Abstract
Background
Opioid use is a topic of growing concern among patients with non-alcoholic fatty liver disease (NAFLD). Given safety concerns of opioids, proactively identifying subgroups of patients with an increased probability of opioid use may encourage practitioners to recommend alternative therapies for pain, thus reducing the likelihood of opioid misuse. This work assessed the prevalence and patient characteristics associated with opioid use in a real-world cohort of patients with NAFLD.
Methods
TARGET-NASH, an observational study of participants at 55 academic and community sites in the United States, includes patients with NAFLD defined by pragmatic case definitions. Opioid use was defined as any documented opioid prescriptions in the year prior to enrollment. The association between patient characteristics and the odds of opioid use were modeled with stepwise multivariable logistic regression and tree ensemble methods (CART/Boosted Tree).
Results
The cohort included 3,474 adult patients with NAFLD including 18.0% with documented opioid use. Variables associated with opioid use included presence of cirrhosis (OR 1.51, 95% CI 1.16–1.98), BMI ≥32 kg/m2 (OR 1.29, 95% CI 1.05–1.59), depression (OR 1.87, 95% CI 1.50–2.33) and anxiety (OR 1.59, 95% CI 1.27–1.98). In the boosted tree analysis, history of back pain, depression, and fibromyalgia had the greatest relative importance in predicting opioid use.
Conclusion
Prescription opioids were used in nearly one of five patients with NAFLD. Given the safety concerns of opioids in patients with NAFLD, alternative therapies including low dose acetaminophen and non-pharmacologic treatments should be considered for these patients.
Keywords: fatty liver, analgesia, narcotics, opiates
Introduction
More than 50 million adults in the United States had painful chronic health conditions in 2016 [1]. Prescribing rates of opioids, commonly used in the management of patients with chronic pain, have increased in the United States since 2006 with more than 70,000 opioid related overdose deaths in recent years [2]. In 2016, to improve patient safety with pain management, the Centers for Disease Control and Prevention (CDC) published guidelines providing recommendations for alternative treatment options [3].
Non-alcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide [4], yet little is known regarding the prevalence of pain and associated opioid use in this population. NAFLD is commonly associated with both obesity and metabolic syndrome due to biomechanical changes which increase the mechanical load on the joints and spine, inflammatory mediators which may alter pain modulation, and poor sleep habits [5–7]. In a single center cohort of patients with NAFLD, 30% experienced pain [8]. Furthermore, model for end stage liver disease (MELD) and Child Pugh score (CPS) have been positively associated with pain and pain-related disability [8–11].
Currently there is a paucity of safety data, societal misconceptions, and fear of litigation associated with the treatment of chronic pain in patients with liver disease [12]. Many patients with severe liver disease receive opioid prescriptions for pain due to the increased risk of hepatic and renal toxicities associated with acetaminophen and non-steroidal anti-inflammatory drug (NSAID) use [8,9,13]. Yet, use of opioids among patients with severe liver disease is associated with an increased risk of complications including prolonged hospitalization, hospital readmission and decreased health-related quality of life [14–16].
Exercise is an alternative non pharmacological therapy for pain, which decreases hepatic steatosis among patients with NAFLD and improves mobility potentially reducing the need for opioids [17–21]. In light of the safety guidelines for opioid use from the CDC, there is promising evidence demonstrating that early physical therapy is associated with a reduction in future opioid use [22]. Therefore, it is possible that proactive identification of characteristics of patients at risk for opioid use may encourage health practitioners to consider alternative non opioid therapies that may improve long term health outcomes in these patients. This study estimated the real world prevalence of prescription opioid use among patients with NAFLD enrolled in the TARGET-NASH cohort. Additionally, this study utilized traditional modeling analyses as well as machine learning techniques to develop a patient phenotype at risk for opioid use.
Methods
Cohort
TARGET-NASH is an ongoing longitudinal cohort of adult and pediatric patients with NAFL, non-alcoholic steatohepatitis (NASH), or NAFLD-related cirrhosis managed according to local practice guidelines across academic and community gastroenterology, hepatology and endocrinology clinics in the United States. A description of the study design has been previously published [23]. Briefly, enrolled patients consent to review of their medical records for three years prior to enrollment and are followed prospectively for five years. Clinical information from the electronic medical records, including patient narratives, laboratory results, pathology reports and imaging data are extracted and uploaded into a secured database at six month intervals from the date of enrollment. Approvals from central and/or local institutional review boards are obtained prior to subject recruitment and at enrollment.
This analysis included all adult patients (≥18 years old) enrolled in TARGET NASH between August 1, 2016 and March 4th, 2019.
Case Definitions
All patients were diagnosed by their treating physician with NAFLD based on liver biopsy and/or pragmatic case definitions. After enrollment, patients were classified as having NAFL, NASH or NAFLD related cirrhosis. NAFL was defined as the presence of hepatic steatosis not attributable to other causes (e.g. heavy alcohol consumption) and without evidence of biochemical and/or histological inflammation. NASH was defined as the presence of steatohepatitis on biopsy or both steatosis and elevated alanine aminotransferase levels in a patient with obesity, type 2 diabetes, dyslipidemia or metabolic syndrome. Cirrhosis was based on clinical determination (i.e. clinical decompensation events, labs, imaging) or liver biopsy (Table S1) [23].
Medication Use
Patients with documentation in the medical record of any use of the following medications codeine, buprenorphine, butorphanol, hydrocodone, dihydrocodeine, oxycodone, fentanyl, hydromorphone, methadone, morphine, oxymorphone, tramadol, pethidine, and tapentadol, within one year up to or at the time of enrollment were considered “opioid users”. We classified these opioids into abuse potential based on scheduling categories set by the United States Drug Enforcement Administration [24]. Information on the duration of prescription opioid use was not available.
Clinical Characteristics
Medication use, medical history, and procedure and laboratory values were ascertained prior to or at enrollment. Psychiatric and medical comorbidities were determined by review of medical records for keywords within clinical notes or relevant medications. For patients with cirrhosis, CPS and MELD were calculated based on labs and clinical assessment at the time of enrollment [25,26].
Statistical Analysis
The proportion of patients with documented opioid and/or non-opioid analgesic use was calculated and stratified by severity of liver disease. The distribution of opioid use by demographic characteristics, patient comorbidities, reported symptoms (abdominal pain, back pain, headache), severity of liver disease and decompensation events was also calculated. A chi-squared test was used to compare the distribution of categorical variables and an analysis of variance (ANOVA) was used to compare the distribution of continuous variables among opioid users versus non-users.
