pain been worsened by the COVID-19 pandemic? For questions 1 and 2, binary (no/yes) variables were computed by combining response options never/rarely (no) vs. occasionally/often/most of the time (yes). Binary variables for questions 3 through 6 were created by combining response options not at all/ slightly (no) vs. moderately/very/extremely (yes).

Covariates

Demographic measures included age, race, gender, and marital status.

Pain Measures

The Brief Pain Inventory (BPI)17,18 measured pain site, severity, and pain interference. Participants reported whether they had pain in 17 different body locations (e.g., neck, lower back, upper back, legs etc.) which was used to define number of pain sites. Pain severity was measured via 4-items: worst in last 30 days, least in last 30 days, pain on average, and current pain. Pain severity was the average of these 4-items on a scale from “0=no pain” to “10=pain as bad as you can imagine.” Seven pain interference questions assessed whether pain has interfered with general activity, mood, walking ability, normal work, relationships, sleep, and enjoyment of life in the last 30 days. The pain interference score was the average of these 7-items on a scale of “0=does not interfere” to “10=completely interferes.”

Prescription Opioid Measures

We adjusted for baseline morphine equivalent dose (MME) and the baseline score on the Prescribed Opioids Difficulty scale (PODS).19 We used only the baseline MME and PODS because patients no longer using opioids at 6-month follow-up did not receive these assessments in follow-up surveys. The PODS measures psychosocial problems and concerns about opioid use. Higher scores indicate greater problems with opioids with scores ≥ 16 considered high.

Mental Health Measures

Mental health measures included generalized anxiety measured via the GAD-7, anhedonia measured via the Snaith-Hamilton Pleasure Scale (SHAPS) and vital exhaustion measured using the Maastricht Vital Exhaustion brief form. The GAD-7 measures general anxiety with higher scores indicating worse anxiety and a score ≥ 15 indicating severe anxiety.20 The SHAPS measures anhedonia with higher scores indicating more severe anhedonia. A score ≥ 3 indicates high anhedonia.21 Vital exhaustion is a measure of “unusual fatigue, increased irritability and feelings of demoralization,” and higher scores indicate worse vital exhaustion. A score of ≥10 indicates high vital exhaustion.22

Social Support and Date of Survey Completion

The PROMIS SF v2.0–Emotional Support 4a scale measured emotional support.23 Higher scores indicate more social support. Using ≥60 as a cut-off, we dichotomized the measure into low and high social support. Last, because the pandemic has waxed and waned, we controlled for the following time periods when surveys were completed: 3/2/20–12/31/20, 1/1/21–11/1/21, 11/2/21–present, or unknown. Measures obtained from baseline and 6-month follow-up are shown in Fig. 1.

Figure 1
figure 1

Timing of assessments used in present study.

Full size image

Analytic Approach

Change scores were computed for pain severity, pain interference, number of pain sites, generalized anxiety, anhedonia, vital exhaustion, and emotional support by subtracting baseline scores from 6-month scores. Therefore, a change score of 1-point represents an increase from baseline to 6-month follow-up. For all measures, except emotional support, decreasing scores represent improvement. All other covariates were measured at baseline only.

Analyses were performed with SAS v9.4 (SAS Institute, Cary, NC) at an alpha level of 0.05. A repeated measures difference-in-difference analysis, using a 2 (baseline, 6-month) × 2 (persist, stop) factorial repeated measures ANOVA for each pain, mental health, and social support variable, was conducted to assess whether change in these variables from baseline to 6-month follow-up is different based on whether patients persisted or stopped opioid use (see e-Table 1).

Bivariate Analyses

Chi-square tests estimated the bivariate association of each COVID-19 item and persistent opioid use at 6-month follow-up. Separate, bivariate logistic regression analyses estimated the association between COVID-19 stressors, demographics, change in pain measures, baseline MME and baseline PODS and change in mental health measures and persistent opioid use using odds ratios and 95% confidence intervals.

Multivariate Analyses

Multivariate logistic regression models estimated the association between COVID-19 stressors and odds of persistent opioid use. Subsequent models first adjusted for demographic variables, then change in pain variables, followed by adjustment for baseline MME and baseline PODS and change in mental health and social support measures. A fully adjusted model included all covariates. The associations between COVID-19 items, covariates, and persistent opioid use were expressed as odds ratios with 95% confidence intervals. Collinearity was evaluated by computing the variance inflation factor (VIF) and tolerance, both of which revealed no evidence of collinearity.

Sensitivity Analyses

Persistent opioid use may include a mix of intermittent and daily opioid users. To determine if COVID-19-related stressors were more strongly associated with persistent daily vs. persistent non-daily opioid use, we computed a fully adjusted multivariate, multinomial model. The primary outcome was no opioid use, non-daily opioid use, and daily opioid use at 6-month follow-up.

