Identifying Patients for Overdose Prevention With ICD-9 Classification in the Emergency Department, Massachusetts, 2013-2014

Abstract

The national rise in opioid overdose deaths signifies a need to integrate overdose prevention within healthcare delivery settings. The emergency department (ED) is an opportune location for such interventions. To effectively integrate prevention services, the target population must be clearly defined. We used ICD-9 discharge codes to establish and apply overdose risk categories to ED patients seen from January 1, 2013 to December 31, 2014 at an urban safety-net hospital in Massachusetts with the goal of informing ED-based naloxone rescue kit distribution programs. Of 96,419 patients, 4,468 (4.6%) were at increased risk of opioid overdose, defined by prior opioid overdose, misuse, or polysubstance misuse. A small proportion of those at risk were prescribed opioids on a separate occasion. Use of risk categories defined by ICD-9 codes identified a notable proportion of ED patients at risk for overdose, and provides a systematic means to prioritize and direct clinical overdose prevention efforts.

Drug overdose is the leading cause of injury death in the United States.¹ Overdose deaths are largely a consequence of the opioid epidemic, which is attributed to a rise in polysubstance use, heroin use, and prescriptions for opioid pain relievers.^2,3 These deaths are preventable. Naloxone, an opioid antagonist that reverses fatal respiratory depression, is the standard of care for patients presenting with opioid overdose. Training new clinicians and nonclinicians to administer the medication will expand access to this lifesaving medication.⁴ Although the effectiveness of community-based distribution programs for reducing deaths from opioid overdose has been demonstrated^5,6 and access to naloxone is a priority of the US Department of Health and Human Services,⁷ efforts to engage drug users in overdose prevention are not widely incorporated into US health-care settings.

The emergency department (ED) is an opportune clinical setting to reach people at risk for opioid overdose. From 2000 to 2011, an estimated 2.9 million adult ED visits were related to opioid overdose, misuse, or dependence.⁸ ED-initiated treatment for opioid dependence has been found to improve engagement in care and has implications for overdose prevention, although few EDs have successfully established naloxone distribution policies.^9,10 A 2012 ED-based overdose prevention pilot in Massachusetts demonstrated the feasibility of engaging those at risk for and likely to witness overdose.¹¹ Another study found that emergency medicine physicians are willing to offer opioid harm-reduction interventions, yet significant barriers (e.g., provider knowledge, time, training, institutional support) remain.¹² Although documented operationalization of these efforts is limited, engaging drug users and their social networks in the ED setting offers a unique opportunity to reach those at risk for overdose who may not otherwise seek out prevention services.

In September 2013, the largest urban safety-net hospital in Massachusetts implemented a policy in partnership with the Massachusetts Department of Public Health to offer patients seen in the ED who are at risk for opioid overdose free intranasal naloxone rescue kits via a standing verbal order. A major barrier to the distribution of naloxone rescue kits identified by ED staff members was a lack of consensus on who should be given the medication. Because identifying the target population is necessary for designing and implementing programs for distributing naloxone kits, clarification of patients at risk for overdose is the first step toward integrating such programs in clinical settings. Use of diagnostic and mechanism-of-injury codes (i.e., International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes)¹³ to identify patients with opioid-related diagnosis is a promising approach for doing so.¹⁴ We used ED ICD-9-CM discharge data to establish and operationalize overdose risk categories with the goal of informing strategies for preventing overdose and distributing naloxone rescue kits in the ED.

Methods

We conducted a retrospective chart review of ED visits for patients aged ≥18 years at an urban safety-net hospital in Massachusetts from January 1, 2013 through December 31, 2014. We examined data on patients with a discharge diagnosis related to opioid overdose, opioid dependence or misuse, high-risk polysubstance use, and patients who were given an opioid prescription at discharge. We used ICD-9-CM codes to identify visits for inclusion in the analysis, establish risk categories, analyze patient demographic characteristics, and examine the number of patients who were categorized into multiple risk categories. This study used de-identified data and was declared not human subjects research by the Boston University Medical Campus Institutional Review Board. The authors thank Linda Rosen for her assistance in obtaining the data.