Stepwise logistic regression was used to estimate the association between patient characteristics and the odds of receipt of opioids among all patients with NAFLD and within the strata of NAFLD cirrhosis using variables determined a priori as well as those identified via univariate analyses (significance level for entry=0.25, significance level to remain in the model=0.1). Collinear variables were removed, and a forward, stepwise regression model was fit. These analyses were performed using SAS version 9.4 (Cary, NC).
Classification and Regression Tree Analysis
The predicted probability of opioid use at enrollment was estimated using a classification and regression tree analysis (CART) among all patients with NAFLD. A CART analysis is a data driven, non-parametric method of estimating the association between independent variables (clinical characteristics) and the predicted probability of a given outcome (opioid use). In a CART analysis, a decision tree is created using a series of dichotomous splits of patient variables that have a similar probability of opioid use [27]. At each terminal split, the algorithm predicts the outcome by classifying the response that creates the greatest difference in the prevalence of the outcome across the splitting category [28].
CART analysis has several advantages over traditional modeling techniques. First, it allows for multiple comparisons and thus can be used to assess outcomes with many possible predictor variables. Second, it does not make assumptions about the distribution or normality of the modeled outcome and it avoids the need to inform the model with transformations of predictor variables. Third, it provides modeling of the complex interactions between predictor variables, resulting in an estimated probability of use among subgroups of patients with similar characteristics, which may be more clinically applicable.
Cross validation was performed to assess the stability of the estimate. The validation data set was created using 75% of the data as the training set and 25% as the validation set with equal balance of opioid users in each group. The misclassification rate and area under the receiver operating curve (AUROC) were calculated for the final model.
In addition to the CART analysis, boosted trees, which applied the classification algorithm over multiple iterations of classification trees, were used to estimate the variable importance of predictors of opioid use while decreasing the prediction error of the algorithm.
The following variables were included in the tree based analysis: age, sex, race, ethnicity, body mass index, recent alcohol use, Alcohol Use Disorders Identification Test (AUDIT) score, history of diabetes mellitus, depression, anxiety, osteoarthritis, rheumatologic diagnosis, dyslipidemia, gastroesophageal reflux disease (GERD), fibromyalgia, nephrolithiasis, headache, back pain, abdominal pain, end stage renal disease, bariatric surgery, and any use of anxiolytics, antipsychotics, and antidepressants. All tree based analyses were performed in JMP SAS (Cary, NC).
Results
Cohort Characteristics
The cohort included 3,474 adult patients with NAFLD including 1,054 (30.3%) with NAFL, 1,303 (37.5%) with NASH, and 1,117 (32.2%) with NAFLD cirrhosis. At the time of enrollment, patients’ mean age (+/− SD) was 56.9 ± 13.0 years, 58.9% were female, 74.5% white, 5.3% black, 12.3% Hispanic, and 83.9% non-Hispanic. The mean BMI was 33.3. ± 7.6 kg/m2, 50.5% of patients had diabetes, and 61.7% had dyslipidemia (Table 1).
Table 1.
Descriptive characteristics of patients with NAFLD by opioid status
| Summary | Opioid Use (N=624) | Other Analgesic Use (N=1324) | No Analgesic Use (N=1526) | All Participants (N=3474) | p-values |
|---|---|---|---|---|---|
| Disease Severity, n (%) | <0.0001 | ||||
| NAFLD Cirrhosis | 267 (42.8) | 419 (31.6) | 431 (28.2) | 1117 (32.2) | |
| NASH | 231 (37.0) | 553 (41.8) | 519 (34.0) | 1303 (37.5) | |
| NAFL | 126 (20.2) | 352 (26.6) | 576 (37.7) | 1054 (30.3) | |
| Age at Study Entry (years)1 | <0.0001 | ||||
| Median (n) | 58.0 (624) | 61.0 (1324) | 56.0 (1525) | 59.0 (3473) | |
| Mean (SD) | 57.2 (11.04) | 59.4 (12.21) | 54.7 (13.89) | 56.9 (12.96) | |
| Min - Max | 19.0 – 83.0 | 18.0 – 91.0 | 18.0 – 87.0 | 18.0 – 91.0 | |
| Age at Study Entry by Category, n (%) | <0.0001 | ||||
| <50 | 139 (22.3) | 233 (17.6) | 507 (33.2) | 879 (25.3) | |
| >= 50 | 485 (77.7) | 1091 (82.4) | 1018 (66.7) | 2594 (74.7) | |
| Not Available | 0 | 0 | 1 (0.07) | 1 (0.03 | |
| Gender, n (%) | <0.0001 | ||||
| Female | 409 (65.5) | 786 (59.4) | 852 (55.8) | 2047 (58.9) | |
| Male | 215 (34.5) | 538 (40.6) | 674 (44.2) | 1427 (41.1) | |
| Not Available | 0 | 0 | 0 | 0 | |
| Race, n (%) | <0.000120 | ||||
| White | 515 (82.5) | 1057 (79.8) | 1016 (66.6) | 2588 (74.5) | |
| Black or African American | 43 (6.9) | 81 (6.1) | 59 (3.9) | 183 (5.3) | |
| American Indian or Alaska Native | 1 (0.2) | 6 (0.5) | 8 (0.5) | 15 (0.4) | |
| Asian | 14 (2.2) | 77 (5.8) | 335 (22.0) | 426 (12.3) | |
| Native Hawaiian or other Pacific Islander | 0 (0.0) | 2 (0.2) | 3 (0.2) | 5 (0.1) | |
| Other | 22 (3.5) | 44 (3.3) | 53 (3.5) | 119 (3.4) | |
