Evaluating chronic pain as a risk factor for COVID-19 complications among New York State Medicaid beneficiaries: a retrospective claims analysis

Abstract

Objective

To assess whether chronic pain increases the risk of COVID-19 complications and whether opioid use disorder (OUD) differentiates this risk among New York State Medicaid beneficiaries.

Design, Setting, and Subjects

This was a retrospective cohort study of New York State Medicaid claims data. We evaluated Medicaid claims from March 2019 through December 2020 to determine whether chronic pain increased the risk of COVID-19 emergency department (ED) visits, hospitalizations, and complications and whether this relationship differed by OUD status. We included beneficiaries 18–64 years of age with 10 months of prior enrollment. Patients with chronic pain were propensity score-matched to those without chronic pain on demographics, utilization, and comorbidities to control for confounders and were stratified by OUD. Complementary log–log regressions estimated hazard ratios (HRs) of COVID-19 ED visits and hospitalizations; logistic regressions estimated odds ratios (ORs) of hospital complications and readmissions within 0–30, 31–60, and 61–90 days.

Results

Among 773 880 adults, chronic pain was associated with greater hazards of COVID-related ED visits (HR = 1.22 [95% CI: 1.16–1.29]) and hospitalizations (HR = 1.19 [95% CI: 1.12–1.27]). Patients with chronic pain and OUD had even greater hazards of hospitalization (HR = 1.25 [95% CI: 1.07–1.47]) and increased odds of hepatic- and cardiac-related events (OR = 1.74 [95% CI: 1.10–2.74]).

Conclusions

Chronic pain increased the risk of COVID-19 ED visits and hospitalizations. Presence of OUD further increased the risk of COVID-19 hospitalizations and the odds of hepatic- and cardiac-related events. Results highlight intersecting risks among a vulnerable population and can inform tailored COVID-19 management.

Introduction

More than 1 in 5 adults in the United States have chronic pain.¹ Individuals with chronic pain have been uniquely challenged by the COVID-19 pandemic, as pandemic-related stressors can exacerbate chronic pain,² and SARS-CoV-2 infection could be associated with post-infection chronic pain.³ Nevertheless, whether preexisting chronic pain predisposes individuals to COVID-19 complications remains unclear yet critical in identifying a potentially vulnerable population for whom tailored public health and clinical interventions might be necessary.⁴

There are several potential mechanisms underlying the relationship between chronic pain and COVID-19 complications. First, chronic pain could be associated with an increased risk of severe COVID-19 because of underlying inflammatory mechanisms of chronic pain.⁵ Similarly, chronic pain could be linked to immune-suppressive effects,⁶ which play a role in the pathophysiological response to COVID-19.⁷ Additionally, individuals with chronic pain often experience physical disability, which negatively impacts access to COVID-19 resources,⁴ including testing kits to detect infection, pulse oximeters to monitor symptoms, medications to alleviate symptoms, and clinical care to manage illness. Immobility might also negatively impact pre-infection thromboembolic and cardiorespiratory function, which could increase COVID-19 complications.⁴

These pathways between chronic pain and COVID-19 complications could be confounded by several factors. Older age⁸ and rural place of residence are associated with increased risk of chronic pain and COVID-19 complications,⁹ while female sex is associated with increased chronic pain¹⁰ but decreased risk of COVID-19 complications.¹¹ Race/ethnicity could also confound this relationship, as African American and Black individuals experience a greater burden of pain¹² and higher rates of severe COVID-19 than do non-Hispanic White individuals.¹³ Poorer access to health care services is also associated with increased risk of chronic pain and COVID-19 complications.¹⁴ Finally, several comorbidities, such as heart disease, smoking, diabetes, and obesity, might also increase the risk of both chronic pain and COVID-19 complications.¹⁵

As the COVID-19 pandemic collides with the opioid epidemic in the United States,¹⁶ the relationship between chronic pain and COVID-19 complications might also differ by the presence of comorbid opioid use disorder (OUD). Opioids play an important role in chronic pain management,¹⁷ and long-term opioid use can result in OUD.¹⁸ OUD increases cardiovascular, pulmonary, and metabolic complications,¹⁹ which are known risk factors for COVID-19 complications. Nonetheless, whether individuals with chronic pain experience differential risks of COVID-19 complications depending on their history of OUD remains unknown.

Overall, three recent United Kingdom (UK)-based studies found that chronic pain is associated with higher risk of COVID-19 infection and potentially hospitalization and death.^4,20,21 However, these studies were limited to UK populations and did not evaluate several relevant COVID-19 complication outcomes, including respiratory complications, acute renal failure, and hospital readmissions, nor did they account for differences based on OUD history.