RESULTS

Sample characteristics at baseline and 6-month follow-up are shown in Table 1. Participants’ average age was 53.9 (±11.4), 67.1% were White race, and 27.4% Black race. The majority were women (70.9%) and nearly half were married (48.8%). The most frequently endorsed COVID-19 item was worry about health (85.7%) and the least often endorsed item was worsening pain due to COVID-19. Among all participants, 65.8% became persistent opioid users. The average baseline pain severity score was 5.9±1.8, average baseline pain interference score was 6.7±2.3, and mean number of pain sites at baseline was 6.0±3.8. Baseline MME dose was 27.9±31.4. The prevalence of specific pain sites is shown in the appendix, e-table 1. The most prevalent site was low back pain.

Table 1 Sample Characteristics (n=477) Who Completed all COVID-19 Stress Questions
Full size table

Difference-in-difference analysis for all continuous measures assessed at baseline and 6-month follow-up by persistent opioid use status is shown in supplementary appendix, e-table 2. Pain severity (p=0.0009) and pain interference (p=0.0085) decreased significantly less among persistent opioid users compared to those who stopped opioid use by 6-month follow-up. Generalized anxiety severity decreased significantly (p=0.0027) among persistent opioid users but remained the same among those who stopped. Change in anhedonia scores (p=0.006) and change in vital exhaustion (p=0.0081) significantly differed by persistent opioid use status. Those who stopped opioids had a significant increase in anhedonia while those who persisted had no change. Conversely, those who stopped opioids had no change in vital exhaustion but those with persistent opioid use had a significant decrease. Change in emotional support differed between those who stopped opioid use vs. those who had persistent opioid use (p=0.0240). Emotional support significantly decreased among those who stopped opioid use but there was no change among persistent opioid users. There was no significant difference in number of pain sites between baseline and follow-up among patients who stopped vs. remained on opioids.

As shown in Figure 2, patients who endorsed COVID-19 related “stressful changes in social/family contacts” were significantly (p=0.019) more likely to be persistent opioid users compared to those who did not endorse this item (70.1% vs. 59.8%). Those who reported “worsened pain due to pandemic” were significantly (p=0.005) more likely to be persistent opioid users compared to those who did not report worsened pain (76.0% vs. 62.2%). Other COVID-19-related stressors were not significantly associated with persistent opioid use.

Figure 2
figure 2

Persistent opioid use at 6-month follow-up by COVID-19-related stressors.

Full size image

The bivariate associations between each COVID-19 stressor and odds of persistent opioid use are shown in Table 2. Participants who did vs. did not report experiencing “stressful change in social/family contacts” were 58% more likely to be persistent opioid users (OR=1.58; 95%CI: 1.08–2.32). Participants who did vs. did not report “worsened pain due to pandemic” were 92% more likely to be persistent opioid users (OR=1.92; 95%CI: 1.21–3.06). There were no significant bivariate associations between other COVID-19 stressors and persistent opioid use.

Table 2 Unadjusted Association Between Individual COVID-19 Stressors, Covariates, and Persistent Opioid Use
Full size table

Changes in pain severity and pain interference were significantly associated with persistent opioid use (OR=1.21; 95%CI: 1.08–1.36 and OR=1.12; 95%CI: 1.03–1.22, respectively). Changes in generalized anxiety, anhedonia, and vital exhaustion and emotional support were significantly associated with small decreases in odds of persistent opioid use. Change in emotional support was associated with a minimal increase in odds of persistent opioid use.

Multivariate models estimating the association between COVID-19 stressors at 6-month follow-up and persistent opioid use at 6-month follow-up before and after adjusting for covariates are reported in Table 3. After simultaneous adjustment for all COVID-19 stressors (model 1), only “worsened pain due to pandemic” remained significantly associated with persistent opioid use (OR=1.83; 95%CI: 1.09–3.04). This association remained largely unchanged after adjusting for baseline demographics (model 2). The association between “worsened pain due to pandemic” and persistent opioid use at 6-month follow-up was attenuated and no longer statistically significant after adjusting for change in pain severity, pain interference, and number of pain sites in model 3. As shown in model 4, after adjusting for baseline MME, baseline PODS, and change in generalized anxiety, anhedonia, vital exhaustion, and emotional support, the association was of greater magnitude and statistically significant (OR=3.52; 95%CI: 1.72–7.20). In the fully adjusted model (model 5), “worsened pain due to pandemic” remained significantly associated with persistent opioid use (OR=2.88; 95%CI: 1.33–6.22). In the full model, a 1-point larger increase in change in pain interference was associated with greater odds of persistent opioid use at 6-month follow-up (OR=1.20; 95%CI: 1.04–1.38). Conversely, a 1-point larger increase in change in vital exhaustion was associated with lower odds of persistent use (OR=0.90; 95%CI: 0.82–0.99). Other covariates were not significantly associated with persistent opioid use.

Table 3 Association Between Type of COVID-19 Stress and Worsening Pain, Mental Health, and Persistent Opioid Use (Outcome Is Opioid Use at 6-Month Follow-up (Yes/No))
Full size table

Multinomial model results are shown in e-table 3. “Worsened pain due to pandemic” was significantly associated with persistent daily opioid use at 6-month follow-up (OR=3.82; 95%CI: 1.61–9.08) but was not significantly associated with non-daily opioid use at 6-month follow-up (OR=2.09; 95%CI: 0.86–5.10).