| Not Reported | 29 (4.6) | 57 (4.3) | 52 (3.4) | 138 (4.0) | |
| Not Available | 0 | 0 | 0 | 0 | |
| Ethnicity, n (%) | 0.1749 | ||||
| Hispanic or Latino | 71 (11.4) | 148 (11.2) | 207 (13.6) | 426 (12.3) | |
| Not Hispanic or Latino | 531 (85.1) | 1124 (84.9) | 1260 (82.6) | 2915 (83.9) | |
| Not Reported | 14 (2.2) | 38 (2.9) | 45 (2.9) | 97 (2.8) | |
| Other | 8 (1.3) | 13 (1.0) | 13 (0.9) | 34 (1.0) | |
| Not Available | 0 | 1 (.08) | 1 (0.07) | 2 (0.06) | |
| BMI (kg/m2) at Enrollment | <0.0001 | ||||
| Median (n) | 34.0 (616) | 33.0 (1305) | 31.0 (1473) | 32.0 (3394) | |
| Mean (SD) | 35.4 (7.9) | 33.8 (7.3) | 31.9 (7.5) | 33.3 (7.6) | |
| Min - Max | 17 – 76 | 16 – 80 | 17 – 103 | 16 – 103 | |
| BMI (kg/m2) Category, n (%) | <0.0001 | ||||
| < 32 | 202 (32.3) | 538 (40.6) | 789 (51.7) | 1529 (44.0) | |
| >= 32 | 414 (66.3) | 767 (57.9) | 684 (44.8) | 1865 (53.7) | |
| Not Available | 8 (1.3) | 19 (14.4) | 53 (3.5) | 80 (23.0) | |
| Alcohol Usage | 0.2298 | ||||
| Never | 242 (38.8) | 431 (32.6) | 479 (31.4) | 1152 (33.2) | |
| Current | 111 (17.8) | 326 (24.6) | 320 (21.0) | 757 (21.8) | |
| Former | 43 (6.9) | 68 (5.1) | 86 (5.6) | 197 (5.7) | |
| Unknown | 48 (7.7) | 98 (7.4) | 126 (8.3) | 272 (7.8) | |
| Not Available | 180 (28.8) | 401 (30.3) | 515 (33.7) | 1096 (31.5) | |
| Smoking Status | <0.0001 | ||||
| Never | 303 (48.6) | 712 (53.8) | 883 (57.9) | 1898 (54.6) | |
| Current | 68 (10.9) | 92 (6.9) | 100 (6.6) | 260 (7.5) | |
| Former | 196 (31.4) | 388 (29.3) | 294 (22.2) | 878 (25.3) | |
| Not Available | 57 (9.1) | 132 (99.7) | 249 (18.8) | 438 (12.6) | |
| Bariatric Surgery, n (%)2 | <0.0001 | ||||
| No | 595 (95.4) | 1287 (97.2) | 1502 (98.4) | 3384 (97.4) | |
| Yes | 29 (4.6) | 37 (2.8) | 24 (1.6) | 90 (2.6) | |
| Renal Failure, n (%)3 | 0.0219 | ||||
| No | 611 (97.9) | 1311 (99.0) | 1514 (99.2) | 3436 (98.9) | |
| Yes | 13 (2.1) | 13 (1.0) | 12 (0.8) | 38 (1.1) | |
| GERD, n (%)4 | <0.0001 | ||||
| No | 180 (28.8) | 519 (39.2) | 884 (57.9) | 1583 (45.6) | |
| Yes | 444 (71.2) | 805 (60.8) | 642 (42.1) | 1891 (54.4) | |
| Dyslipidemia, n (%)5 | <0.0001 | ||||
| No | 202 (32.4) | 348 (26.3) | 780 (51.1) | 1330 (38.3) | |
| Yes | 422 (67.6) | 976 (73.7) | 746 (48.9) | 2144 (61.7) | |
| Diabetes, n (%)6 | <0.0001 | ||||
| No | 237 (38.0) | 542 (40.9) | 941 (61.7) | 1720 (49.5) | |
| Yes | 387 (62.0) | 782 (59.1) | 585 (38.3) | 1754 (50.5) | |
| Autoimmune/Rheumatologic, n (%)7 | <0.0001 | ||||
| No | 542 (86.9) | 1231 (93.0) | 1464 (95.9) | 3237 (93.2) | |
| Yes | 82 (13.1) | 93 (7.0) | 62 (4.1) | 237 (6.8) | |
| Fibromyalgia, n (%)8 | <0.0001 | ||||
| No | 530 (84.9) | 1246 (94.1) | 1497 (98.1) | 3273 (94.2) | |
| Yes | 94 (15.1) | 78 (5.9) | 29 (1.9) | 201 (5.8) | |
| Chronic Fatigue, n (%)9 | 0.2791 | ||||
| No | 621 (99.5) | 1320 (99.7) | 1523 (99.8) | 3464 (99.7) | |
| Yes | 3 (0.5) | 4 (0.3) | 3 (0.2) | 10 (0.3) | |
| Chronic Fatigue/Fibromyalgia, n (%)10 | <0.0001 | ||||
| n | 624 | 1324 | 1526 | 3474 | |
| No | 528 (84.6) | 1243 (93.9) | 1495 (98.0) | 3266 (94.0) | |
| Yes | 96 (15.4) | 81 (6.1) | 31 (2.0) | 208 (6.0) | |
| Osteoarthritis, n (%)11 | <0.0001 | ||||
| No | 509 (81.6) | 1148 (86.7) | 1462 (95.8) | 3119 (89.8) | |
| Yes | 115 (18.4) | 176 (13.3) | 64 (4.2) | 355 (10.2) | |
| Nephrolithiasis, n (%)12 | <0.0001 | ||||
| No | 537 (86.1) | 1219 (92.1) | 1428 (93.6) | 3184 (91.7) | |
| Yes | 87 (13.9) | 105 (7.9) | 98 (6.4) | 290 (8.3) | |
| Any Mental Health Diagnosis, n (%)13 | <0.0001 | ||||
| No | 157 (25.2) | 607 (45.8) | 1012 (66.3) | 1776 (51.1) | |
| Yes | 467 (74.8) | 717 (54.2) | 514 (33.7) | 1698 (48.9) | |
| Any Depression Diagnosis, n (%)14 | <0.0001 | ||||
| No | 233 (37.3) | 762 (57.6) | 1146 (75.1) | 2141 (61.6) | |
| Yes | 391 (62.7) | 562 (42.4) | 380 (24.9) | 1333 (38.4) | |
| Any Anxiety Diagnosis, n (%)15 | <0.0001 | ||||
| No | 341 (54.6) | 946 (71.) | 1262 (82.7) | 2549 (73.4) | |
| Yes | 283 (45.4) | 378 (28.5) | 264 (17.3%) | 925 (26.6) | |
| Any Other Mental Health Diagnosis, n (%)16 | <0.0001 | ||||
| No | 423 (67.8) | 1036 (78.2) | 1347 (88.3) | 2806 (80.8) | |
| Yes | 201 (32.2) | 288 (21.8) | 179 (11.7) | 668 (19.2) | |
| Headache, n (%)17 | <0.0001 | ||||
| No | 556 (89.1) | 1246 (94.1) | 1481 (97.1) | 3283 (94.5) | |
| Yes | 68 (10.9) | 78 (5.9) | 45 (2.9) | 191 (5.5) | |
| Back Pain, n (%)18 | <0.0001 | ||||
| No | 453 (72.6) | 1132 (85.5) | 1443 (94.6) | 3028 (87.2) | |
| Yes | 171 (27.4) | 192 (14.5) | 83 (5.4) | 446 (12.8) | |
| Abdominal Pain, n (%)19 | <0.0001 | ||||
| No | 391 (62.7) | 995 (75.2) | 1258 (82.4) | 2644 (76.1) | |
| Yes | 233 (37.3) | 329 (24.8) | 268 (17.6) | 830 (23.9) | |
| Acetominophen Use, n (%) | |||||
| No | 362 (58.0) | 1164 (87.9) | 1526 (100.0) | 3052 (87.9) | |
| Yes | 262 (42.0) | 160 (12.1) | 0 (0.0) | 422 (12.1) | |
| NSAID Use, n (%) | |||||
| No | 263 (42.1) | 229 (17.3) | 1526 (100.0) | 2018 (58.1) | |
| Yes | 361 (57.9) | 1095 (82.7) | 0 (0.0) | 1456 (41.9) |
Opioid use reported within one year prior to or at enrollment. Other medication use, medical history, procedures and laboratory values reported on or prior to enrollment.