Risks of negative health outcomes related to chronic pain, with or without OUD, can be pronounced in low-income individuals without access to sufficient care.²² The present study therefore evaluates New York State (NYS) Medicaid beneficiaries, an underserved population that experienced high exposure to COVID-19 in 2020.²³ We tested whether chronic pain was a risk factor for COVID-19 complications and whether this relationship differed in individuals with and without OUD. We estimated the effects of chronic pain, adjusting for the aforementioned confounders, including demographics, baseline resource utilization, and comorbidities, and stratified by OUD history status. We hypothesized that (1) chronic pain increases the risk of COVID-19 complications and (2) patients with chronic pain and comorbid OUD would have a greater risk of COVID-19 complications than would patients with chronic pain alone.

Methods

Data source and study design

Using a retrospective cohort design, we analyzed NYS Medicaid claims from March 1, 2019, through December 31, 2020 (study window). We included patients who met the following criteria: enrollment in NYS Medicaid in February 2020 (to identify a baseline population enrolled before the start of the pandemic in March 2020) and for ≥10 months between March 2019 and February 2020, and 18–64 years of age as of February 29, 2020. We excluded patients who met any of the following criteria: dually eligible for Medicaid and Medicare (to ensure we did not miss health care encounters for patients); missing county of residence; or having recent history of cancer, palliative care, or long-term care from March 1, 2019, through February 29, 2020 (baseline period; Supplemental Appendix 1). Beginning on March 1, 2020 (index date), we followed adults until the earliest of (1) first COVID-19 emergency department (ED) visit or hospitalization (evaluated separately), (2) death from any cause (where data were available), (3) final disenrollment from their Medicaid health plan (where gaps in enrollment were excluded from analyses, but patients could re-enter analyses if they re-enrolled in the Medicaid health plan during the study window), or (4) December 2020 (end of study window). The study protocol was reviewed and approved by the New York University Langone Health Institutional Review Board (ID: i18-01221).

Study population

Supplemental Appendix 2 presents the cohort attrition steps and counts for cohort selection. On the basis of previously published algorithms,^24–28 patients were included in the chronic pain cohort if they had ≥2 baseline claims (from March 1, 2019, through February 29, 2020) that were between 90 and 366 days apart^27,29,30 for chronic pain; these claims were based on presence of ICD-10-CM diagnosis codes³¹ in the inpatient or outpatient settings within the same category and body region of chronic pain for the following conditions: arthritis, back pain, neck pain, unspecified back and/or neck pain, neurological pain, chronic headache, or miscellaneous chronic pain (Supplemental Appendix 1). Patients who satisfied selection criteria but did not have any diagnoses for chronic pain were eligible controls. We further stratified patients with chronic pain and controls by the presence of an OUD diagnosis at baseline based on ICD-10-CM diagnosis codes (Supplemental Appendix 1). We conducted analyses at the patient-month level.

Outcome measures

ED visits for COVID-19 were identified on the basis of claims in the ED setting with a primary diagnosis of COVID-19 (ICD-10-CM diagnosis code U071). Hospitalizations for COVID-19 were identified on the basis of claims in the inpatient setting with a primary or admitting COVID-19 diagnosis. Within each hospital encounter, complications were identified on the basis of diagnosis and ICD-10-PCS procedure and Current Procedural Terminology (CPT) codes for the following: ventilation (for respiratory distress/failure/arrest), non-ventilation respiratory complications (including pneumonia, bronchitis, asphyxia, and hypoxemia), acute renal failure, sepsis, and other complications (including rhabdomyolysis and hepatic failure, and cardiovascular complications, such as venous thrombosis, acute and viral carditis, and cardiac arrest; Supplemental Appendix 1). Readmissions were identified on the basis of inpatient hospital claims for any cause after a discharge from a primary COVID-19–related hospitalization within 0-30 days, 31-60 days, and 61-90 days.

Other measures

Demographics included age, race/ethnicity, sex, NYS health service area, and individuals’ status of aged/blind/disabled based on Medicaid’s benefits provision of “community Medicaid” eligibility.³² Health care utilization during the baseline period was captured as each of the following: proportion of patients with ≥1 ED visit, proportion with ≥1 hospitalization, mean number of primary care visits, log drug-related health expenditures, and log non–pharmacy-related health expenditures.

History of comorbidities thought to be confounders in the relationship between chronic pain and COVID-19 was captured from March 2017 to February 2020 on the basis of claims in any setting with diagnosis codes (Supplemental Appendix 1). Conditions were selected from the Centers for Disease Control and Prevention list of underlying medical conditions that increase the risk of severe COVID-19 illness as of March 2021, as well as mental health and substance use conditions hypothesized to confound the relationship between chronic pain and COVID-19 severity. These conditions included anxiety disorders; bipolar disorder; major depressive disorders; post-traumatic stress disorder; nicotine dependence; history of traumatic life events; heart conditions, including cardiomyopathies, coronary artery disease, and heart failure; chronic kidney diseases; chronic lung diseases; chronic obstructive pulmonary disease; Down syndrome; HIV/AIDS and other immunocompromised conditions; overweight/obesity; sickle cell disorders; diabetes mellitus types 1 and 2; solid organ or blood stem cell transplants; and pregnancy (evaluated only within 3 months before and after index, as it is not a chronic condition).³³

