Among cancer patients taking prescription opioids, opioid prescribing patterns are associated with the risk of opioid overdose death. Medicinal marijuana has been shown to have analgesic properties, and specifically for cancer patients, has demonstrated relief from adverse effects of therapy like nausea and anorexia, with few reports even suggesting antineoplastic activity. Recent research among Medicaid beneficiaries suggests that medical and adult-use marijuana has the potential to lower opioid prescriptions. As of 2016, approximately 60% of the US population now resides in states with legalized use of medicinal marijuana, which highlights increasing public support given its promising medical benefits. A cross-sectional survey of adult cancer patients in Washington State showed that nearly a quarter of patients reported active cannabis use. Classification of marijuana as a Schedule I controlled substance , however, makes large-scale clinical studies challenging. While marijuana use appears to be quite promising in the management of chronic and neuropathic pain, there are associated adverse effects including the potential for addiction, impairment of memory and judgement, and the potential to exacerbate psychiatric illness including depression and anxiety. There is limited population-based or epidemiologic data on marijuana and other substance use specifically in patients with cancer. The primary objectives of this study were to examine the associations between cancer and marijuana use as well as between cancer and prescription opioid use in a population-based setting. We also sought to examine trends in marijuana and opioid use over a 10-year period given the evolving legislation for marijuana legalization and dynamic temporal changes in prescription opioid use. We compiled population-based datasets from the US National Health and Nutrition Examination Survey ,commercial grow setup a survey designed to assess the health and nutritional status of non-institutionalized adults and children in the US.
This nationally representative, biennially administered survey interviews 10,000 individuals per two-year cycle about demographic characteristics , substance use, and medical conditions. We compiled five biennial datasets from 2005-2014 and included all respondents aged 20-60 years , which includes all respondents that were asked to report on a cancer diagnosis and marijuana use . Respondents missing a definitive ‘yes’ or ‘no’ response to cancer diagnosis were excluded. Table 1 summarizes the NHANES variables considered in the analyses. Respondents were grouped by reported diagnosis of cancer. For respondents with multiple cancer diagnoses, primary cancer site was defined as the first site reported. Demographic variables of interest included age, gender, race, education, self-reported health status, low income, and insurance coverage. Age was analyzed as a continuous variable. Race was categorized as non-Hispanic white, non-Hispanic black , Hispanic, and other. Education was dichotomized as less than college-level education versus college-level education or beyond. Self-reported health status was dichotomized as “good” versus “poor” . Low income was categorized as annual household income of less than $20,00031 versus $20,000 and above given the average federal poverty line for a family of four from 2005-2014. Insurance coverage status was categorized as covered or not covered. Current marijuana use was defined as use within the past 30 days and recent marijuana use as use within the past year. Prescription opioid use was defined per the Prescription Medication subsection of the survey on use of prescription medications during a one-month period prior to the survey date and included the following generic drug names: morphine, hydrocodone, codeine, oxycodone, fentanyl, dihydrocodeine, hydromorphone, meperidine, and methadone. Additional substance use variables included cigarette smoking, binge alcohol use, and illicit drug use. Cigarette smoking was defined as having smoked at least 100 cigarettes in a lifetime.
Binge alcohol use was defined as drinking an average of more than 5 drinks/ drinking day in the last year for men and more than 3 drinks/drinking day for women. Illicit drugs included cocaine, heroin, and methamphetamines . Current illicit drug use was defined as use within 30 days. The primary explanatory variable of interest was diagnosis of cancer, while the primary outcome variables were marijuana use and prescription opioid use. Other associated variables explored included previously-described demographic variables and other substance use including alcohol, smoking, and current illicit drug use. Given the potential for poly substance use in this cohort,we also investigated the relationship between our primary outcomes of marijuana and opioid use. Propensity score matching was performed to compare respondents with cancer to controls . A 1:2 matching was performed based on a nearest-neighbor matching algorithm with a caliper width of 0.1 of the propensity score with age, gender, race, education, and self-reported health status as co-variables. These demographics were chosen to better estimate the association between cancer diagnosis and marijuana and prescription opioid use, especially given the tendency of NHANES to over sample certain groups . Cancer respondents and propensity score matched controls were compared for primary outcome measures of current marijuana use and prescription opioid use using Pearson chisquare tests for categorical data and independent sample t-tests for continuous data . Univariable and multi-variable logistic regressions were used to evaluate significantly associated variables of marijuana and prescription opioid use among both cancer and non-cancer matched controls . Demographic and substance use co-variables that were not significant at level P<0.10 on multi-variable analyses were removed via backward stepwise elimination from the final multi-variable logistic regression models 36. Conditional logistic regression models were used when analyzing the propensity score matched cohort to account for the matched pairs.