Age calculated based on year of consent minus birth year.
From procedures coded as ‘Gastric banding’ or ‘Gastric bypass’.
Medical history coded as ‘Renal failure’.
Medical history coded as ‘Gastrooesophageal reflux disease’ or any use of a proton pump inhibitor.
Medical history coded terms indicative of dyslipidaemia or any use of medications indicative of dyslipidemia.
Medical history coded to diabetes (excluding prediabetes or insulin resistant diabetes), medications used to treat diabetes or HbA1c>6.5.
Medical history coded to terms indicative of autoimmune/rheumatologic conditions.
Medical history coded to ‘Fibromyalgia’.
Medical history coded to ‘Chronic fatigue syndrome’.
Medical history coded to ‘Chronic fatigue syndrome’ or ‘Fibromyalgia’.
Medical history coded to ‘Osteoarthritis’.
Medical history coded to ‘Nephrolithiasis’.
Medical history coded to ‘Psychiatric disorders’ or medications used to treat mental health disorders.
Medical history indicative of depression or medications used to treat depression.
Medical history indicative of anxiety or medications used to treat anxiety.
Medical history coded to ‘Psychiatric disorders’ that are not indicative of depression or anxiety, or medications used to treat schizophrenia.
Medical history coded terms indicative of headache.
Medical history coded terms indicative of back pain.
Medical history coded terms indicative of abdominal pain.
Race grouped as: white, black or African American, or other.
Characteristics of Patients by Opioid Use
Among all patients with NAFLD, 56.1% of patients were users of analgesics and 18.0% had documentation of opioid use within a year prior to or at enrollment. Of all opioid prescriptions, 12.2% were DEA schedule II, 1.4% were schedule III, and 7.7% were schedule IV. Opioid use was more common in patients with more advanced liver disease (12.0% NAFL, 17.7% NASH, 23.9% NAFLD cirrhosis, p<0.001). Opioid use was more often documented among females (20.0%) compared to males (15.1%) (p<0.001) and opioid users had higher mean BMI (35.4 vs 33.8 kg/m2, p<0.001) compared to users of non-opioid analgesics. The prevalence of opioid use was highest among black race (23.5%) compared to white (19.9%) and other races (6.5%) (p<0.001) (Table 1).
Non-opioid analgesic use was documented in 38.1% of patients with NAFLD. Ninety percent of non-opioid analgesic users were users of NSAIDs and 12.1% were users of acetaminophen. The proportion of non-opioid analgesic use was highest among patients with NASH (42.4%) followed by NAFLD cirrhosis (37.5%) (p<0.001) (Table 2). Patients who were users of non-opioid analgesics were 27.5% and 47.1% less likely to have a history of a mental health diagnosis or back pain respectively compared to opioid users (Table 1).
Table 2.
Frequency of Analgesic Use Overall and by NAFLD Category
| Overall (N=3474) | NAFLD Cirrhosis (N=1117) | NASH (N=1303) | NAFL (N=1054) | p-values | |
|---|---|---|---|---|---|
| N (%) | N (%) | N (%) | N (%) | ||
| Opioid Use* | 624 (18.0) | 267 (23.9) | 231 (17.7) | 126 (12.0) | <0.001 |
| Non-opioid analgesic use† | 1324 (38.1) | 419 (37.5) | 553 (42.4) | 352 (33.4) | <0.001 |
| No analgesic use | 1526 (43.9) | 431 (38.6) | 519 (39.8) | 576 (54.6) | <0.001 |
Multivariable Model of Opioid Use in NAFLD
In the multivariable adjusted logistic regression model, variables associated with opioid use included NAFLD cirrhosis vs NAFL (OR 1.51, 95% CI 1.16–1.98), BMI ≥32 kg/m2 (OR 1.29, 95% CI 1.05–1.59), depression (OR 1.87, 95% CI 1.50–2.33), anxiety (OR 1.59, 95% CI 1.27–1.98), osteoarthritis (OR 1.42, 95% CI 1.07–1.89), rheumatologic disease (OR 1.95, 95% CI 1.40–2.70), GERD (OR 1.28, 95% CI 1.03–1.60), fibromyalgia (OR 2.17, 95% CI 1.55–3.04), nephrolithiasis (OR 1.69, 95% CI 1.25–2.29), back pain (OR 2.25, 95% CI 1.76–2.87) and abdominal pain (OR 1.58, 95% CI 1.28–1.95) (Table 3).
Table 3.
