nlennertz-blog

[nlennertz-blog]

ANOVA Analysis of Opioid Use

For this ANOVA analysis, my null hypothesis is that there is no relationship between race or gender and the age at which a person experiences the onset of opioid abuse.  In these cases the race (ETHRACE2A) and gender (SEX) will be the categorical predictive variables and age will be the quantitative response variable (S3CD3Q13A).

To perform the ANOVA procedures, I used the following statements:

The resulting output is:

For ethnicity, we see that the sample means are not equal but very close indicating that the age of onset of opioid abuse appears constant regardless of ethnicity.  From the ANOVA results, we see that the P Value is 0.57 which is far above 0.05 leading me to say that we cannot reject the null hypothesis that the means are equal.

For gender:

Similar to the results for ethnicity, a person’s gender does not appear to influence the age at which opioid abuse starts.  The sample means are nearly identical and the P value is 0.1263 which is far above 0.05.  The null hypothesis that the means are equal cannot be rejected.

[nlennertz-blog]

Univariate and Bivariate Analysis

For this section we are graphing variables to examine characteristics of the response and explanatory variables as well as relationships between variables.  My primary research questions is centered on whether major depression is a risk factor for opioid abuse and my secondary question is about whether marijuana use is a risk factor for opioid abuse.

For the primary question, I am using the quantitative variable S4AQ6A, “AGE AT ONSET OF FIRST DEPRESSION EPISODE", as the explanatory variable and the quantitative variable S3BD3Q2A, "AGE FIRST USED OPIOIDS", as the response variable.

My current SAS program looks this way:

The first univariate graph is for age first used opioids:

The mean is 23.5, median is 20.0 and the mode is 18.  The standard deviation is 10.43 so there can be a wide variation for the given mean.

The second univariate graph is for age first depression episode:

The mean is 24.97, median is 23.0 and the mode is 18.  The standard deviation is 11.55 so there can be a wide variation for the given mean.

The last univariate chart is for age first used cannabis:

The mean is 16.5, median is 16.0 and the mode is also 16.  The standard deviation is 4.47 so the values are tightly centered around the mean.

Looking at the scatter plots between the two explanatory variables and the response variable of age of opioid use we have:

There appears to be a large cluster of results in the younger ages for first use, but the spread is very wide indicating that there is not a strong relationship between first episode of depression and first opioid abuse.  This lack of correlation is also implied by the mean of first depression episode being higher than that for opioid indicating that opioid use generally precedes a depression episode.  Their modes are the same but depression generally occurs at a later age.

Graphing cannabis use against opioid use, we see a tighter concentration than with depression.  Given the tight clustering of the values it is hard to imply any correlation.  Although the mean and mode of the cannabis use is at an earlier age than those of opioid abuse, we can’t infer that cannabis use is a significant risk factor for opioid use.

[nlennertz-blog]

Next iteration of SAS program

The output of this iteration is:

In this iteration I added several variables to look at the ages at which people first used opioids or cannabis, and the age of their first depression episode.  I then binned the age variables in increments of 10 years so that they could be compared with the same bin ranges.    Although I had marked them as missing data earlier, for each of the age variables, I had to explicitly exclude blank values from the results when binning them into age ranges.  Each age variable had reached a cumulative percentage of almost 50% by the age of 19 for opioid and depression but almost 90% for cannabis use by age 19. There is an apparent lack of use of medicines or drugs to improve low mood in the last 12 months which may indicate that depression is not a major factor in starting opioid use.

[nlennertz-blog]

First SAS Program

For the second week’s assignment, below is my SAS program to look at distribution metrics for several of the variables involved with my analysis of risk factors for opioid abuse and marijuana use.

After running the program, I get the following results.  The first set of variables are the simpler ‘have you ever used’ variable for opoids and marijuana.

For Ever Used Opioids, I have limited the results to only those surveyed who had ever used opioids of which there are 1815 results.  Of those 1409 have also used cannabis, 405 had not, and 1 is unknown.

In the second set of variables, I am looking more specifically at use in the last 12 months.

For both of these variables,the variables are defined as:

0. No diagnosis 1. abuse only 2. dependence only 3. abuse and dependence 

For the 1815 total results, only a small percentage have abused or been dependent upon opioids or cannabis in the last 12 months and more had abused or been dependent upon cannabis than opoids.  A large proportion of the results had neither an opioid or cannabis diagnosis

[nlennertz-blog]

Correlation of Psychological Risk Factors with Heroin Abuse

I will be working with the NESARC Wave 1 codebook to learn more about opiate use and addiction.  I don’t know which variables I will use but will probably focus on family and economic factors.

After hearing a lot in the news about epidemic opioid abuse in the United States, I am interested in what are risk factors leading to opioid abuse and addiction.  I want to investigate what personal, geographic, or economic factors like income contribute to starting opioid use. I have seen news programs that spotlighted perfectly happy and normal teens like straight A Boy Scouts that lose everything when they start opiates.  Pain relief is a primary risk factor but I am also interested in how much psychological factors influence opioid abuse.

Secondarily, I am interested to see if there is any statistical evidence that marijuana is a ‘gateway’ drug towards opioid use.  I will look for correlation between marijuana and opiate use and whether there is any information that marijuana preceded opiates.

For a literature review I use Google Scholar and Google Search to identify several relevant studies and articles based on the search for risk factors for non-prescription opiod abuse:

1)      Relationship between Nonmedical Prescription-Opioid Use and Heroin Use Wilson M. Compton, M.D., M.P.E., Christopher M. Jones, Pharm.D., M.P.H.,

and Grant T. Baldwin, Ph.D., M.P.H. New England Journal of Medicine Jan 14, 2016

 2)      Risk and Protective Factors Associated with the Non-Medical Use of Prescription Drugs SAMHSA’s Center for the Application of Prevention Technologies

January, 2015

3)      A Model to Identify Patients at Risk for Prescription Opioid Abuse, Dependence, and Misuse J. Bradford Rice, PhD,* Alan G. White, PhD, Howard G. Birnbaum, PhD,* Matt Schiller, BA,* David A. Brown, PhD, MPH,† and Carl L. Roland, PharmD‡ Pain Medicine, 2012

 4)      DrugAbuse.com http://drugabuse.com/legalizing-marijuana-decreases-fatal-opiate-overdoses/

Legalizing Marijuana Decreases Fatal Opiate Overdoses, Study Shows

5)      DrugAbuse.com http://drugabuse.com/library/prescription-opioid-addiction/

Prescription Opioid Addiction: What is Causing the Epidemic?

The first three resources made extensive use of data and study evidence whereas the two web sites provided higher level discussions of risk factors.  Based on  reviewing these sources, I hypothesize that the abuse of opiod drugs like heroin will be positively correlated to psychological issues such as depression, which is a form of pain although not a totally physical form.  I also hypothesize that there will be a positive correlation between marijuana use and opiate use but possibly less correlated than psychological risk factors.

There are a large number of variables in the NESARC database related to drug use in relation to psychological factors and other drug use.  These are primarily in sections 3B: Medicine Use and section 4A: Major Depression.