Logistic regressions were used to investigate trends in marijuana and opioid use over the 10- year time-period for all NHANES respondents as well as cancer respondents, and to investigate differences in these trends between respondents with cancer and matched controls by using an interaction term of year and cancer diagnosis. Survey sampling weight, strata, and clusters were accounted for in any analysis of non-propensity score matched cohorts . Two tailed P<.05 was considered significant for all analyses. All statistical analyses were done using SAS v9.4 . In an era of rapidly evolving marijuana legislation and a growing opioid epidemic, it has become critically important to understand and quantify current substance use patterns. To our knowledge, this is the first population-based analysis of the prevalence of marijuana and prescription opioid use in people with a cancer diagnosis. Among cancer respondents, 8.7% and 40.3% reported using marijuana in the last 30 days and one year, respectively. This contrasts with a recent survey of cancer patients in Washington State which found that 24% used cannabis in the last year and 21% in the last 30 days. While cancer respondents in this study self-reported more current and recent use of marijuana than non-cancer matched controls, cancer was not significantly associated with current marijuana use. This may be in part because our data do not specify medicinal versus recreational marijuana use, the former being more associated with managing cancer-related symptoms, including pain. Among cancer patients surveyed in Washington State, active users reported using cannabis most frequently for pain. Also, we analyzed years 2004-2015, so perhaps with future datasets reflecting the evolving role of marijuana in oncology18 and broadening legalization, the association of cancer and marijuana use may change. Nearly 14% of cancer respondents reported prescription opioid use in the last month,vertical grow racks for sale and cancer diagnosis was the only variable significantly associated with opioid use. Indeed, opioid analgesics are critical to the management of moderate to severe cancer-related pain,and we cannot draw conclusions regarding the association between cancer status and opioid misuse from this analysis presented here. However, it is becoming increasingly important to identify risk factors for opioid misuse, such as younger age and higher pain levels, which have previously been identified among cancer patients being treated for pain. We did find that insurance status trended towards a significant association with opioid use, likely reflecting access to a prescribing provider. A previous study found that uninsured and low income adults had a higher prevalence of prescription opioid misuse and substance use disorders. While there are no randomized trials of marijuana compared with prescription opioids for cancer-related pain, patients are increasingly reporting the use of cannabis as a substitute for prescription opioids.
Oncology patients may have apprehensions about opioids including fear of dependence and potential side effects. Indeed, the most commonly reported motivation for opioid misuse is pain relief, yet these fears introduce potential barriers to effective cancer pain management. Medical marijuana legalization has been associated with a 23% reduction in hospitalizations related to opioid dependence or abuse, suggesting that if patients are in fact substituting opioids with marijuana, this substitution may reduce the risks of opioid-related health problems. However, most large-scale randomized trials of marijuana use for pain are limited to non-cancer pain17, and there may be potential adverse effects of marijuana use that should be considered. We found an increase in the proportion of marijuana users between 2005-2006 and 2013-2014 with a significantly increased likelihood of 5% each two-year study period among all survey respondents. This finding reflects increased US support of marijuana legalization and changes to local and state legislation over this decade. In 2005, 36% of the population supported marijuana legalization; in 2014, 51% of Americans were supportive. Between 2005-2014, seven states legalized medical marijuana, while four states and Washington, DC legalized marijuana for recreational use. By November 2014, nearly 175 million people lived in areas where recreational or medical marijuana were fully legal or decriminalized. This phenomenon is particularly relevant for oncology, as prior studies have shown that legalization is an important factor in cancer patients’ decision to use cannabis. Given the current opioid epidemic with sales of opioid pain relievers quadrupling between 1999 and 2010, it is interesting that there was no significant increase in the proportion of respondents using prescription opioids between 2005-2006 and 2013-2014. This outcome echoes a recent Centers for Disease Control report, which found that recent annual opioid prescribing rates actually decreased by 13.1% between 2012 and 2015, yet still remained three times as high compared to 1999. A recent observational study over a 6 year period found that doses of opioids prescribed to cancer patients had decreased. These recent decreases suggest heightened awareness among physicians and all patients about the risks associated with opioid pain relievers. The increase in marijuana use measured in this study in the context of stable opioid use highlights the significance of increasing marijuana usage between 2005-2006 and 2013-2014. This study has several limitations. Given the cross-sectional study design, our findings are associations and not indicative of a causal relationship between cancer and marijuana or opioid use. Future studies that further investigate these relationships should consider investigating additional clinical characteristics not accounted for here but previously shown to predict opioid abuse, such as number of opioid prescriptions, number of opioid prescribers, early opioid refills, and psychiatric diagnoses51. Second, data currently available from NHANES does not include results beyond 2015. Thus we are unable to capture time and prevalence trends after some of the most recent legislative changes in marijuana legalization and responses to opioid epidemic. With NHANES data we cannot discern between medicinal and recreational marijuana use. The cancer variable for our analysis is not confirmed with medical records but instead is self-reported and subject to recall bias. Thus, we do not have additional information about respondent cancer status that may impact substance use and it is possible that these data may not be generalizable to all cancer patients with a verified diagnosis. However, NHANES data has been used to investigate cancer in other studies. Finally, we defined opioid use based on filling a prescription within the last 30 days, which may be an under representation of total opioid use. While the complex, multistage probability sampling method of NHANES data collection introduces statistical challenges, our analysis effectively accounts for confounding variables via propensity score matching and multi-variable analyses. Ultimately, while the NHANES data is self-reported and subjective to sampling bias , we are able to investigate the outcome of substance use in this representative population otherwise not previously documented. Currently, medical marijuana is legal in 25 states and Washington DC, with retail marijuana legalised in four states and Washington DC. On 1 January 2014, Colorado became the first state to legally sell retail marijuana to people 21 years or older. Shifting regulations have been accompanied by technological innovations, including electronic vaporisers for tobacco and marijuana. These developments are likely to transform use of these substances, especially among young adults.