Odds of Opioid Use among patient with NAFLD by patient characteristics*
| OR | 95% CI | ||
|---|---|---|---|
| NAFLD Severity | NAFL | reference | reference |
| NASH | 1.15 | 0.89–1.50 | |
| NAFLD cirrhosis | 1.51 | 1.16–1.98 | |
| Age ≥59 years (vs <59) | 0.86 | 0.70–1.06 | |
| BMI ≥32 kg/m2 (vs <32) | 1.29 | 1.05–1.59 | |
| Diabetes mellitus | 1.21 | 0.98–1.49 | |
| Depression | 1.87 | 1.50–2.33 | |
| Anxiety | 1.59 | 1.27–1.98 | |
| Osteoarthritis | 1.42 | 1.07–1.89 | |
| Rheumatologic diagnosis | 1.95 | 1.40–2.70 | |
| Fibromyalgia | 2.17 | 1.55–3.04 | |
| GERD | 1.28 | 1.03–1.60 | |
| Nephrolithiasis | 1.69 | 1.25–2.29 | |
| Back pain | 2.25 | 1.76–2.87 | |
| Abdominal pain | 1.58 | 1.28–1.95 |
Variables included in the step-wise model: disease severity, age(years), sex, race (white vs non-white), BMI, type 2 diabetes, osteoarthritis, depression, autoimmune/rheumatologic disease, hyperlipidemia, GERD, fibromyalgia, nephrolithiasis, anxiety, headache, back pain and abdominal pain: significance level to remain in the model=0.1
Characteristics of Opioid Use in NAFLD Cirrhosis
The prevalence of splenomegaly (17.2% vs 10.1%, p=0.003), hepatic encephalopathy (34.8% vs 23.4%, p<0.001), and ascites (43.1% vs 34.4%, p=0.010) was significantly higher among opioid users compared to patients without documentation of opioid use. The prevalence of hepatocellular carcinoma was not significantly different among patients with cirrhosis who had documentation of opioid use compared to non-users (3.4% vs 2.0%, p=0.214). Among opioid users and non-users with NAFLD cirrhosis, there was no significant difference in mean MELD score (10.3 vs 10.1, p=0.501) or CPS (8.0 vs 7.7, p=0.195) (Table 4).
Table 4.
Characteristics of Opioid Users vs. Non-Users with NAFLD Cirrhosis*
| Opioid Use (n=267) | No Opioid Use (n=850) | p-value | |
|---|---|---|---|
| N (%) | N (%) | N (%) | |
| MELD (Mean) (SD) | 10.3 (4.5) | 10.1 (4.0) | 0.5008 |
| Child-Pugh Score (Mean) (SD) | 8.0 (2.0) | 7.7 (1.9) | 0.1950 |
| Hepatic Encephalopathy (%) | 93 (34.8) | 199 (23.4) | 0.0003 |
| Ascites | 115 (43.1) | 292 (34.4) | 0.0104 |
| Variceal Bleeding | 19 (7.1) | 50 (5.9) | 0.4718 |
| Hepatocellular Carcinoma | 9 (3.4) | 17 (2.0) | 0.2138 |
| Splenomegaly | 46 (17.2) | 86 (10.1) | 0.0025 |
MELD: model for end stage liver disease
Multivariable Model of Opioid Use in NAFLD Cirrhosis
In the multivariate analysis of opioid use among patients with NAFLD cirrhosis, age, sex, BMI, osteoarthritis, rheumatologic diagnosis, depression, MELD, ascites, HCC and hepatic encephalopathy were considered in our initial model (Table 5). In our final model, age >62 years (OR:0.70, 95% CI 0.51–0.96), depression (OR 2.29, 95% CI 1.66–3.18), rheumatologic disease (OR 2.53, 95% CI 1.48–4.30), and osteoarthritis (OR 2.30, 95% CI 1.53–3.46) were associated with opioid use (Table 5).
Table 5.
Odds of Opioid Use among patient with NAFLD Cirrhosis by patient characteristics*
| OR | 95% CI | |
|---|---|---|
| Age ≥62 years (vs <62) | 0.70 | 0.51–0.96 |
| BMI ≥34 kg/m2 (vs <34) | 1.24 | 0.90–1.71 |
| Depression | 2.29 | 1.66–3.18 |
| Osteoarthritis | 2.30 | 1.53–3.46 |
| Rheumatologic diagnosis | 2.53 | 1.48–4.30 |
| Ascites | 1.26 | 0.89–1.78 |
| Hepatic encephalopathy | 1.39 | 0.97–2.00 |
Variables included in the step-wise model: age (years), sex, BMI, osteoarthritis, rheumatologic diagnosis, depression, MELD, ascites, HCC, and hepatic encephalopathy; significance level to remain in the model=0.1
Classification and Boosted Tree Analysis
In the boosted tree analysis, a history of fibromyalgia, depression, or back pain in the medical record had the greatest variable importance for predicting opioid use over 100 trees at a sampling rate of 0.75. The AUROC was 0.76 for the learning sample and 0.73 for the validation sample. The misclassification rate was 0.18 for both the training and the validation sample (Figure 1).
Figure 1. Variable importance predictive of opioid use among patients with NAFLD using boosted trees.
AUDIT: Alcohol Use Disorders Identification Test (AUDIT); GERD: Gastroesophageal Reflux Disease (GERD); Gini improvement measure of relative importance estimates the difference in the prevalence of the outcome in the child nodes after the classification split [27]; AROClearning: 0.7602; AROC validation:0.7321; Misclassification Rate learning:0.1796; Misclassification Rate Validation:0.1797. Any opioid use at enrollment or one year prior; 100 layers, Multiple splits per tree 3, Sampling rate 0.60 for rows and 0.40 columns;
**medical history of or drug use on or prior to enrollment
*medical history on or prior to enrollment
Forty six percent of patients with NAFLD who had a history of depression and back pain were opioid users. The probability of opioid use among patients with a combined history of depression and fibromyalgia without back pain was 44%. Among patients without depression, the probability of opioid use was approximately 20% for patents with anxiety, abdominal pain, or back pain. The misclassification rate in the training sample and validation sample was 0.18 (Table 6).
Table 6.
