Overview and Participants
Data for the study were derived from the WVCS survey, described in detail elsewhere.35, 36 Eligible participants were military veterans of the OEF/OIF/OND era who enrolled for care in VHA settings between September 12, 2001 and September 30, 2012 (n = 625,082). A subsample of veterans participated in three surveys were administered yearly and recruited primarily via mailings sent to 8465 potential participants from eight states; female veterans were oversampled as a means to equalize numbers of male and female participants. A total of 767 veterans provided consent, and 662 completed the baseline WVCS survey between July 2008 and December 2011, for an overall response rate of 7.7%. This secondary analysis used baseline data from the WVCS survey, and only those who endorsed having pain for 3 months or longer were included, yielding a sample of 460 veterans. The study was approved by the Institutional Review Boards at the VA Connecticut Healthcare System and Yale University School of Medicine.
The demographic section of the baseline WVCS survey included age, sex, race/ethnicity, marital status, employment status, and education.
Use of Non-Pharmacological Modalities (NPMs)
The survey asked of respondents used the following approaches to manage pain in the past year: physical therapy, chiropractic, acupuncture, psychotherapy, massage, educational classes, or exercise. Veterans responded based on all use of NPMs, inside or outside of VHA. Using categories defined by the VA SOTA, we created three variables: use of psychological/behavioral therapies, including psychotherapy or educational classes; use of exercise/movement therapies, including physical therapy or exercise; and use of manual therapies, including chiropractic, acupuncture, or massage.
The Brief Pain Inventory (BPI) 37 was used to measure worst, least, average, and current pain intensity in the past week on 0 (no pain) to 10 (worst pain imaginable) scales. Participants also indicated location(s) of their pain by responding to a checklist of possible types (i.e., back pain, neck pain, headache or migraine, stomach ache or abdominal, joint pain, chest pain, facial ache or pain, or whole body pain); finally, participants also indicated the length of time they had experienced pain using categories ranging from 3 to 6 months up to more than 4 years.
Veterans reporting chronic pain were asked about past-week use of the following opioid analgesics, listed verbatim: morphine (MS Contin, Roxanol, Oramorph), oxycodone (OxyContin, Percocet, Roxicet, Tylox, Endocet), hydrocodone (Vicodin, Lortab, Norco), methadone (Dolophine), fentanyl (Duragesic, Actiq), codeine (Tylenol #3,Tylenol #4), propoxyphene (Darvon, Darvocet), or tramadol (Ultram, Ultracet). Veterans were instructed to report on all medication use, regardless of VHA/non-VHA prescriber status. We created a variable “opioid pain medication use” to indicate whether a veteran used at least one opioid pain medication in the past week.
Non-Opioid Pain Medication Use
Veterans endorsing chronic pain were asked about past-week use of the following non-opioid analgesics, listed verbatim: non-steroidal anti-inflammatory drugs (NSAIDs) (e.g., ibuprofen [Motrin], naproxen [Naprosyn, Aleve], indomethacin [Indocin], meloxicam [Mobic], etodolac, or Celebrex), acetaminophen (Tylenol), herbal or nutritional supplements, tricyclics (e.g., amitriptyline [Elavil], nortriptyline [Pamelor], doxepin, imipramine), muscle relaxants (e.g., cyclobenzaprine [Flexeril], methocarbamol [Robaxin], carisoprodol [Soma]), anti-epileptics (e.g., gabapentin [Neurontin], Lyrica, Topamax), and SNRIs (e.g., Cymbalta, venlafaxine [Effexor]). We did not ask about topical analgesics. Veterans were instructed to respond based on all medication use, regardless of VHA/non-VHA prescriber status. We created a variable “any non-opioid pain medication use” to indicate whether a veteran used at least one non-opioid pain medication in the past week.
Mental Health Symptoms
The survey included the Patient Health Questionnaire-9 (PHQ-9),38, 39 a self-report measure of depressive symptoms, and the PTSD Symptom Checklist-Military version (PCL-M),40, 41 a self-report measure of PTSD symptoms. In this sample, these two measures were highly correlated (r = 0.79). Therefore, we elected to use a symptom composite to avoid concerns of multicollinearity. PHQ-9 scores and PCL-M scores were converted into z scores (M = 0, SD = 1) and averaged to create a mental health composite score representing overall mental health symptom severity.
Descriptive statistics for the entire sample were calculated for all variables. Less than 1% of data were missing. We examined bivariate relationships between frequency of use of NPMs and demographic/clinical variables (e.g., age, race/ethnicity, sex, education, pain severity, mental health symptom severity, opioid and non-opioid pain medication use) using independent sample t tests, correlations, and chi square tests as appropriate. We used logistic regression to examine demographic and clinical factors associated with the use of each category of NPMs.