Matching strategy

Propensity score matching was used to account for confounding by creating groups that were similar on confounding covariates but differed on chronic pain status. Patients with chronic pain were matched to eligible controls through the use of 5:1 nearest-neighbor Mahalanobis distance (MD) matching with replacement based on demographics, health care utilization, and comorbidities.³⁴ Variables included in the propensity score are presented in Supplemental Appendices 3–5. Within matched groups, MD matching was conducted on characteristics measured at baseline. Nearest-neighbor matches were restricted to the area of common support, overlap weights were calculated, and the propensity score model was re-estimated with these weights. Standardized mean differences (SMDs) of overlap-weighted means were assessed and considered well balanced when <0.1.³⁵

Pre- and post-match frequencies and SMDs of demographics, baseline health care utilization, and prevalence of clinical characteristics were tabulated. Weighted counts and proportions were used to describe categorical variables; weighted means and standard deviations (SDs) were used to describe continuous variables.

Secondary analysis

We conducted an exploratory, secondary analysis to examine the risks of COVID complications among patients with chronic pain and OUD, while additionally stratifying the population by receipt of medication for opioid use disorder (MOUD) at baseline to address potential clinical heterogeneity among individuals with OUD depending on their treatment status. MOUD was identified during the 12-month period before the COVID-19 pandemic and defined as having at least one outpatient pharmacy prescription filled for buprenorphine or injected naltrexone using National Drug Codes or a visit to a methadone maintenance treatment program or opioid treatment program using rate codes (Supplemental Appendix 1).

Statistical analyses

We computed weighted descriptive characteristics and outcome frequencies by chronic pain status, with and without OUD. Hazard ratios (HRs) for COVID-19-related ED visits and hospitalizations were calculated with discrete time-proportional hazards, or complementary log–log transformed models,³⁶ to account for person-month-level data; Kaplan-Meier survival curves were constructed to visualize time to ED visits and hospitalizations.³⁷ Results were compared for patients with chronic pain and those without chronic pain, overall. To assess results stratified by the presence of chronic pain and baseline OUD, standardized mean differences of overlap-weighted means were reassessed within each stratum of patients, and any imbalanced covariates were further adjusted within each model evaluating chronic pain and OUD strata (Supplemental Appendix 6).

The sample was limited to patients hospitalized for COVID-19, and covariate balance was reassessed between cohorts with and without chronic pain (Supplemental Appendix 7). Imbalanced covariates were included in subsequent models. For models stratified by presence of chronic pain and OUD, weighted balance was reassessed (Supplemental Appendix 8), and remaining imbalanced covariates were included in the models. Logistic regression models were estimated to calculate the odds ratios (ORs) of experiencing each complication: ventilation, non-ventilation respiratory complications, acute renal failure, sepsis, and other hepatic and cardiac-related events. To estimate the odds of hospital readmissions within 0–30, 31–60, and 61–90 days, the hospitalized sample was restricted to patients hospitalized with sufficient follow-up time in each respective readmission window. All models included propensity score matching weights and robust standard errors.

Analyses were conducted with R statistical software (version 1.4.1717) and Stata (StataCorp, 2020; Stata Statistical Software: Release 17, College Station, TX, USA, StataCorp, LLC). Given that this analysis was exploratory, the statistical significance was set at P < 0.05.

Results

Sample characteristics

Among 773 880 matched patients with and without chronic pain, all demographics, utilization, and comorbidities were well-balanced (SMD < 0.1; Table 1). In the unweighted sample, 236 391 (30.5% of the matched sample) patients with chronic pain were matched to 537 489 controls without chronic pain. In the weighted sample, a total of 386 940 patients with chronic pain were compared to the same number without chronic pain, of whom 4561 were hospitalized for COVID-19 in the study window (54.4% of whom had chronic pain). Total weighted frequencies and proportions of all outcomes are presented in Supplemental Appendix 3 and are further stratified by chronic pain and OUD in Supplemental Appendix 4.

Table 1.

Demographics and baseline clinical characteristics in matched chronic pain versus no chronic pain cohorts, before and after match.