Classification and Regression Tree (CART) estimating the predicted probability of opioid use among patients with NAFL*
| Patient Phenotypes | Probability of Opioid Use |
|---|---|
| Patients with Depression | |
| Depression, Back Pain | 0.4578 |
| Depression, No Back pain, Fibromyalgia | 0.4368 |
| Depression, No Back Pain, No Fibromyalgia | 0.2348 |
| Patients without Depression | |
| No depression, Abdominal Pain | 0.2142 |
| No depression, No Abdominal Pain, Back Pain | 0.2518 |
| No depression, No Abdominal Pain, No Back Pain, Anxiety | 0.2067 |
| No depression, No Abdominal Pain, No Back Pain, No Anxiety, White or African American Race | 0.0710 |
| No depression, No Abdominal Pain, No Back Pain, No Anxiety, Other Race | 0.0210 |
Variable included in the model: age, sex, race, ethnicity, body mass index, severity of liver disease, recent alcohol use, Alcohol Use Disorders Identification Test (AUDIT) score, history of diabetes mellitus, depression, anxiety, osteoarthritis, rheumatologic diagnosis, dyslipidemia, gastroesophageal reflux disease (GERD), fibromyalgia, nephrolithiasis, headache, back pain, abdominal pain, end stage renal disease, bariatric surgery, and any use of anxiolytics, antipsychotics, and antidepressants. AUROC test:0.7257, AUROC validation:0.7172, Misclassification Rate test:0.1796, Misclassification Rate test:0.1797. Depression: medical history or medication use for depression. Anxiety: medical history or medication use for anxiety
Discussion
This cross-sectional study estimates real-world opioid use among patients with NAFLD. Nearly one-fifth of those with NAFLD and one-quarter of patients with NAFLD cirrhosis reported recent or ongoing opioid use. Opioid use was twice as common among those with NAFLD cirrhosis compared with NAFL and was associated with higher BMI, painful comorbidities and psychiatric disease.
In the CART analysis, which allows for the identification of distinct sub groups of patients with NALFD who are likely to use opioids, both depression and back pain were associated with opioid use across patient subgroups. Approximately half of patients who had both a history of depression as well as a history of fibromyalgia were opioid users. The probability of opioid use among patients with back pain and depression was twice that of patients with back pain who were not depressed.
The results of both analyses, CART and logistic regression, suggested that depression and back pain were significantly associated with opioid use in patients with NAFLD. A logistic regression analysis estimated the average effect for the association between independent variables and the odds of an outcome, while a CART analysis can define clinical subgroups of patient’s that are associated with an increased probability of opioid use. The results of this classification analysis suggest that patients with NAFLD who also suffer from back pain, fibromyalgia, and/or depression might be at particularly high risk of future opioid use and may benefit from proactive alternative therapies to address pain, such as scheduled low-dose acetaminophen or referral to physical therapy and/or a psychologist where appropriate.
Among patients with cirrhosis, the proportion of patients who were users of opioids in our cohort is slightly lower than prior studies reporting a 25–47% prevalence of opioid use among patients with cirrhosis [8,9,13,29]. This may be due to differences in how opioid use was ascertained or a reflection of our “real world” population of patients with NAFLD who were receiving care from both academic and community practices. These clinical practices may have less severe liver disease and fewer comorbidities than cohorts drawn exclusively from tertiary referral centers. In addition, given that our cohort was enrolled recently, this may be a reflection of the national trend towards reduced opioid prescribing [30].
This work supports previously findings in the literature which have suggested that opioid use was more common as liver disease progressed, with the highest reported prevalence of use in patients with NAFLD cirrhosis [8]. This could reflect an appropriate avoidance of NSAIDS given their risk of precipitating acute kidney injury and gastrointestinal bleeding or misinformation about the harms of acetaminophen use in the setting of cirrhosis. However, the higher reported use of non-opioid analgesics, including acetaminophen and NSAIDs, among patients with NASH and cirrhosis compared to NAFL suggests that this observation may be due to an increased prevalence of pain in those with higher BMI or lower levels of physical activity, both of which promote liver disease progression.
Another potential explanation for opioid use among patients with NAFLD is the association between metabolic syndrome and painful complications such as osteoarthritis due to increased mechanical load or diabetic neuropathy. Even after adjusting for BMI, osteoarthritis was significantly associated with opioid use. A lower pain threshold may also be attributable to higher levels of inflammatory cytokines seen in NAFLD [9,31]. It is possible that mood disturbances, seen more frequently in cirrhosis, and sleep disturbances, a common manifestation of hepatic encephalopathy, could contribute to opioid use even in the absence of pain [32]. In our cohort, fibromyalgia, anxiety, and depression were all associated with opioid use and hepatic encephalopathy was more commonly reported in opioid users.
Opioid use in patients with NAFLD is concerning given the potential role of opioids in sedentary behavior. Large prospective studies have demonstrated that the development of obesity is associated with a higher risk of chronic low back, neck, and shoulder pain and that this effect may be attenuated by exercise [33]. In our cohort, opioids were associated with obesity, suggesting that opioid use may predispose to the development or worsening of NAFLD.
While there are situations in which opioid prescription may be appropriate, particularly for acute pain, there are many reasons why opioids should be used cautiously in patients with NAFLD and cirrhosis [15,34]. First, opioids have limited efficacy data for the treatment of chronic non-cancer pain (>3 months) [34,35]. Additionally, the use of long-acting opioids for non-cancer pain is associated with increased risk for opioid abuse, opioid dependence, opioid overdose, and all-cause mortality compared to no opioid treatments or non-opioid analgesics [34,36]. Lastly, opioids have specifically been associated with poor outcomes in patients with cirrhosis including an increase in hepatic encephalopathy, length of stay and hospital readmission [14,37,38].
Given opioids’ limited long-term efficacy data and substantial risks, alternative options for patients with NAFLD and pain should be fully explored [35]. Non-pharmacologic options should be considered first line options in patients with NAFLD [3]. Exercise, which remains a cornerstone of treatment for NAFLD, has the dual benefits of improving hepatic steatosis in patients with NAFLD and decreasing back pain, joint pain and symptoms of fibromyalgia [17–19,39–42]. This may be facilitated through early referral to physical therapy which is associated with reduced future opioid use [22]. Other non-pharmacologic treatment options that deserve additional study in patients with NAFLD and cirrhosis include multidisciplinary rehabilitation, mindfulness-based stress reduction and cognitive behavioral therapy [34]. Lastly, treatment for coexisting depression, anxiety, substance use disorders and sleep disturbances may improve pain and reduce inappropriate opioid use. Acetaminophen (≤2 g/day) should be considered a first-line non-opioid analgesic option in cirrhosis given its safety and efficacy data [43]. Further study is needed to determine alternative analgesics or analgesic combinations that optimize physical function and limit adverse effects in patients with NAFLD and cirrhosis.
Our study has several strengths. It includes a large, real-world cohort of patients with NAFLD and, to our knowledge, represents the largest study to date examining use of opioids in this population. Many prior observational studies of patients with NAFLD have been limited to tertiary centers or large administrative databases, which may be more prone to referral bias, misclassification bias and limited generalizability [44,45]. In contrast, our cohort encompasses patients with NAFLD diagnosed by histology and clinical data, rather than billing codes, and includes patients from both community and academic settings. Therefore, our estimates of opioid use are likely to more accurately reflect patients with NAFLD receiving care in usual clinical practice.
