Sunday, September 4, 2011

Missing Data

By Matt Curtis

Looking back to the situation five or ten years ago, we’ve made major progress in building an evidence base around overdose, both in terms of the prevalence of overdose in different kinds of drug using communities and the efficacy of prevention and response programs. More research is happening now than ever before, at a bigger scale and in more places.

A random graph not actually associated with this article
As with any new area that epidemiology sticks its nose into the first wave of research is often all about counting: how big is the problem, what kind of people is it affecting, where and in what way? Although there’s work to be done still on these kinds of most basic questions, we have a pretty good idea of the scale of the overdose problem in many parts of the world and enough solid information to take action at a public health level without fearing we may be wildly off base. (Some people would disagree with that statement, which may deserve a fuller discussion on OPA; for now I’ll only note that doubters of overdose prevention programming are nowadays mostly confined to the wing of public health usually referred to as “Wrong.”)

So what should come next? Here, thinking out loud, are eight areas that I think researchers could fruitfully investigate:

(1) Smoking and cardiovascular and lung disease: How much is tobacco smoking-related cardiovascular disease related to cocaine (and other stimulant) overdose?  Similarly, how much is smoking, COPD, chronic bronchitis and other chronic lung disease related to opioid overdose?  My gut tells me the answer in both cases is ‘probably a lot, especially in combination with other health problems,’ but there’s very little published information on these subjects. And I don’t know a single harm reduction organization that has a serious tobacco harm reduction or smoking cessation program (though that may be just because I need to get out more). Are we failing people on this because everyone’s just thinking that smoking is pretty punk rock and part of the scene and all I need is nic-fitting crack users running around my drop-in center?

(2) Risk factors for fatal overdose: By now, many studies have examined risk factors for nonfatal overdose, especially for opioids, and I think we’ve got a pretty good picture of the main issues (more on which in an upcoming OPA article). But what about fatal overdose? We have some evidence for basic risk factors, like older age, but a less than clear understanding of why risk is heightened (e.g. with age, probably comorbidities, but we need details). It’s much more complicated and potentially expensive to organize studies on overdose death, but doing so will fill gaps in our understanding of how to best target overdose services.

(3) We need more prospective studies: When is someone going to do a great time-to-event study (‘survival analysis’) or in general prospective (forward-looking) studies to look at how people are using the information and naloxone we give them? Wouldn’t this be a better way of assessing whether certain kinds of training or naloxone distribution systems or prevention messages are more effective than others, compared to all those damned cross-sectional surveys common to underfunded harm reduction organizations?  The N-ALIVE study in England is kind of doing that – and most likely there will be lots of interesting data to examine for years to come from that study.

One thing such studies could be used to investigate is (4) whether overdose prevention programs reduce overdose incidence.  We now have a growing and notably consistent body of evidence at an ecological level that strongly suggests scale-up of overdose education and naloxone availability reduces overdose mortality. But it’s much less clear that current overdose prevention efforts are reducing how often nonfatal overdoses occur (and for whom, in what circumstances, what kinds of overdoses, etc.).

(5) What would a ‘complete package’ overdose prevention program look like?: Related to mortality reduction is the question of what might a truly comprehensive overdose reduction effort look like.  Naloxone has gotten most of the attention in recent years, but it’s certainly not the only important tool available, and isn’t relevant for non-opioid users, doesn’t on its own do anything for prevention, and may not even be the most important tool to reduce overdose mortality risk for opioid users (if we had to choose only one intervention, which we don’t). Opioid substitution therapies and safer injection sites are both known to greatly reduce overdose risk but we have strangely few examples of where both those services are also available alongside naloxone distribution efforts (maybe in Germany?). And what else is there? How important are other drug treatment modalities? Access to psychiatric services? Social or economic assistance? What do we do about cocaine or prescription opioid users?

Every locality will have its own nuances, but if we’ve been able to come up with evidence-based core packages of services and policies for the prevention, care and treatment of numerous infectious and chronic diseases, why not for overdose?

(6) Are prevention messages mostly wishful thinking?: Speaking of overdose prevention education, I often have the feeling that everything we say about not mixing, not using alone, minding one’s tolerance, and so on is very closely akin to saying “Remember not to get high today.” That’s not an argument for skipping prevention education, and I have a deep (and potentially unfounded) belief that behavior around overdose risk will change in communities over time – just as behavior around injection and HIV risk has changed – the more we talk about it. But I’d really love to see some actual evidence for this. We need a team of anthropologists or sociologists running around looking at what people actually do with the overdose information we give them.

(7) Do overdose services bring more people into health care?  This is a question beloved, for not entirely bad reasons, by AIDS funders who we trick into supporting overdose programs. There’s some anecdotal evidence that new overdose services are exciting and draw more people into other kinds of health services, especially in environments where harm reduction programs aren’t well developed beyond basic syringe exchange. I’ve heard LOTS of stories from outreachers at this point about waves of new participants running through the door once naloxone was being offered – and subsequently getting hooked up with clean injecting gear, VCT, primary healthcare, etc. We need more than nice stories though, and we can start by designing a simple way to identify and quantify new clients when overdose services are started.

(8) Which gear to pack?: This is a smaller issue than the things described above, but it’s been bugging me: Do people really and truly need to use intramuscular needles to inject naloxone?  I hear about lots of people using insulin needles, and basically no cases of that not being effective. I think there are a few studies that are starting to better incorporate questions about what needles/syringes were used in OD response, but I’m not sure I’ve seen anything published, and definitely not anything that’s looked specifically at the efficacy of short vs. long needles in naloxone administration.