We defined and applied three risk categories to patients in the ED visit sample. Risk categories were defined and ranked based on likelihood of future overdose, as informed by the literature.^15–18 The first risk category, for opioid overdose, was defined as patients with ICD-9-CM discharge codes of poisoning or adverse effects of heroin, methadone, or other opiates (965.00, 965.01, 965.02, 965.09, 970.1, E850.00, E850.01, E850.02, E935.00, E935.10, E935.20, and E940.10). The second risk category, for opioid dependence or misuse and/or high-risk polysubstance misuse, was defined as patients with opioid or combination opioid dependence and abuse and patients with a high-risk polysubstance-related discharge involving sedatives, barbiturates, or a mixed/unspecified combination in addition to opioid dependence (304.00, 304.01, 304.02, 304.70, 304.71, 304.72, 305.50, 305.51, 305.52, 304.80, 304.81, 304.82, 305.90, 305.91, 305.92, 305.40, 305.41, and 305.42). The third risk category was defined as patients who were prescribed opioids during their ED visit (i.e., fentanyl, hydromorphone, meperidine, morphine, oxycodone, hydrocodone, propoxyphene, and oxymorphone).

We ranked each risk category according to level of risk. Opioid overdose was defined as the highest risk category because prior overdose is a major predictor of future overdose^17,18; opioid/polysubstance misuse was defined as the second highest risk category. We analyzed demographic characteristics by assigning each patient with an opioid-related diagnosis to only one risk category; if a patient fell into more than one category during the time period, he or she was assigned to the highest risk category. For example, if a patient met criteria for all three risk categories, the patient was assigned to the opioid overdose category for the purposes of demographic analysis. Among patients who met criteria for one or more risk categories and had multiple ED visits, we calculated the frequency of visits and the overlap between risk categories. We also compared the characteristics of ED patients who had an opioid-related diagnosis with all patients aged ≥18 years who visited the ED during the study period. We used SAS^® version 9.3 for analysis,¹⁹ and we used eulerAPE²⁰ for the Venn diagram to demonstrate risk category overlap.

Results

A total of 206,482 ED visits were made by 96,419 unique patients during the 2-year study period. Of these 96,419 patients, 18,622 (19.3%) met criteria for at least one risk category; 640 (0.7%) were categorized as opioid overdose, 3,828 (4.0%) were categorized as opioid/polysubstance misuse, and 14,154 (14.7%) were prescribed opioids. The two highest risk groups comprised 4,468 (4.6%) ED patients. Compared with all ED patients, those in the opioid overdose and opioid/polysubstance misuse categories were disproportionately non-Hispanic white and male. Patients who received an opioid prescription were similar to the total ED population in age, sex, and race/ethnicity (Table 1).

Table 1.

Demographic characteristics of patients aged ≥18 years seen in the emergency department (ED) of a Massachusetts safety-net hospital, by overdose risk category, January 1, 2013, to December 31, 2014^a

Characteristic	Total no. of unique ED patients (percent)	No. admitted for opioid overdoseb (percent)	No. admitted for opioid/polysubstance misusec (percent)	No. given an opioid prescription at discharged (percent)
Total	96,419 (100.0)	640 (0.7)	3,828 (4.0)	14,154 (14.7)
Mean number of ED visits per patient; range	2.1 (1-237)	1.1 (1-5)	1.5 (1-18)	1.2 (1-17)
Mean age, in years; range	42.2 (18-107)	46.6 (18-97)	40.5 (18-93)	42.4 (18-98)
Sex
Female	48,641 (50.5)	261 (40.8)	1,441 (37.6)	7,045 (49.8)
Male	47,778 (49.5)	379 (59.2)	2,387 (62.4)	7,109 (50.2)
Race/ethnicity
Non-Hispanic white	26,071 (27.1)	353 (55.2)	2,248 (58.7)	3,354 (23.7)
Non-Hispanic black	46,576 (48.3)	196 (30.6)	1,004 (26.2)	7,326 (51.8)
Hispanic/Latino	16,206 (16.8)	71 (11.1)	474 (12.4)	2,527 (17.9)
Asian	2,712 (2.8)	8 (1.2)	23 (0.6)	320 (2.3)
Other/unknown	4,854 (5.0)	12 (1.9)	79 (2.1)	627 (4.3)
Insurance
Medicaid	40,739 (42.3)	254 (39.7)	1,952 (51.0)	6,345 (44.8)
Medicare	12,939 (13.4)	68 (10.6)	484 (12.6)	1,741 (12.3)
Private	19,281 (20.0)	35 (5.5)	134 (3.5)	2,598 (18.4)
Uninsured	16,443 (17.1)	32 (5.0)	179 (4.7)	2,384 (16.8)
Other/unknown	7,017 (7.2)	251 (39.2)	1,079 (28.2)	1,086 (7.7)

^aCategories are mutually exclusive based on highest risk ranking.

^bDefined by ICD-9-CM codes: 965.00, 965.01, 965.02, 965.09, 970.1, E850.00, E850.10, E850.20, E935.00, E935.10, E935.20, and E940.10.