	Before match			After match
	Chronic pain	No chronic pain	SMDa	Chronic pain	No chronic pain	SMDa
	n = 236 391	n = 1 831 479		n = 236 391	n = 537 489
Male sex, %	35.4%	47.5%	–0.24	36.9%	36.9%	0.00
Race, %
Unknown	20.2%	30.1%	–0.22	21.3%	21.3%	0.00
White	29.7%	28.0%	0.04	30.2%	30.2%	0.00
Black / African American	17.4%	17.4%	0.00	17.3%	17.3%	0.00
Asian	14.3%	11.5%	0.09	14.0%	14.0%	0.00
Other	5.4%	5.3%	0.01	5.4%	5.4%	0.00
Hispanic	13.0%	7.7%	0.19	11.7%	11.7%	0.00
Health service area, %
Central NY	6.1%	5.5%	0.03	6.1%	6.1%	0.00
Finger Lakes	6.0%	5.1%	0.04	6.0%	6.0%	0.00
Mid-Hudson	7.8%	10.0%	–0.07	8.1%	8.1%	0.00
Nassau-Suffolk	8.2%	9.9%	–0.06	8.4%	8.4%	0.00
New York City	58.4%	57.4%	0.02	58.0%	58.0%	0.00
Northeastern NY	4.8%	5.1%	–0.01	5.0%	5.0%	0.00
NY-Pennsylvania	1.0%	1.0%	0.01	1.1%	1.1%	0.00
Western NY	7.5%	6.1%	0.06	7.3%	7.3%	0.00
Age, years (mean)	46.4	36.8	0.72	44.8	44.8	0.00
Aged, blind, or disabled, %	18.8%	6.7%	0.45	15.6%	15.6%	0.00
≥1 Baseline hospitalization, %b	14.4%	7.3%	0.26	13.3%	13.3%	0.00
≥1 Baseline emergency department visit, %b	40.8%	24.1%	0.38	39.1%	39.1%	0.00
Number of baseline primary care visits (mean)b	8.43	2.69	0.47	7.07	7.07	0.00
Log baseline nonpharmacy health care costs, $ (mean)b	$7.41	$4.42	0.84	$7.17	$7.17	0.00
Log baseline prescription drug costs, $ (mean)b	$6.33	$2.94	1.12	$5.94	$5.94	0.00
Preexisting comorbidities, %c
Anxiety	39.7%	18.9%	0.51	37.2%	37.2%	0.00
Obesity	44.2%	21.6%	0.53	40.6%	40.6%	0.00
Major depressive disorder	17.6%	6.3%	0.43	15.4%	15.4%	0.00
Bipolar disorder	9.3%	3.8%	0.27	8.6%	8.6%	0.00
Chronic kidney disease	8.8%	2.2%	0.38	7.0%	7.0%	0.00
Chronic lung diseased	12.0%	2.8%	0.48	9.6%	9.6%	0.00
Type 2 diabetes	37.1%	10.1%	0.80	30.9%	30.9%	0.00
Down syndrome	0.1%	0.1%	–0.01	0.1%	0.1%	0.00
Heart conditionse	12.4%	3.0%	0.48	9.9%	9.9%	0.00
HIV or immunocompromisedf	4.0%	1.8%	0.15	3.8%	3.8%	0.00
Post-traumatic stress disorder	8.6%	3.1%	0.29	7.8%	7.8%	0.00
Pregnancyg	1.9%	2.0%	–0.01	2.1%	2.1%	0.00
Sickle cell disease	0.8%	0.5%	0.04	0.8%	0.8%	0.00
Smoking	34.2%	19.7%	0.35	32.8%	32.8%	0.00
Traumatic life events	1.1%	0.6%	0.07	1.1%	1.1%	0.00

^aSMD = standardized mean difference to assess covariate balance after matching, where <0.1 is considered well balanced.

^bOne-year lookback period from March 2019 through February 2020 (inclusive).

^cThree-year lookback period from March 2017 through February 2020 (inclusive).

^dIncludes chronic obstructive pulmonary disease, emphysema, bronchitis, and pulmonary emphysema.

^eIncludes heart failure, coronary artery disease, cardiomyopathies, and Takotsubo syndrome.

^fIncludes organ transplants, immunodeficiencies involving blood and blood-forming organs, and immunoregulatory T-cell disorders.

^gBased on having ≥1 diagnosis ±3 months from index date of February 29, 2020.

Hazard ratios for COVID-19 ED visits and hospitalizations

Chronic pain was significantly associated with COVID-19 ED visits (HR = 1.22 [95% CI: 1.16–1.29]; Table 2; Figure 1A) and hospitalizations (HR = 1.19 [95% CI: 1.12–1.27]; Table 2; Figure 1C). Although individuals with both chronic pain and OUD together, relative to neither condition, did not experience increased hazards of COVID-19–related ED visits (Table 2; Figure 1B), they had 1.25 times increased hazards of being hospitalized for COVID-19 (95% CI: 1.07–1.47, Table 2; Figure 1D). Mean follow-up time for ED visits was 10.74 and 10.67 months for patients with and without chronic pain, respectively (10.35 for those with chronic pain and OUD; 10.69 for those with chronic pain and without OUD). For hospitalizations, mean follow-up time was 10.73 and 10.67 months for those with and without chronic pain, respectively (10.61 for those with chronic pain and OUD; 10.69 for those with chronic pain and without OUD).

Table 2.

Complementary log–log regression model hazard ratios for time to first COVID-19–related ED visit and hospitalization among patients with chronic pain versus no chronic pain and further stratified by OUD status.