DISCUSSION

Among a cohort of persons with non-cancer pain and a new period of 30–90-day prescription opioid use at baseline, we observed that after multivariate adjustment, patients who perceived the COVID-19 pandemic worsened their pain were nearly 3 times more likely to remain on opioids at 6-month follow-up. Change in pain interference and vital exhaustion remained significant in multivariate models. Pain interference declined less among persistent opioid users compared to opioid quitters which is consistent with our observation that for each unit increase in change in pain interference there was 20% greater odds of persistent opioid use. Vital exhaustion scores did not change in those who stopped opioids and decreased in persistent users which is consistent with evidence that each unit increase in change in vital exhaustion was associated with 10% lower odds of persistent opioid use.

There was some evidence that perceiving the pandemic worsened pain had a greater association with daily vs. non-daily persistent opioid use at 6-month follow-up. It is possible that those who attribute greater pain to the pandemic were more likely to be daily opioid users. However, the multinomial model led to smaller cell sizes and large confidence intervals which prohibit strong conclusions regarding these associations.

This study advances current understanding of the numerous COVID-19 pandemic–related stressors thought to be associated with prescription opioid use. In bivariate models, only stressful change in social/family contacts was significantly associated with persistent opioid use. However, this association did not remain after accounting for the perception that the pandemic worsened pain and other covariates. Clinicians screening for LTOT risk factors related to the COVID-19 pandemic may consider focusing on asking patients if they attribute increasing pain to the pandemic.

There are several potential explanations for the strong association between perceiving pain was made worse by the pandemic and persistent opioid use. Barriers to non-pharmacological pain treatments have occurred during the pandemic7,8 and patients may attribute worsening pain due to COVID-19 if access to pain management was restricted. A byproduct of the pandemic is chronic stress which is a risk factor for chronic pain and long-term opioid use.24,25 Another explanation for our results could be that those who reported pain was worsened by the pandemic also perceived such pain as an injustice (e.g., belief that the pandemic was mismanaged). Persons who perceive pain is unjust are less likely to accept pain in their lives, exhibit more expressive pain, and are more likely to be prescribed opioids.26,27 Last, patients may attribute worsening pain due to the pandemic because they became infected with COVID-19; in severe cases, this can lead to posttraumatic stress disorder and depression,28 both conditions associated with worse pain outcomes and greater risk for LTOT.

Our results are largely consistent with an exploratory study of change in pain experiences before, during, and after the first COVID-19 wave29 which revealed chronic pain was linked to cognitive overload and impaired ability to cope with pandemic related stress.29 However, the same patients had little change in pain severity and experienced improvement in pain interference.29 A longitudinal study of 1500 persons with chronic pain revealed no significant decline in pain severity and pain interference during the first year of the pandemic.6 The limited variation in pain severity and pain interference is consistent with our results.

Although changes in pain interference and vital exhaustion were significantly associated with persistent opioid use, examination of mean scores for both measures reveals very small differences between those who stopped vs. continued opioid use at follow-up. Thus, it is not clear that clinically meaningful change occurred or that clinically meaningful differences exist between those who quit vs. those who persisted.

Strengths and Limitations

It is possible that non-response biased our study. However, the current 6-month retention rate is about 82% which reduces concerns about non-response bias. The Pathways Study is primarily designed to determine if long-term prescription opioid use leads to incident depression. We did not include depression in the current study because it is measured using a diagnostic interview at baseline and 12-month follow-up and thus with only 6-month follow-up data we would not be able to model change in depression.

The cohort was recruited from two large metropolitan areas in the middle of the USA and may not generalize to other locations. Results could differ in regions, such as the eastern U.S. where excess mortality was greatest during the early phase of the pandemic.30 However, the cohort’s racial diversity improves generalizability. Self-reported opioid use could be biased; however, research on the validity of self-reported medications that could be stigmatizing, such as antidepressants and benzodiazepines, indicates excellent agreement between self-report and pharmacy records and medical claims.31,32 We are unable to draw conclusions about the temporal direction of COVID-19 stress and persistent opioid use because both were measured at the 6-month follow-up survey, and our change scores in this observational cohort study do not estimate causal effects.33

Conclusions

Patients who attributed worsening pain to the COVID-19 pandemic were more likely to remain prescription opioid users over a 6-month observation period and may be more likely to be persistent daily as compared to non-daily opioid users. Worsening pain interference, but not pain severity, was a risk factor for persistent opioid use. Further quantitative and qualitative research is needed to identify patient experiences with the COVID-19 pandemic that explain these associations. Clinicians who prescribe opioids may consider discussing pain in the context of the pandemic to identify patients at risk for long-term opioid use. Further research is warranted to determine if pandemic-related stress mediates or moderates the relationships between mental illness and risk for opioid use disorder and between long-term prescription opioid use and worsening mood and depression.