As discussed later, other substance use disorders have been consistently found to confer the highest risk of opioid use disorder [22, 23]. Tobacco and alcohol are the most commonly abused substances in this country with 64 million tobacco users in 2015, 66.7 million “binge,” and 17.3 million “heavy” alcohol users [24]. Numerous large studies over the past couple decades have shown fairly consistently that tobacco use is associated with a higher rate of opioid use and abuse [25–27]. A recent large (n > 24,000) study showed that smokers were greater than three times more likely to report opioid misuse and three to five times more likely to meet OUD criteria relative to non-smokers [27].

While numerous investigations suggest increased risk of OUD among patients with comorbid alcohol use disorder, large-scale epidemiologic data and even smaller individual retrospective or prospective studies are very scarce. This is likely due in part to a number of factors, including ubiquity of alcohol use in the general population, well-documented underreporting of alcohol use, and also lack of consensus for alcohol use disorder definition/cutoffs.

Among 1883 patients using opioids daily in a managed care environment on the West Coast, 12.4% admitted to concurrent use of alcohol, defined as having a drink within 2 h of consuming opioids [28]. Within this study, however, there was no difference between concurrent drinkers vs. nondrinkers in terms of diagnosis or self-report of substance use disorder, other than concurrent drinkers being statistically significantly more likely to show an AUDIT-C (a self-report tool for quantitation of alcohol consumption) score consistent with alcohol use disorder. Coexisting alcohol use disorder has been reported to be as prevalent as one-third among patients undergoing opioid maintenance therapy [29]. Daily alcohol use has been associated with increased risk of prescription drug misuse in general [30], and an early investigation using NESARC data reported a 5% and 15% prevalence of prescription opioid misuse among alcohol-abusing and alcohol-dependent individuals, respectively, compared to 0.6% for those who had abstained from alcohol in the previous year [31]. A recent study using data from NESARC Waves 1 and 2 shows an association between early-onset alcohol abuse and later development of prescription drug, including opioid misuse, [32] and makes a strong case for the frequently cited “gateway” phenomenon of drugs such as alcohol and marijuana leading to OUD and other substance use disorders.

Cannabis is the most commonly illicit substance of abuse in this country, and its use has been shown in several recent investigations to be strongly predictive of opioid use [33–35].

Benzodiazepines are frequently co-prescribed with opioids; the ever-growing body of evidence [36–38] from emergency department visit and postmortem analyses of overdose victims demonstrates the tremendous and generally unacceptable risk of this combination. Nonetheless, the practice often continues, not infrequently due to ignorance of (hopefully not apathy concerning) other providers’ treatment plans that may include benzodiazepines. The use of benzodiazepines has been shown in multiple investigations to confer increased risk of opioid use disorder [39, 40], and a large (n = 17,074) Norwegian study published at the beginning of this decade showed an unadjusted odds ratio of 7:7 for chronic opioid use from previous benzodiazepine prescription [41]. After adjusting for alcohol and tobacco use, chronic pain, and socioeconomic variables, the effect was reduced by a little over 50% (odds ratio 3.1); nonetheless this sample suggested that benzodiazepine use exceeds even chronic pain as a risk factor for opioid use.

Benzodiazepine use of course probably bears significant association with opioid use in that it is a surrogate for anxiety and a predilection for “chemical coping” with distress. The association between anxiety disorders and opioid use disorders has been reported for decades [42–45]. Whether pre-existing anxiety (or other psychiatric disorder) precedes and predicts the development of OUD or vice versa has been debated in the literature, and support for both pathways exists, as discussed below. “Self-medication” of emotional distress including anxiety by opioids has long been recognized as a significant issue driving OUD; conversely, anxiety almost universally accompanies the development of dependence and withdrawal states. A common underlying vulnerability to both issues has also been postulated but awaits proof. The recent availability of large-scale population databases (e.g., NESARC) has allowed for some attempt to investigate the temporal directions/progression of these associations. One investigation examining the question of whether psychiatric disorders including anxiety preceded the development of OUD or vice versa showed support for both directional hypotheses [46]. In this study using NESARC Wave 1 data, nonmedical use of opioids was associated with a threefold higher rate of development of anxiety, and on the other hand, the odds of developing OUD was 6-fold higher in patients with anxiety disorders in general and nearly 11-fold higher in patients with generalized anxiety disorder. A follow-up study using NESARC Waves 1 and 2 showed similar results [47].

