Overdose in Patients Prescribed Chronic Opioid Therapy

The rate of overdose in patients who have been prescribed chronic opioid therapy is not known and neither is the relationship between the prescribed dose and the risk of overdose. The Annals of Internal Medicine published a cohort study which attempts to provide more data on this question in a population of patients with chronic, non-cancer related pain (unfortunately entitled Opioid Prescriptions for Chronic Pain and Overdose). It's funded by the National Institute of Drug Abuse.

The cohort is derived from the CONSORT (Consortium to Study Opioid Risks and Trends) study with all subjects belonging to the Group Health Cooperative (GHC) in Washington State at the time of enrollment. Subjects were eligible for inclusion if they were prescribed three consecutive months of opioids at the onset of opioid therapy and if they were given a diagnosis of chronic pain within two weeks of starting therapy. Patients with cancer were excluded. Information on opioid prescriptions was obtained from the GHC automatied pharmacy files. Potential overdose* cases were identified from the electronic medical record using one of two definitions, both relying on ICD codes (or combinations of codes) that might suggest opioid overdose. The first definition includes codes related to "opioid-related poisoning." The second definition includes a combination of codes that indicate an adverse opioid-related event plus a diagnosis code on the same date "considered to identify an overdose." Examples of the latter codes include the ICDs for drug induced delirium, altered mental status, pneumonia, and dyspnea. Thus, a patient on a stable, well tolerated opioid regimen who presents to the hospital with delirium and pneumonia might be a potential overdose case if given both the ICD for "adverse effect of opioid" and "pneumonia." (Even if the opioid was not the cause of delirium.)

After finding potential cases, they reviewed charts to classify the likelihood of each case being an overdose (definite, probable, uncertain, probably not, and definitely not), relying largely on the assessment and documentation of the treating clinician. (So a case like the one above, if documented "delirious secondary to morphine, community acquired pneumonia" would likely be labeled as a probable or definite overdose.)

They stratified opioid "exposure" using an "average daily opioid dose dispensed" over a 90-day exposure period: none, 1 to 19 mg, 20 to 49 mg, 50 to 99 mg, or >100 mg.

Here's more about the cohort:

• 9940 subjects included (Average age 54 with 60% females).
• Mean follow up was 42 months. 61% of subjects completed follow-up with most of the rest leaving the cooperative during the study.
• 2/3 of cohort had diagnosis of back or extremity pain.
• Mean daily dose of opioids prescribed was 13 mg (oral morphine equivalent) and most common opioid was hydrocodone.
  

And here's what they found:

• 6 fatal overdoses were identified (a rate of 17 per 100 000 person-years). Characteristics of this group of subjects are not described.
• 51 total overdoses (fatal + non-fatal) were labeled as "definite" or "probable" (a rate of 148 per 100 000 person-years).
• Many of the overdoses involved patient misuse of the prescribed opioids (suicide attempts, accidental excess ingestion, using opioids not prescribed, and misusing fentanyl patches).
• The most common adverse effect was delirium (in 23 patients).
• Average daily opioid dose dispensed seemed to correlate with the rate of overdose. Estimated annual overdose rates were 0.2%, 0.7%, and 1.8% among patients receiving less than 20 mg/d, 50 to 99 mg/d, and more than 100 mg/d of opioids, respectively.
• Having recently received sedatives/hypnotics correlated with increased risk of OD
• A history of depression was the only covariate examined that correlated significantly with an increased OD rate. The youngest and oldest (under 45 years, over 64 years) patients were at higher risk for OD than those in the middle, but this was not statistically significant. When comparing subjects who were in the "high dose" group to the "low dose" group, the high dose group had more men, smokers, history of depression treatment, substance abuse, and a higher Charlson comorbidity score. The relationship between comorbidity and OD was not reported.

It would be tempting make the plausible conclusion that higher doses of opioids cause more overdoses but as the authors state, the difference may be related to patient characteristics (and this is where I'd like to see more about the OD group, especially comorbidity scores). Additionally, I'm not sure how I feel about the use of physician documentation to determine the likelihood of OD. The palliative care practitioner routinely sees patients that have 2-3 simultaneous causes of delirium, and I think I'm less prone to blaming the opioid than other non-PC colleagues (whether that is due to experience or bias or both, I can't say for sure). How that clinical experience applies to this "healthier" population is unclear to me without seeing how frequently delirious subjects in the study had comorbid causes of delirium or other adverse effects that bought the label "overdose." On the other hand, it is possible that this method misses some OD subjects because the clinician may have missed it.

Speaking of a more classic palliative care population, I wonder how that group would differ from the results above (including patients with advanced cancer who may be on much higher average daily doses of opioid). Since that population is at higher risk for many of the diagnoses that were thought to be indicative of possible overdose (delirium, dyspnea, falls, others), I suspect the number of patients who would "rule in" as possible subjects would be much higher. This might in turn lead to more confirmed cases, although the importance of accurate diagnosis and documentation becomes an even bigger issue.

The 6 fatal overdoses are a concern and really I'd like to see more information about patient characteristics as well as evidence of aberrent opioid use in those specific cases.

The authors' modest conclusion is that in a chronic pain population, patients require careful observation and instruction on appropriate use. Although it's pretty nice to have some data on the topic, hopefully you've reached the same conclusion before reading this. The associated editorial suggests that it's " time to change our prescribing practices." Certainly, we need more research in this area to guide practice (more research into novel non-opioid pharmaceuticals, non-pharmaceutical therapies, and the management of pain in patients with depression or substance abuse, plus lots more). We also need better systems for monitoring chronic opioid use/prescriptions and to prevent diversion (which seems to be implicated in the majority of accidental overdoses, at least based on this study). And we already know that patients with chronic pain shouldn't be continued on opioid medications if they aren't leading to some reasonable, agreed upon outcome. So I don't think this article should result in a sea-change in practice or attitudes, unless you're a provider who has mistakenly believed that opioids don't have potentially serious adverse effects (just like many other medications that doctors prescribe).

For more Pallimed posts that touch on issues related to the topic of chronic pain, see here, here, here, and here.

*Rhetorical question: If a patient takes 15 mg of morphine and becomes delirious, is this an "overdose" or an "adverse effect"? What about the same response to 25 mg of amitriptyline?