	HR (95% CI)	P value
ED: chronic pain vs no chronic pain
No chronic pain	REF
Chronic pain	1.22 (1.16–1.29)	<0.001
ED: chronic pain vs no chronic pain, stratified by OUDa
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	0.66 (0.55–0.79)	<0.001
Chronic pain (+), OUD (–)	1.20 (1.14–1.27)	<0.001
Chronic pain (+), OUD (+)	0.97 (0.81–1.16)	0.751
Hospitalization: chronic pain vs no chronic pain
No chronic pain	REF
Chronic pain	1.19 (1.12–1.27)	<0.001
Hospitalization: chronic pain vs no chronic pain, stratified by OUDa
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	0.75 (0.62–0.90)	0.003
Chronic pain (+), OUD (–)	1.15 (1.08–1.22)	<0.001
Chronic pain (+), OUD (+)	1.25 (1.07–1.47)	0.004

Abbreviations: ED= emergency department; HR= hazard ratio; OUD= opioid use disorder.

^aComplementary log–log regression models evaluating chronic pain vs no chronic pain by presence vs absence of OUD adjusted for the following covariates that were imbalanced after matching: sex, race, aged/blind/disabled, health service area, number of baseline primary care visits, log health care costs, log drug costs, anxiety, major depressive disorder, bipolar disorder, chronic kidney disease, chronic lung disease, heart conditions, immunologic disorders, and post-traumatic stress disorder.

Figure 1.

Kaplan-Meier curves for time to first COVID-19–related ED visits among patients with chronic pain versus no chronic pain (A) and further stratified by OUD (B) and time to first COVID-19–related hospitalization among patients with chronic pain versus no chronic pain (C) and further stratified by OUD (D).

Odds ratios for morbidity complications

Table 3 presents the ORs of hospital-based complications for those with versus without chronic pain among patients hospitalized for COVID-19, further stratified by OUD, after adjustment for imbalanced covariates. Chronic pain, overall, was not associated with increased odds of acute renal failure; however, when stratified by baseline OUD status, patients with both chronic pain and OUD had greater odds of acute renal failure than did those with neither condition (OR, 1.45 [95% CI: 1.00–2.09]). Individuals with chronic pain and OUD also experienced increased odds of other hepatic and cardiac-related complications (OR, 1.74 [95% CI: 1.10–2.74]).

Table 3.

Adjusted logistic regression of the relationship between chronic pain, and chronic pain and opioid use disorder, with COVID-19 hospital-based complications.

	OR (95% CI)	P value
Ventilation: chronic pain vs no chronic paina
No chronic pain	REF
Chronic pain	1.08 (0.93–1.25)	0.321
Ventilation: chronic pain vs no chronic pain with vs without OUDb
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	1.49 (0.95–2.34)	0.086
Chronic pain (+), OUD (–)	1.13 (0.97–1.32)	0.114
Chronic pain (+), OUD (+)	1.26 (0.87–1.83)	0.230
Respiratory complications (non-ventilation): chronic pain vs no chronic paina
No chronic pain	REF
Chronic pain	1.12 (0.97–1.29)	0.113
Respiratory complications (non-ventilation): chronic pain vs no chronic pain with vs without OUDb
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	1.03 (0.67–1.58)	0.894
Chronic pain (+), OUD (–)	1.16 (1.00–1.34)	0.053
Chronic pain (+), OUD (+)	1.16 (0.83–1.63)	0.383
Acute renal failure: chronic pain vs no chronic paina
No chronic pain	REF
Chronic pain	1.06 (0.92–1.22)	0.396
Acute renal failure: chronic pain vs no chronic pain with vs without OUDb
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	0.93 (0.58–1.48)	0.747
Chronic pain (+), OUD (–)	1.08 (0.94–1.26)	0.285
Chronic pain (+), OUD (+)	1.45 (1.00–2.09)	0.048
Sepsis: chronic pain vs no chronic paina
No chronic pain	REF
Chronic pain	1.00 (0.84–1.18)	0.963
Sepsis: chronic pain vs no chronic pain with vs without OUDb
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	1.24 (0.73–2.11)	0.430
Chronic pain (+), OUD (–)	1.01 (0.85–1.21)	0.871
Chronic pain (+), OUD (+)	1.18 (0.76–1.84)	0.464
Other complications (hepatic failure, cardiac-related events): chronic pain vs no chronic paina
No chronic pain	REF
Chronic pain	1.02 (0.84–1.25)	0.809
Other complications (hepatic failure, cardiac-related events): chronic pain vs no chronic pain with vs without OUDb
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	2.11 (1.25–3.57)	0.005
Chronic pain (+), OUD (–)	1.05 (0.86–1.29)	0.629
Chronic pain (+), OUD (+)	1.74 (1.10–2.74)	0.017
0–30 Day readmission: chronic pain vs no chronic paina,c
No chronic pain	REF
Chronic pain	1.06 (0.85–1.32)	0.605
0–30 Day readmission: chronic pain vs no chronic pain with vs without OUDb,c
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	0.84 (0.45–1.56)	0.582
Chronic pain (+), OUD (–)	0.97 (0.77–1.23)	0.817
Chronic pain (+), OUD (+)	1.45 (0.94–2.22)	0.092
31–60 Day readmission: chronic pain vs no chronic paina,d
No chronic pain	REF
Chronic pain	1.18 (0.88–1.58)	0.273
31–60 Day readmission: chronic pain vs no chronic pain with vs without OUDb,d
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	0.76 (0.36–1.62)	0.477
Chronic pain (+), OUD (–)	1.20 (0.88–1.64)	0.254
Chronic pain (+), OUD (+)	0.78 (0.45–1.36)	0.381
61–90 Day readmission: chronic pain vs no chronic paina,e
No chronic pain	REF
Chronic pain	0.99 (0.72–1.35)	0.938
61–90 Day readmission: chronic pain vs no chronic pain with vs without OUDb,e
Chronic pain (–), OUD (–)	REF
Chronic pain (–), OUD (+)	2.79 (1.44–5.40)	0.002
Chronic pain (+), OUD (–)	1.23 (0.87–1.74)	0.251
Chronic pain (+), OUD (+)	0.88 (0.48–1.60)	0.672