A cognitive distortion common to anxiety disorders (and also PTSD, discussed below, which was previously categorized within the anxiety disorders in the DSM-IV) is pain catastrophization , defined as a negative perceptual filter applied to actual or anticipated pain. Components include feelings and thoughts of inevitability of and helplessness about pain, rumination on pain, and magnification of pain. Pain catastrophization has been associated in numerous studies with increased incidence and severity of chronic pain [48–50] and also independently with increased risk of opioid misuse and OUD [51–53]. In the study by Martel et al., catastrophizing conferred statistically significant risk for opioid misuse even after controlling for pain severity, anxiety, and depressive symptoms [52].

Post-traumatic stress disorder (PTSD) as defined in the DSM-5 is a syndrome of persistent reexperiencing of distress (including nightmares, intrusive thoughts, flashbacks, etc.) and other symptom clusters of avoidance, negative alterations in cognition and mood, and alterations in arousal and reactivity following exposure to actual or threatened death, serious injury, or sexual violence [54].

PTSD has strong independent associations with chronic pain [55–57] and has been shown for decades to be associated with higher rates of substance abuse. PTSD confers a higher risk of heroin use [58], and those with this dual diagnosis are more prone to overdose and otherwise show worse treatment outcomes [59]. More recently, specific association of PTSD with general opioid misuse and OUD has been shown in both military and civilian populations [60–63], although the latter two studies, both using NESARC data, showed that full-blown OUD (again, subject to that database’s limitations of self-report and lay interviewers) occurred in the PTSD population only among females.

Moderate to severe depression has long been understood to both contribute to and also stem from chronic pain [64–66], and, similarly, earlier literature on the association between depression and opioid misuse could not clearly identify/support directionality of risk [67, 68]. As discussed in Chap. 4, more recent data support a causal association between chronic opioid use and resultant depression [69, 70], whereas the converse (depression increasing risk for opioid misuse/OUD) has also been shown [46, 47]. Given the depressant effects of opioids, it seems intuitively less likely that patients suffering with depressive disorders would self-medicate with this drug class compared to those with anxiety and PTSD, and the odds ratios bear that out. Complex comorbidities abound however, with depression often intertwined with chronic pain, substance use disorders in general, anxiety, and other risk factors. Fink et al. used data from the 2011–2012 National Survey of Drug Use and Health (n = 113,665) to show that patients suffering from comorbid depression and OUD were more likely to be female, of low annual income, not currently married, and to report an alcohol use disorder or other drug use [71]; numerous confounders between these variable exist of course, rendering conclusions difficult.

Associations between bipolar disorder and chronic pain are far less frequently reported, although some reports exist [72]. Substance use disorders on the other hand are extremely comorbid with bipolar disorder; their coexistence is regarded by many as “the norm” [73]. While opioids do not appear to be the drug of choice among this population, with international literature showing a preference for alcohol, cannabis, and cocaine all greater than opioids [74], the condition does remain a significant risk factor for the development of OUD [46, 47].

Personality disorders have been associated with OUD; borderline personality disorder (BPD ) in particular has shown some consistency as a predictive variable within the literature [75–77]. Heightened sensitivity to physical pain sensation in addition to lowered emotional distress threshold and heightened impulsivity have both been suggested as contributors to this association. A recent population-based review using NESARC data identified borderline, schizotypal, and antisocial personality disorders as all being predictive of opioid misuse and OUD [78].

Various personality factors have also been reported to confer both increased and decreased risk for opioid misuse and OUD. As indicated above, impulsivity appears to be associated with OUD irrespective of DSM diagnosis [79–81]. A study of 312 opioid addicts in Serbia (compared to 346 controls) linked high novelty seeking and low reward dependence, as well as self-transcendence to OUD [82]. A Dutch study comparing 161 opioid misusers without a diagnosis of OUD to 402 methadone- or heroin-maintained addicts (and a third group of 135 non-heroin users) found that the misusers were more likely to report increased novelty seeking, self-transcendence, harm avoidance, and less self-directedness than healthy non-using controls. Conversely, they reported greater social reward dependence and self-directedness than diagnosed addicts [83].

