Opioid Epidemic or Pain Crisis?

Using the Virginia All Payer Claims Database to Describe Opioid Medication Prescribing Patterns and Potential Harms for Patients With Cancer

Virginia T. LeBaron, PhD; Fabian Camacho, MS; Rajesh Balkrishnan, PhD; Nengliang (Aaron) Yao, PhD; Aaron M. Gilson, PhD


J Oncol Pract. 2019;15(12):e997-e1009. 

In This Article


Aim 1a: Describing POM Prescribing Patterns

Patients. Our patient sample included 4,324 unique patients with cancer (Table 2). Of these, 958 patients (22.16%) were prescribed a C-II POM at least three times in any study year (referred to as: three-or-more POM category patients) and by a prescriber likely to be treating cancer pain. The majority of these patients (n = 958) resided in HPD 1 (n = 530 [55.32%]); were 45 to 64 years old (n = 580 [60.54%]); were female (n = 491 [51.25%]); and were diagnosed with solid malignances (n = 839 [87.57%]), primarily breast (n = 134 [13.98%]) or lung (n = 126 [13.15%]) cancers (Data Supplement). Consistent with the composition of the CV-APCD data set, the largest payer source was Medicaid (n = 424 [44.26%]). Significant differences between the POM frequency groups were related to age, sex, payer source, county of residence within HPD 2 (P < .001), and tumor type (P = .024).

Prescribers. For three-or-more-POM category patients, C-II prescriptions for cancer-related pain were issued by 1,125 health care providers over the study period (Data Supplement). Physicians of any specialty (n = 847 [75.29%]) comprised the largest group of C-II prescribers (n = 14,618 prescriptions) followed by advanced practice registered nurses (n = 206), who wrote 3,520 C-II prescriptions. Of note, the prescriber/prescription ratio is highly similar for physicians (0.057) and advanced practice registered nurses (0.059).

Claims. A total of 222,000 prescriptions for any medication (opioid or nonopioid) were written for the three-or-more category patients, with 18,660 (8.4%) being for C-II POMs (Data Supplement). Of the C-II POMs, 2,567 (13.76%) were for extended-release C-II POMs, with 850 (4.56%) containing a deterrent formulation (eg, crush-resistant tablet). The three most frequently prescribed C-IIs included oxycodone-acetaminophen (n = 5,925 [31.75%]), hydrocodone-acetaminophen (n = 4,156 [22.27%]), and oxycodone hydrochloride (n = 3,545 [19.0%]). Eight prescriptions (0.04%) were written for meperidine tablets, and 270 patients (28.0%) were prescribed benzodiazepines three or more times in the same calendar year as that in which they were prescribed the C-II POMs. Longitudinal trends in absolute numbers of patients given C-II prescriptions show increases in 2014, most likely explained by the DEA upscheduling of hydrocodone (Data Supplement). When C-II prescription claims are considered per patient (to account for larger populations living in HPD 1 v HPD 2), no difference in the overall yearly trends was detected (P = .2022). However, grand means per HPD were significantly different (P = .0034), suggesting higher C-II use in patients residing in HPD 1, as were grand means per year (P = .0016), suggesting a decreasing trend in POM use across time (Figure 1; Data Supplement).

Figure 1.

No. of Controlled Substance Schedule II (C-II) claims, per patient, per Health Planning District (HPD), per year (2011–2015), for three-or-more–prescription opioid medication (POM) category patients.

Aim 1b: Exploring Predictors of POM Prescription Frequency

For both female and male patients, significant predictors of being a three-or-more POM category patient included cancer type, age, and county of residence (Table 3). For female patients, the odds of being a three-or-more POM category patient were significantly higher if the patient was diagnosed with multiple malignancies (odds ratio [OR], 2.13; P = .003) and significantly lower if the patient was diagnosed with breast (OR, 0.60; P = .009), CNS (OR, 0.26; P = .023), or skin (OR, 0.33; P < .000) cancer. For male patients, the odds of being a three-or-more POM category patient were significantly lower if the patient was diagnosed with colorectal (OR, 0.59; P = .035), GI (OR, 0.47; P = .031), genital-urinary (OR, 0.52; P = .003), prostate (OR, 0.29; P ≤ .000), or skin (OR, 0.27; P = .000) cancer. Both male and female patients had lower odds of being a three-or-more-POM patient if they were living in Buchanan (male: OR, 0.44; P = .000; female: OR, 0.64; P = .018) or Tazewell (male: OR, 0.49; P = .000; female: OR, 0.42; P < .001) counties and higher odds if they were between the ages of 30 and 44 years (male: OR, 2.87; P = .014; female: OR, 3.31; P = .001) or 45 to 64 years (male: OR, 2.58; P = .014; female: OR, 3.55; P = .000). Finally, male patients with Medicaid had lower odds of being a three-or-more-POM patient (OR, 0.69; P = .002).

Aim 2: Exploring POM-related Harms

Of all patients (n = 4,324), a total of 652 (15.1%) experienced 1,599 hospitalizations for any reason (mean hospitalizations = 2.45; SD, 2.22). Of these 652 patients, 10 or fewer patients were admitted for 11 OUD-related hospitalizations between 2011 and 2015. For each POM category, 10 or fewer patients were hospitalized for an OUD admission. For the one or two per year and three-or-more-POM categories, all patients admitted for an OUD were between the ages of 30 and 64 years and had been prescribed C-II POMs before, or the same year as, their OUD hospitalization. Although we examined distributions for age, sex, and tumor type for all POM categories as related to OUD hospitalizations, the results are restricted by the low number of OUD hospitalizations and by CV-APCD data suppression requirements that prohibit reporting on cells of 10 or fewer patients to preserve patient privacy.

Aim 3: Mapping Geospatial Patterns

In the choropleth maps (Figure 2), county level disparities exist between the highest number of DEA C-II authorized prescribers (Wise; City of Norton) and the highest rates of fatal POM overdose rates (Dickenson; Buchanan). It is interesting to note that Buchanan county has both the highest POM overdose rates and the lowest malignant cancer incidence rates, suggesting that patients with cancer may not be a significant contributing factor to the county's high number of POM overdoses. Heat maps (Figure 3) reveal the highest probability of a provider issuing a C-II prescription and being a patient who uses a C-II prescription in Wise County (consistent with where medical care and patient population is concentrated), whereas portions of Buchanan, Tazewell, and Lee counties reveal the lowest probability rates.

Figure 2.

Choropleth maps, per county. C-II, Controlled Substance Schedule II; DEA, Drug Enforcement Administration; POM, prescription opioid medication.

Figure 3.

Heat maps of prescription opioid medication (POM) probability rates for three-or-more POM category patients, by zip code. C-II, Controlled Substance Schedule II.