Every time I ask educated types about this they say something to the effect of, oh yeah, definitely go with an IM needle, and if you must use an insulin or tuberculin needle, inject into the tongue. Fine advice in the absence of definitive evidence, but it’s a major practical problem in a lot of places where I work, where people absolutely refuse to carry one extra potentially unnecessary piece of drug gear because on a day to day basis they are very likely to have a cop greedily searching their pockets and looking for any excuse to make their day very bad. My question is do we really need to be doing the work to ensure that people are able to carry extra IM syringes because they really are a better delivery device for naloxone, or can we drop it and tell people it’s OK to use the syringes they’re already carrying for their drugs?


  1. Great points, Matt. I'm going to try to respond to a few...

    1&2) There are selected studies demonstrating elevated risk of overdose fatality associated with selected co-morbidities (e.g. HIV). This issue is complicated and I think has to be understood based on specific diseases rather than a general sense of one's overall health (I tried that latter approach and it didn't give much for results). I suspect that the association with cocaine is cardiovascular disease. If I'm right, it may help to explain some of the demographic trends among those who die of cocaine overdose.

    3) Prospective studies: There are actually several, and still some good data coming from Baltimore. But I agree the current state of research for naloxone distribution is disappointing. The major U.S. funding agencies have missed the boat on ideal studies - involving site-based control groups - and now any studies are hamstrung by ethical considerations: how many IRBs would be comfortable with a randomized control group? I certainly would not be.

    If we want to measure a reduction in population-level mortality, we're going to need a large, multi-site trial that runs at least 5 years. New York City, Baltimore, and Newark would have been the optimal U.S. sites given the large numbers of deaths in those cities. I'd be happy to talk about power calculations more based on some modeling I've been doing.

    4 and 7) Do OD prevention programs reduce all overdose, not just fatal? This is a big question and thus far we have to assume that, no, they don't. But the cool part about this question is that a single site, reasonably-sized trial could actually answer this question. The only obstacle is funding – this would be an “RO1” which means an expensive federal study. Everything costs more in research and such a study would exceed a million. Notwithstanding claims of interest in overdose research, the subject itself remains homeless.

    With regard to bringing people into healthcare, I think this is a resounding yes. And probably easy to document although again would require actual research dollars (perhaps as a modest addendum to NX research?).

    8) Regarding needle length, a convincing study in Vancouver established that subcutaneous injection is acceptable for naloxone delivery (, with no delay in response. It is also possible that ancillary resuscitative efforts might be more important with this mode of administration and, more importantly, the researchers compared 0.4mg IM to 0.8mg SC. I would not hesitate to administer naloxone subcutaneous (i.e. with a short needle) if necessary, and perhaps short needles are an option with the 10mL (large) vials of naloxone since a larger dose of naloxone could be given … but I would hesitate to advocate distributing only short needles with the single dose 0.4mg vials.

  2. Re #2: Being out of line of sight of another conscious person at the point you lose consciousness. Davidson, P., 1999. Circumstances of death: An assessment of the viability of using non-toxicological coronial data to investigate opiate overdose risk factors, Perth, Western Australia: National Center for Research into the Prevention of Drug Abuse.

  3. Matt, this is great.

    Yes, please.

    I remember we made a video at our needle exchange a million years ago back in Massachusetts and one of the women who hung out every day was talking about how she loved the overdose groups "because even though she still overdosed all the time, at least she knew what she was doing wrong." That always stuck with me. Does this pretty much sum up the effectiveness of our "prevention" messages? Not to mention the fact that although she laughed after she said it, it still made me realize that she was attaching a moral judgement to her behavior, that she thought she was doing things "wrong." Yikes, good job harm reductionists.

    What Phillip said. Subcutaneous is fine with an insulin syringe. Better than the tongue, which I have actually heard some doctor-types (who I love) say was not really effective and you risk getting bitten or hurting the person anyway.

  4. great post, matt! my 2 cents -- given the recent stats from nyc that 98% of od's involve 2+ drugs and 50ish% involve 3+ drugs, i think we need to start picking that apart, study wise. (my speculative) examples:

    heroin + cocaine: what is evidence of increased od risk for this combo? is it related to greater half life of cocaine vs heroin? could use of longer lasting opioids mitigate risk?

    cocaine + alcohol: forms metabolite (cocaethylene) with similar action as cocaine but longer lasting. how does this effect od risk?

    psych meds: many people i talk to are prescribed or buy off street benzos and/or seroquel (generic: quetiapine), what are interaction effects with opioids and stimulants?

    and re 1: i agree we need to work on tobacco harm reduction! cocaine related death is mostly cardiac mediated (not counting cocaine associated accidental or violence related trauma) so we need to understand both acute interaction of nicotine and cocaine in od events and their interaction with regards to atherosclerosis/cardiovascular disease. there is a lot of evidence that cocaine (like nicotine) accelerates atherosclerotic plaque formation (clogging of arteries).

    re 6: i also like to believe people change their behavior after my training raps but given the usual results around public health education that's likely wishful thinking. that said, people often seem to have "aha!" moments in my trainings when they realize that oxys are like heroin, not like xanax, ie. it's about drug class, not delivery mechanism; and tolerance is why their friends od when they get out of jail. maybe tracking a change in post incarceration od rates is a way to get at the education effectiveness question.

    re 8: i'm still in favor of im needles. if you need .8mg to work sc you'll still need bigger syringes (do they make 2cc's?) or a stronger formulation like IN dose (1mg/ml). IN delivery seems the best way to get around needle issues. and i always put the needles on the naloxone script when i write it up, which puts people at ease though i don't know if it actually makes a difference in actual police interactions. and obviously is not relevant in places where naloxone distribution doesn't need one...

    thanks opa for setting up this blog!