There are a few limitations of this work. The cross-sectional design of this study does not allow one to determine the temporal relationship between patient characteristics and current opioid use. Prospective studies, including the TARGET-NASH cohort, will help to shed light on the use of opioids in patients with NAFLD and the potential for associated complications. Moreover, the definition of opioid use is based on reports of a prescription for opioids at or within a year of enrollment and not on pharmacy fills. It is possible that patients who were prescribed but never took opioids were misclassified as opioid users and those who took non-medical prescription opioids were misclassified as non-users. Our cohort included primarily White and Non-Hispanic patients who were actively receiving care in largely Gastroenterology and Hepatology practices across the United States. Patterns of opioid use may differ among other racial/ethnic groups or patients receiving care in other specialty or primary care clinics. Lastly, we were unable to differentiate between short-term and chronic opioid use.
Opioid use occurred in one of five patients with NAFLD and was more common among those who have more severe liver disease, anxiety or depression. Given the safety concerns associated with opioid treatment, use of non-pharmacologic options and non-opioid analgesics should be considered for the treatment of pain in patients with NAFLD.
Supplementary Material
Acknowledgements
TARGET-NASH is sponsored by TARGET PharmaSolutions Inc. We would like to thank the study staff, nurses, health care providers, and participants for their contribution to this study. This is a collaboration of academic and community investigators, pharmaceutical partners, and NASH patient community advocates.
Grant Support: This research was supported in part by grants from the National Institutes of Health, T32 DK007634.
ASL: has received research grants from BMS, Gilead and TARGET PharmaSolutions (to University of Michigan) and is an advisor for TARGET PharmaSolutions
Abbreviations
- AUDIT
Alcohol Use Disorders Identification Test
- AUROC
area under the receiver operating curve
- BMI
body mass index
- CART
classification and regression tree
- CPS
Child Pugh score
- DEA
Drug Enforcement Agency
- GERD
gastroesophageal reflux disease
- HCC
hepatocellular carcinoma
- NAFL
non-alcoholic fatty liver
- NAFLD
non-alcoholic fatty liver disease
- NASH
non-alcoholic steatohepatitis
- NSAIDs
nonsteroidal anti-inflammatory drugs
- MELD
model for end-stage liver disease
Footnotes
Conflicts of Interest:
AM: No conflicts related to this work
SW is an employee of TARGET PharmaSolutions
RFM: No conflicts related to this work
HT: No conflicts related to this work
JK: No conflicts related to this work
CS: is an employee of TARGET PharmaSolutions
BAN-T: No conflicts related to this work
ASB is a consultant for TARGET PharmaSolutions
Statement of Ethics
This study was approved by WCG (WIRB Copernicus Group); reference number:20161381. Clinical information from the electronic medical records, including patient narratives, laboratory results, pathology reports and imaging data are extracted and uploaded into a secured database at six month intervals from the date of enrollment. Approvals from central and/or local institutional review boards are obtained prior to subject recruitment and at enrollment.
References
- 1.Dahlhamer J, Lucas J, Zelaya C et al. Prevalence of Chronic Pain and High-Impact Chronic Pain Among Adults - United States, 2016 MMWR Morb Mortal Wkly Rep. 2018;67:1001–1006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Scholl L, Seth P, Kariisa M, Wilson N, Baldwin G. Drug and Opioid-Involved Overdose Deaths - United States, 2013–2017 MMWR Morb Mortal Wkly Rep. 2018;67:1419–1427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Dowell D, Haegerich TM, Chou R. CDC Guideline for Prescribing Opioids for Chronic Pain--United States, 2016 JAMA. 2016;315:1624–1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Perumpail BJ, Khan MA, Yoo ER, Cholankeril G, Kim D, Ahmed A. Clinical epidemiology and disease burden of nonalcoholic fatty liver disease World J Gastroenterol. 2017;23:8263–8276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Stone AA, Broderick JE. Obesity and pain are associated in the United States Obesity (Silver Spring). 2012;20:1491–1495. [DOI] [PubMed] [Google Scholar]
- 6.Smuck M, Kao MC, Brar N, Martinez-Ith A, Choi J, Tomkins-Lane CC. Does physical activity influence the relationship between low back pain and obesity? Spine J. 2014;14:209–216. [DOI] [PubMed] [Google Scholar]
- 7.Hitt HC, McMillen RC, Thornton-Neaves T, Koch K, Cosby AG. Comorbidity of obesity and pain in a general population: results from the Southern Pain Prevalence Study J Pain. 2007;8:430–436. [DOI] [PubMed] [Google Scholar]
- 8.Rogal SS, Winger D, Bielefeldt K, Szigethy E. Pain and opioid use in chronic liver disease Dig Dis Sci. 2013;58:2976–2985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Rogal SS, Bielefeldt K, Wasan AD et al. Inflammation, psychiatric symptoms, and opioid use are associated with pain and disability in patients with cirrhosis Clin Gastroenterol Hepatol. 2015;13:1009–1016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gutteling JJ, de Man RA, van der Plas SM, Schalm SW, Busschbach JJ, Darlington AS. Determinants of quality of life in chronic liver patients Aliment Pharmacol Ther. 2006;23:1629–1635. [DOI] [PubMed] [Google Scholar]
- 11.Rakoski MO, McCammon RJ, Piette JD et al. Burden of cirrhosis on older Americans and their families: analysis of the health and retirement study Hepatology. 2012;55:184–191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rakoski MO, Goyal P, Spencer-Safier M, Weissman J, Mohr G, Volk ML. Pain Management in Patients with Cirrhosis Clinical Liver Disease. 2018;11:135–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rogal SS, Beste LA, Youk A et al. Characteristics of Opioid Prescriptions to Veterans With Cirrhosis Clin Gastroenterol Hepatol. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Acharya C, Betrapally NS, Gillevet PM et al. Chronic opioid use is associated with altered gut microbiota and predicts readmissions in patients with cirrhosis Aliment Pharmacol Ther. 2017;45:319–331. [DOI] [PubMed] [Google Scholar]
- 15.Dwyer JP, Jayasekera C, Nicoll A. Analgesia for the cirrhotic patient: a literature review and recommendations J Gastroenterol Hepatol. 2014;29:1356–1360. [DOI] [PubMed] [Google Scholar]
- 16.Tapper EB, Baki J, Parikh ND, Lok AS. Frailty, Psychoactive Medications, and Cognitive Dysfunction Are Associated With Poor Patient-Reported Outcomes in Cirrhosis Hepatology. 2019;69:1676–1685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Busch AJ, Barber KA, Overend TJ, Peloso PM, Schachter CL. Exercise for treating fibromyalgia syndrome Cochrane Database Syst Rev. 2007:CD003786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fransen M, McConnell S, Hernandez-Molina G, Reichenbach S. Exercise for osteoarthritis of the hip Cochrane Database Syst Rev. 2014:CD007912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. Exercise for osteoarthritis of the knee Cochrane Database Syst Rev. 2015;1:CD004376. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hayden JA, van Tulder MW, Malmivaara A, Koes BW. Exercise therapy for treatment of non-specific low back pain Cochrane Database Syst Rev. 2005:CD000335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Smart NA, King N, McFarlane JR, Graham PL, Dieberg G. Effect of exercise training on liver function in adults who are overweight or exhibit fatty liver disease: a systematic review and meta-analysis Br J Sports Med. 2018;52:834–843. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sun E, Moshfegh J, Rishel CA, Cook CE, Goode AP, George SZ. Association of Early Physical Therapy With Long-term Opioid Use Among Opioid-Naive Patients With Musculoskeletal Pain JAMA Netw Open. 2018;1:e185909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Barritt AS, Gitlin N, Klein S et al. Design and rationale for a real-world observational cohort of patients with nonalcoholic fatty liver disease: The TARGET-NASH study Contemp Clin Trials. 2017;61:33–38. [DOI] [PubMed] [Google Scholar]
- 24.United States Drug Enforcement Administration. Drug Scheduling, Available at: https://www.dea.gov/drug-scheduling. Accessed June 28, 2020.
