Advice on studying infections in persons who inject drugs

Hello, I am one of the second year ID fellows at VUMC. I’ve been working on a project this year to try and look at possible trends of serious infections in persons who inject drugs. In the midst of the opioid epidemic, we are seeing significant cases of young people having to spend a lot of time in the hospital with severe, life threatening infections.
Our plan is to look at this in several ways with several hypotheses.
One, we plan to look at the trends here in the state of Tennessee. To do this, we will use hospital discharge data from the dept of health from 2010-2017. We will look at ICD 9/10 codes for severe infections (inpatient admissions for endocarditis, osteomyelitis, septic arthritis, epidural abscess, extremity skin/soft tissue infection). Using this data, we’ve planned to look at the total numbers and also the numbers of infections in persons who also carry a diagnosis of hepatitis C or substance use disorder (primarily opioids). We plan to look at this on a case per 100,000 persons.
Secondly, we want to look at differences in the numbers of these infections among the grand divisions in TN (East, Middle, West TN). The thought is that East TN has been the epicenter of the opioid crisis in TN, so we thought it would be interesting to look at this to see if admissions for infections were correlating with this. We are planning to use the same ICD 9/10 codes but just separating the patients into a Grand Division based on the county they were discharged from.
Thirdly, we want to look at how the numbers of these infections in TN compare to the nation as a whole. To do this, we will look at a national database called HCUP. Unfortunately, this dataset is a little less sophisticated. We can basically enter ICD 9/10 codes and generate numbers that match those codes. We can break those down based on age, race, sex, and payer though, which we plan to do and look at.

I know I am leaving a lot of details out. We know we have some limitations to our study, notably determining a denominator. For that, we’ve planned to try and use a diagnosis of either hepatitis C or substance use disorder as surrogates for intravenous drug use (this has been done in other studies). Also, reliance on administrative data has it’s inherent constraints.

Our hope is to try and produce the most impactful study we can with what we have available to us here. We think that there is something interesting to uncover here. If anyone has any thoughts or tips on how to potentially make this more meaningful we are all ears and certainly appreciate it

1 Like

I had a Nigerian colleague once who never seemed to be at a loss for a Nigerian proverb apt for every situation. One of these was, “When the desirable is not available, the available becomes desirable.” It seems to me that much of your thinking about this project is propelled by this principle. Most of your discussion is at the ‘bare metal’ level of the data and the calculations you can do with them, lacking higher-level scientific guidance. (Imagine approaching the diagnosis of a patient primarily on the basis of what diagnostic equipment you have available, instead of based on the chief complaint!)

You mention 2 things in your description of the project, which I would encourage you to elaborate. First, you mention “hypotheses”. What scientific hypotheses would you like to explore? (Maybe some have come to mind while you cared for patients with infections caused by injection drug use?) Secondly, you mention “impact”. What policies would you like to inform or advocate based on your research? I’m aware of certain IDU behaviors that increase the likelihood of certain kinds of infections (e.g., licking the needle). But are there readily available behaviors that could reduce the infection risks of IDU? Can these be encouraged in a way that is helpful? Are there certain injection drug users who suffer repeated infections, but others who do not? What are the determinants of risk?

1 Like