Abbreviations: OR= odds ratio; OUD= opioid use disorder.

^aLogistic regression models evaluating chronic pain vs no chronic pain adjusted for the following covariates that were imbalanced after matching: baseline hospitalization, log baseline health care costs, diabetes, and cardiovascular conditions.

^bLogistic regression models evaluating chronic pain vs no chronic pain by presence vs absence of OUD adjusted for the following covariates that were imbalanced after matching: sex, race, aged/blind/disabled, baseline hospitalization, baseline ED visit, number of baseline primary care visits, log health care costs, log drug costs, anxiety, major depressive disorder, bipolar disorder, chronic kidney disease, chronic lung disease, cardiovascular conditions, immunologic disorders, post-traumatic stress disorder, smoking status, traumatic events.

^c30-Day readmissions were evaluated among patients whose COVID-19 hospitalization occurred between March and November 2020.

^d60-Day readmissions were evaluated among patients whose COVID-19 hospitalization occurred between March and October 2020.

^e90-Day readmissions were evaluated among patients whose COVID-19 hospitalization occurred between March and September 2020.

Within 0–30, 31–60, and 61–90 days after COVID-19 hospitalization, chronic pain overall, as well as stratified by OUD status, was not associated with any-cause readmissions. Of the most frequent readmission diagnoses among individuals with and without chronic pain were conditions that could be risk factors for, or symptoms of, COVID-19, including shortness of breath (R06.02),³⁸ chest pain (R07.9),³⁹ chronic obstructive pulmonary disease with acute exacerbation (J44.1),⁴⁰ and pneumonia (J18.9)⁴¹; in addition, alcohol dependence (F10.20; F10.239) was relatively common.

Secondary analysis

Table 4 presents secondary exploratory analyses, which assessed whether significant associations from the primary analyses differed by receipt of MOUD. The hazards of ED visits among patients with chronic pain and OUD were consistent in magnitude and significance when stratified by MOUD receipt as compared with overall OUD stratification findings. As compared with the overall population with chronic pain and OUD, patients with chronic pain and OUD who did not receive MOUD experienced even higher hazards of COVID-19–related hospitalizations (HR = 1.36 [95% CI: 1.12–1.66]); those who received MOUD did not have increased hazards of hospitalizations (HR = 1.18 [95% CI: 0.94–1.48]). Similarly, among patients with chronic pain and OUD who did not receive MOUD, we observed exacerbated odds of acute renal failure (OR = 1.85 [95% CI: 1.19–2.95]) and other hepatic and cardiac conditions (HR = 2.07 [95% CI: 1.20–3.55]); among patients with chronic pain and OUD who received MOUD, the odds were attenuated and no longer significant for both sets of complications.

Table 4.

Complementary log–log regression model HRs^a for time to first COVID-19–related ED visit and hospitalization and adjusted logistic regression ORs^b for hospital-based complications among patients with chronic pain and OUD, stratified by receipt of MOUD^c.