Disruption of normal childhood development by various aberrancies has long been held to confer later-life psychological/psychiatric dysfunction. Plentiful evidence of childhood sexual and physical abuse leading to various disturbances including substance abuse and dependence exist and are considered in a later section. Harm needs not occur solely in the context of these gross violations; however, verbal and emotional abuse and neglect may also predict opioid misuse and OUD. An Australian study of nearly 1000 opioid-dependent individuals showed an odds ratio of 1.9 for frequent emotional abuse in male OUD subjects (no difference for female OUD subjects vs. controls) [84]. Outright child neglect, as well as detached/disengaged parenting styles, has been shown to increase the risk of substance abuse in general [85, 86] as has a less encouraging/more rejecting parental role [87]. (Significant overlap between nonphysical forms of abuse with physical and sexual abuse exists of course, which confounds the associations.) Parental loss through death or divorce and even family discord seem to increase the risk as well [85, 88], and adoptee status is associated with nearly a twofold increase in substance use disorder risk, with an adjusted odds ratio of 2.2 for opioids specifically [89]. A fascinating link between childhood parental loss (and separation anxiety disorder in general) with disruption of the endogenous opioid system has been elucidated by the panic disorder research community [90, 91]. This deficiency in endogenous opioid activity, with its complex ramifications upon the neuroimmunoendocrine system, may explain in part the vulnerability of these individuals to exogenous opioids in later life.

Data show that women receive more opioid prescriptions than men [92, 93] perhaps associated with a higher incidence of severe pain complaints [94, 95]. Numerous studies over the past decade have investigated whether gender is associated with opioid misuse and dependence [24, 25, 96–101]. Despite variability in settings, most show a consistent pattern of increased prescription opioid misuse and dependence in men compared to women. A recent large investigation using NESARC data [62] revealed greater prevalence of opioid misuse among men than women but no difference in the prevalence of OUD between the sexes.

Jamison et al. reported similar degrees of aberrant opioid use between men and women, but a greater association among women to misuse opioids is “due to emotional issues and affective distress while men tend to misuse opioids due to legal and problematic behavioral issues [98].”

Among numerous tragic consequences of childhood physical and sexual abuse, a substantial body of literature bears witness to markedly increased rates of substance abuse in victims [102–105]. More recently childhood physical and sexual abuse has been linked with increased risk of opioid abuse and dependence, specifically [84, 106]. So well known is this association that screening for preadolescent sexual abuse is a component of opioid misuse risk stratification instruments, such as the Opioid Risk Tool (ORT) discussed below. While self-medication of physical pain symptoms (not uncommon in abusive relationships, which tend to correlate with substance abuse) is certainly present in many situations, the literature on the subject draws considerable attention/lends support to the very plausible theory that a tremendous amount of opioid misuse stems from an attempt to long-standing emotional wounds and distress.

While not as well represented in the literature, there is growing awareness that post-childhood sexual abuse [107–109] and other interpersonal violence particularly toward women [110] are associated with an increased incidence of opioid misuse and dependence. Unidirectional association of abuse preceding opioid misuse/OUD clearly seen in children (who do not use opioids for the most part) is not as clear with adults, in whom there is likely a circular pattern of abuse preceding opioid misuse and that misuse likely facilitating further abuse, etc.

The Opioid Risk Tool (ORT), published in 2005 by Dr. Lynn Webster, was designed specifically “to predict the probability of a patient displaying aberrant behaviors when prescribed opioids for chronic pain” [134]. The ORT is a self-report questionnaire designed specifically for new (not at all synonymous with opioid-naïve) patient screening and assesses both personal and family history of substance abuse including alcohol, prescription and illicit drugs, age, history of preadolescent sexual abuse, and specific psychiatric diagnoses. Positive answers are assigned a weighted point value for overall risk contribution, and the sum is tabulated and categorized into low-risk (score 0–3), moderate-risk (score 4–7), and high-risk (score >7) groups. This original publication reported results from a sample of 185 patients new to the authors’ pain clinic. Subsequent aberrant drug-related behaviors (including soliciting prescriptions from other providers, using unauthorized/illicit opioids, abnormal drug screening, unsanctioned dose escalation, missing visits) were identified after a 12-month period in 6% of patients categorized as low risk, 28% of patients categorized as moderate risk, and 91% of those categorized as high risk. Increasing number of aberrant behaviors also correlated with increasing score.