- 25.Wiesner R, Edwards E, Freeman R et al. Model for end-stage liver disease (MELD) and allocation of donor livers Gastroenterology. 2003;124:91–96. [DOI] [PubMed] [Google Scholar]
- 26.Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices Br J Surg. 1973;60:646–649. [DOI] [PubMed] [Google Scholar]
- 27.Lemon SC, Roy J, Clark MA, Friedmann PD, Rakowski W. Classification and regression tree analysis in public health: methodological review and comparison with logistic regression Ann Behav Med. 2003;26:172–181. [DOI] [PubMed] [Google Scholar]
- 28.Strobl C, Malley J, Tutz G. An introduction to recursive partitioning: rationale, application, and characteristics of classification and regression trees, bagging, and random forests Psychol Methods. 2009;14:323–348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Konerman MA, Rogers M, Kenney B et al. Opioid and benzodiazepine prescription among patients with cirrhosis compared to other forms of chronic disease BMJ Open Gastroenterol. 2019;6:e000271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Guy GP Jr., Zhang K, Bohm MK et al. Vital Signs: Changes in Opioid Prescribing in the United States, 2006–2015 MMWR Morb Mortal Wkly Rep. 2017;66:697–704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kugelmas M, Hill DB, Vivian B, Marsano L, McClain CJ. Cytokines and NASH: a pilot study of the effects of lifestyle modification and vitamin E Hepatology. 2003;38:413–419. [DOI] [PubMed] [Google Scholar]
- 32.Schuh-Hofer S, Wodarski R, Pfau DB et al. One night of total sleep deprivation promotes a state of generalized hyperalgesia: a surrogate pain model to study the relationship of insomnia and pain Pain. 2013;154:1613–1621. [DOI] [PubMed] [Google Scholar]
- 33.Nilsen TI, Holtermann A, Mork PJ. Physical exercise, body mass index, and risk of chronic pain in the low back and neck/shoulders: longitudinal data from the Nord-Trondelag Health Study Am J Epidemiol. 2011;174:267–273. [DOI] [PubMed] [Google Scholar]
- 34.Centers for Disease Control and Prevention. CDC Guideline for Prescribing Opioids for Chronic Pain - United States, 2016 Morbidity and Mortality Weekly Report. 2016;65. [DOI] [PubMed] [Google Scholar]
- 35.Busse JW, Wang L, Kamaleldin M et al. Opioids for Chronic Noncancer Pain: A Systematic Review and Meta-analysis Jama. 2018;320:2448–2460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Ray WA, Chung CP, Murray KT, Hall K, Stein CM. Prescription of Long-Acting Opioids and Mortality in Patients With Chronic Noncancer Pain JAMA. 2016;315:2415–2423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Moon AM, Jiang Y, Rogal SS, Becker J, Barritt ASt. In inpatients with cirrhosis opioid use is common and associated with length of stay and persistent use post-discharge PLoS One. 2020;15:e0229497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Moon AM, Jiang Y, Rogal SS, Tapper EB, Lieber SR, Barritt ASt. Opioid prescriptions are associated with hepatic encephalopathy in a national cohort of patients with compensated cirrhosis Aliment Pharmacol Ther. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Romero-Gomez M, Zelber-Sagi S, Trenell M. Treatment of NAFLD with diet, physical activity and exercise J Hepatol. 2017;67:829–846. [DOI] [PubMed] [Google Scholar]
- 40.St George A, Bauman A, Johnston A, Farrell G, Chey T, George J. Independent effects of physical activity in patients with nonalcoholic fatty liver disease Hepatology. 2009;50:68–76. [DOI] [PubMed] [Google Scholar]
- 41.Senna MK, Sallam RA, Ashour HS, Elarman M. Effect of weight reduction on the quality of life in obese patients with fibromyalgia syndrome: a randomized controlled trial Clin Rheumatol. 2012;31:1591–1597. [DOI] [PubMed] [Google Scholar]
- 42.Mork PJ, Vasseljen O, Nilsen TI. Association between physical exercise, body mass index, and risk of fibromyalgia: longitudinal data from the Norwegian Nord-Trondelag Health Study Arthritis Care Res (Hoboken). 2010;62:611–617. [DOI] [PubMed] [Google Scholar]
- 43.Benson GD. Acetaminophen in chronic liver disease Clin Pharmacol Ther. 1983;33:95–101. [DOI] [PubMed] [Google Scholar]
- 44.Lazo M, Hernaez R, Eberhardt MS et al. Prevalence of nonalcoholic fatty liver disease in the United States: the Third National Health and Nutrition Examination Survey, 1988–1994 Am J Epidemiol. 2013;178:38–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Adams LA, Lymp JF, St Sauver J et al. The natural history of nonalcoholic fatty liver disease: a population-based cohort study Gastroenterology. 2005;129:113–121. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.