	CP vs no CP: no MOUD		CP vs no CP: MOUD
	HR (95% CI)	P value	HR (95% CI)	P value
ED: chronic pain vs no chronic pain, stratified by OUD
Chronic pain (–), OUD (–)	REF		REF
Chronic pain (–), OUD (+)	0.65 (0.54–0.78)	<0.001	0.70 (0.58–0.85)	<0.001
Chronic pain (+), OUD (–)	1.20 (1.14–1.27)	<0.001	N/A
Chronic pain (+), OUD (+)	1.17 (0.94–1.46)	0.164	0.80 (0.60–1.06)	0.126
Hospitalization: chronic pain vs no chronic pain, stratified by OUD
Chronic pain (–), OUD (–)	REF		REF
Chronic pain (–), OUD (+)	0.74 (0.61–0.90)	0.003	0.77 (0.63–0.94)	0.009
Chronic pain (+), OUD (–)	1.15 (1.08–1.22)	<0.001	N/A
Chronic pain (+), OUD (+)	1.36 (1.12–1.66)	0.002	1.18 (0.94–1.48)	0.154
	OR (95% CI)	P value	OR (95% CI)	P value
Acute renal failure: chronic pain vs no chronic pain with vs without OUD
Chronic pain (–), OUD (–)	REF		REF
Chronic pain (–), OUD (+)	0.87 (0.54–1.40)	0.567	0.88 (0.53–1.45)	0.606
Chronic pain (+), OUD (–)	1.08 (0.93–1.25)	0.296	N/A
Chronic pain (+), OUD (+)	1.87 (1.19–2.95)	0.007	0.91 (0.52–1.57)	0.729
Other complications (hepatic failure, cardiac-related events): chronic pain vs no chronic pain with vs without OUD
Chronic pain (–), OUD (–)	REF		REF
Chronic pain (–), OUD (+)	2.04 (1.20–3.48)	0.008	2.35 (1.29–4.26)	0.005
Chronic pain (+), OUD (–)	1.05 (0.86–1.29)	0.619	N/A
Chronic pain (+), OUD (+)	2.07 (1.20–3.55)	0.008	1.52 (0.74–3.11)	0.250

Abbreviations: CP= chronic pain; ED= emergency department; HR= hazard ratio; MOUD= medications for opioid use disorder; OR= odds ratio; OUD= opioid use disorder.

^aComplementary log–log regression models evaluating chronic pain vs no chronic pain by presence vs absence of OUD adjusted for the following covariates that were imbalanced after matching: sex, race, aged/blind/disabled, health service area, number of baseline primary care visits, log health care costs, log drug costs, anxiety, major depressive disorder, bipolar disorder, chronic kidney disease, chronic lung disease, heart conditions, immunologic disorders, and post-traumatic stress disorder.

^bLogistic regression models evaluating chronic pain vs no chronic pain by presence vs absence of OUD adjusted for the following covariates that were imbalanced after matching: sex, race, aged/blind/disabled, baseline hospitalization, baseline ED visit, number of baseline primary care visits, log health care costs, log drug costs, anxiety, major depressive disorder, bipolar disorder, chronic kidney disease, chronic lung disease, heart conditions, immunologic disorders, post-traumatic stress disorder, smoking status, and traumatic events.

^cMOUD was identified during the 12-month period before the COVID-19 pandemic and defined as having at least one outpatient pharmacy prescription filled for buprenorphine or injected naltrexone using National Drug Codes or a visit to a methadone maintenance treatment program or opioid treatment program using rate codes.

Discussion

Our study identified chronic pain as a risk factor for COVID-19 ED visits and hospitalizations among NYS Medicaid beneficiaries. We also provided new evidence that individuals with chronic pain and OUD experienced an increased risk of hospitalizations. At a time in the United States when 50 million people experience chronic pain and 2.1 million have OUD, the impact of the continuing COVID-19 pandemic on these populations is concerning.¹

Our estimated 1.22 and 1.19 times increased hazards of chronic pain on COVID-19 ED visits and hospitalizations, respectively, are consistent with estimates from a recent UK Biobank study.²⁰ To our knowledge, our study was the first to stratify the risk of chronic pain on COVID-19 complications by OUD. The increased risk of chronic pain and OUD, together, on COVID-19 hospitalizations (HR of 1.25) highlights the vulnerability of patients with both conditions. Our study estimated the effect of chronic pain on COVID complications, after adjusting for demographics, baseline resource utilization, and comorbidities that could confound this relationship. In these identified associations, chronic pain could be a proxy for unmeasured risk factors or undiagnosed and/or untreated disease, such as comorbid neuropsychiatric disorders and poorer general health.^21,42–44 As such, these results suggest the need for more thorough screening and exploration of underlying health conditions among individuals presenting to clinical settings with chronic pain.

Patients with chronic pain overall who were hospitalized with COVID-19 did not experience increased odds of ventilation or acute renal failure. When stratified by OUD status, patients with chronic pain and OUD had 1.45 times increased odds of acute renal failure compared with those with neither condition. This could be because chronic opioid use and nonfatal overdose result in increased renal stress associated with greater toxicity, dehydration, and urinary retention.⁴⁵ Similarly, individuals with chronic pain and OUD experienced greater odds of other hepatic and cardiac conditions, such as hepatic failure, acute carditis, and cardiac arrest. Patients with OUD might be more likely to use other substances, such as alcohol and cocaine, which are not well captured in administrative claims data but might increase the odds of hepatic and heart-related complications over and above the health risks posed by chronic pain alone.⁴⁵ By contrast, chronic pain overall or stratified by OUD status was not an independent risk factor for sepsis or respiratory complications.⁴⁶

In our secondary exploratory analyses of COVID-19 complications among patients with chronic pain and OUD stratified by receipt of MOUD, our findings suggest that the real risk is within those with untreated OUD. Although chronic pain and OUD are complex comorbid conditions to manage, MOUD can minimize the potential toxicity and organ damage caused by otherwise unsafe or unregulated opioid consumption.⁴⁷ Our preliminary results highlight important downstream effects and benefits of MOUD, not only for OUD-related outcomes but also for ensuing complications that this population experiences. Further research is warranted to elucidate the underlying mechanisms and provide a comprehensive explanation for these observed exploratory findings.