Moore et al. [135] conducted a small but important study of 48 patients new to their pain clinic who were initially prescribed opioids but subsequently had opioid therapy terminated for aberrant behaviors. Besides an initial interview by one of the psychologists, the sample was subjected to the ORT, and two other risk assessment tools described below in a head-to-head comparison of the instruments’ sensitivity for predicting aberrancy. In this analysis, when evaluating patients assigned an initial ORT risk category of moderate or high risk, the instrument showed a sensitivity value of 0.45 (21 of the 48 patients accurately identified a priori.) The sensitivity was reduced to 0.10 when evaluating only patients categorized as high risk by the instrument.

Two follow-up studies comparing the same instruments in larger samples (n = 132 and 263) and using the same methodology [136] showed similar low sensitivity for the ORT (0.10 and 0.18, respectively) but superior specificity (0.88) among the screening tools. Thus, while more likely to miss patients at risk for opioid misuse and abuse, the ORT has the lowest likelihood of false positives among commonly used tools.

A German study evaluating a cancer population (n = 114) compared the predictive value of the ORT for urine drug testing (UDT) abnormalities [137]. This study found a higher proportion of aberrant UDT (positive primarily for cannabis) in patients categorized as moderate risk (69%) and high risk (59%) compared to those categorized as low risk (7%). Significant limitations of this analysis however included the fact that some patients did not fill out the ORT themselves, with questionnaire completed retrospectively by staff. Furthermore, not all patients underwent urine drug screening, with the test biased toward those patients assigned a higher-risk categorization (79% of high-risk patients undergoing UDT compared to 52% in the moderate-risk group and 21% in the low-risk group).

The Screener and Opioid Assessment for Patients with Pain (SOAPP) is a self-report questionnaire designed to predict opioid misuse and abuse among chronic pain patients considered for long-term opioid therapy. The original instrument (SOAPP) was designed using eight concept clusters listed here in descending order of predictive importance: antisocial behaviors/history, substance abuse history, medication-related behaviors, doctor-patient relationship factors, psychiatric history, emotional attachment to pain medicine, personal care and lifestyle issues, and finally psychosocial problems [138].

An initial validation study by the developers [138] reported good sensitivity (0.91) and specificity (0.69) at a cutoff score of 7 or greater in predicting aberrant drug-related behaviors after 6 months. A second validation study by the same group [139] showed markedly lower performance, with sensitivity of 0.68 and specificity of 0.39 for a cutoff score of 8 or greater; the authors reported however that 10% of the sample did not complete the form and were excluded, and the comparison standard (UDT) was not applied to all patients in this sample.

In the analysis of Moore et al. [135], the SOAPP achieved a relatively high sensitivity value (0.73), with 35 of 48 of the aberrancy-displaying/discharged patients having received a high-risk rating (score greater than 6) at baseline.

The initial version, however, was perceived to be excessively vulnerable to deceptive answers and furthermore was conceptually flawed in that predictive validity which was tested primarily against self-reported aberrant behaviors at follow-up [140]. As such, SOAPP subsequently underwent revision (SOAPP-R) which included a focus on eliminating “admission of socially unacceptable behaviors” and incorporating more “subtle” predictors (deemed less transparent to respondents) from the literature such as impulsivity, anger, resentment, and boredom to complement or update the initial instrument (Fig. 10.1). Scoring categories based upon subsequent analysis introduced below include low-risk (score <10), moderate-risk (score 10–21), and high-risk (score >21).

In the initial validation study [140], an outcome measure (“aberrant drug behavior index” or ADBI) was created with score based upon self-report, physician assessment, and UDT. In the final analysis, 223 patients’ SOAPP-R data were compared to their ADBI score, and at the recommended cutoff score (18) for positivity, sensitivity was shown to be 0.81 with a specificity of 0.68. Subsequent cross-validation analysis in a different sample showed comparable predictive value and reliability [141].