^cDefined by ICD-9-CM codes: 304.00, 304.01, 304.02, 304.70, 304.71, 304.72, 305.50, 305.51, 305.52, 304.80, 304.81, 304.82, 305.90, 305.91, 305.92, 305.40, 305.41, and 305.42.

^dDefined by prescription of fentanyl, hydromorphone, meperidine, morphine, oxycodone, hydrocodone, propoxyphene, and oxymorphone

Of 640 patients with opioid overdose, 284 (44.4%) came in on a separate occasion either before or after the overdose for opioid/polysubstance misuse, 61 (9.5%) were given an opioid prescription on a separate occasion, and 42 (6.6%) met criteria for all three categories on separate visits during the study period (Figure 1). Of the 3,828 patients with opioid/polysubstance misuse discharges, 486 (12.7%) were given an opioid prescription on a separate occasion. Some patients left the ED against medical advice: 66 (10.3%) with opioid overdose and 529 (13.8%) with opioid/polysubstance misuse. Of the 640 overdose patients, 11 (1.7%) had fatal overdoses.

Figure 1.

Unique patients aged ≥18 years seen in the emergency department of a Massachusetts safety-net hospital (n = 18,622), by opioid risk category, January 1, 2013, to December 31, 2014^a. Abbreviation: OD, overdose.

Discussion

Our analysis identified ED patients at risk for opioid overdose and provides insight into how interventions can identify target populations for overdose prevention using risk categories. Although this analysis did not account for the chronological order of multiple visits to the ED, some patients in the two highest risk categories were also given opioid prescriptions in the ED at some point during the study period. This small but not insignificant proportion of patients represents the group at highest risk for a future opioid overdose and in most need of targeted prevention efforts. Findings of this analysis also underscore the importance of screening, intervention, and referral to treatment in the ED setting, as well as consideration of prescribing opioids to patients at risk for overdose. The demographic characteristics of patients in this analysis are consistent with those of such patients described in the literature.²¹ The slightly higher rate of opioid prescriptions for non-Hispanic black people compared with non-Hispanic white people is inconsistent with national data, which show a substantial disparity between these two groups in pain treatment with opioid pain medication.²²

Clearly defining the target population for overdose prevention efforts is necessary for successful integration within the ED setting, and doing so must appropriately balance prioritization of the population most at risk while reaching the greatest number of those at risk. Designating only patients with opioid overdose as the target population would be a missed opportunity to engage those with opioid, high-risk polysubstance, or other drug misuse who are at comparable risk. Although not all patients who receive a prescription for opioids require overdose prevention efforts, those with a history of opioid misuse, polysubstance misuse, opioid overdose, or those prescribed high-dose opioids (defined as >100 mg of morphine equivalent) may be appropriate candidates for intervention. Overdose prevention efforts may be best directed after presentation at the ED or toward the social networks (i.e., emergency contacts) of overdose survivors. Our findings show that 13.3% of patients in the two highest risk categories left the ED against medical advice. Thus, some patients at risk for overdose may refuse a naloxone rescue kit. Ultimately, prioritization is prudent for guiding the allocation of limited time and resources. Use of risk categories defined by ICD-9-CM discharge codes may facilitate prioritization of the distribution of naloxone rescue kits and could be supported by an electronic medical record alert for clinicians working with patients who are potentially at risk for overdose. Categories can be adapted and applied to International Classification of Diseases, Tenth Revision codes, which capture greater diagnostic detail and, as such, may be better suited to identify those at risk for opioid overdose.

Limitations

This study was limited by its reliance on medical record review data, specifically ICD-9-CM discharge codes and prescription records. Because some physicians are reluctant to document substance dependence, our analysis may have captured data on only the most severe cases, thereby underestimating the true number of patients with drug misuse or dependence. Use of these data to define risk categories is, however, a potentially powerful tool. Additionally, the dataset did not allow us to examine the chronological order of the ED encounters; as such, we could not determine whether overdose-related encounters occurred before or after other opioid- or polysubstance-related encounters.

Conclusion

Until recently, the ED has been overlooked as a venue to engage individuals at risk for overdose in training on use of naloxone and distributing naloxone rescue kits. This study confirms that a substantial proportion of the ED patient population is at risk for overdose, emphasizing the need to integrate prevention services into this setting. By operationalizing overdose risk categories with a sample of ED patients, we hope to inform efforts to identify the target population for ED-based overdose prevention interventions.²³ Additional research is needed to determine how to best use this risk-classification system to direct overdose prevention interventions for high-risk populations. Unlike other substance use disorders, opioid use disorder is, in part, driven by the healthcare delivery system. Engaging providers and people who use drugs in overdose prevention efforts in the ED setting is, therefore, just one component of a system-wide response to the opioid epidemic.