Finally, we did not identify associations between chronic pain and hospital readmissions within 3 months after initial COVID-19 hospitalization. Frequency statistics of admitting diagnoses suggested that potential long–COVID-19 sequelae (eg, shortness of breath and chest pain), as well as drug and alcohol overdose and toxicity, drove readmissions among those initially hospitalized for COVID-19, highlighting an important phenomenon in the wake of the dual public health crises.³⁹

Overall, our findings suggest that chronic pain increases the risk of attending the ED and being admitted to the hospital for COVID-19. However, overall, it does not affect the risk of experiencing most subsequent complications after hospitalization. This conclusion is consistent with previous UK Biobank findings.⁴ History of OUD could increase the effect of chronic pain on COVID-19 hospitalizations, as well as the risk of chronic pain on hepatic and cardiac conditions.

Limitations

This study has some limitations, including its focus on adults with NYS Medicaid coverage, which might not be generalizable to those with commercial or Medicare insurance, those who are uninsured, or those residing in other geographic regions. Additionally, the study relied on administrative diagnosis codes for chronic pain and OUD, which are subject to provider coding errors. Analyses allowed patients to have enrollment gaps of any length to impose the fewest assumptions, as hospitals were unlikely to disenroll Medicaid beneficiaries from coverage while they experienced COVID-19 health issues.⁴⁸ We did not evaluate COVID-19 diagnoses outside of ED visits or hospitalizations as a primary outcome given diagnostic biases and testing limitations in 2020,⁴⁹ nor did we evaluate COVID-19 deaths given inherent limitations in claims data, which do not always contain reliable information on reasons for the end of enrollment, including mortality information. Additionally, although the present study provides quantitative estimates of the risks associated with chronic pain, it provides insights into an earlier COVID-19 period, with a different virus variant, and did not account for subsequent clinical and public health interventions, such as vaccines (which were not readily available during the study window), though these interventions could impact the magnitude of our estimates. Evaluating the mediating and moderating effects on COVID-19 outcomes of pain treatments was beyond the scope of this study. Our adjusted results might still incorporate unmeasured confounding. Nevertheless, whether the effect of chronic pain is direct or mediated through underlying poor health status, the results highlight the need for careful clinical and health systems attention to risks for COVID-19. Patients with chronic pain could also have other important patient characteristics, including poorer general health and mental health complications, which might increase the individual’s vulnerability to COVID-19 complications.²¹

Conclusions

This study demonstrated that adults with chronic pain experienced increased risk of ED visits and hospitalizations for COVID-19. Patients with chronic pain and comorbid OUD experienced greater risk of COVID-19 hospitalizations, as well as increased odds of acute renal failure. The risks posed by chronic pain, and exacerbated by OUD, highlight an increased vulnerability to colliding public health emergencies affecting a large segment of US adults. Results suggest a need to design targeted public health and clinical interventions for these high-risk groups.

Supplementary material

Supplementary material is available at Pain Medicine online.

Funding

This study was supported by the National Institute on Drug Abuse (NIDA) of the National Institutes of Health under award number R01DA045872-01A1S2; the National Institute on Drug Abuse (NIDA) of the National Institutes of Health under award number 5T32DA031099-10.

Conflicts of interest: The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the New York State Department of Health. Examples of analysis performed within this article are only examples. They should not be used in real-world analytic products. N.K. is involved in ongoing opioid litigation. A.P. has no financial disclosures. K.W.M. has no financial disclosures. K.T. has no financial disclosures. N.K. has no financial disclosures; N.K. is involved in ongoing opioid litigation. V.J. has no financial disclosures. D.S.H. acknowledges support from Syneos Health for unrelated research on the risk for prescription opioid misuse, abuse, and addiction among patients with chronic pain. C.N. has no financial disclosures. Z.L.M. has no financial disclosures. L.V.D. has no financial disclosures. J.R.P. has no financial disclosures. T.N.T. has no financial disclosures. S.C. has no financial disclosures. S.S.M. has no financial disclosures. M.C. has no financial disclosures.

The study protocol was reviewed and approved by the New York University Langone Health Institutional Review Board (ID: i18-01221).

Data availability

The data that support the findings of this study are available from the New York State Department of Health, but restrictions apply to the availability of these data, which were used under license for the present study and so are not publicly available.